TSPE 0.700 Public Beta out now!

[Ventolin]

Hi folks,

After a couple years of quiet development in the darkest corners of the Internet, another public beta of TSPE is now available!

http://www.bitstreamtools.com

For those that don't know, TSPE (Transport Stream Packet Editor) is a non-linear Editor / Analyser.

So, what's new...

Supported Formats:
MPEG2 / H.264 / VC-1 / AC-3 / DDP / MLP / DTS / DTS-HD / DTS-HD MA / AAC / MPEG1 Audio / MPEG2 Audio / PCM (everything? :)

.TS and .M2TS support including stripping the m2ts header to make normal 188 byte .TS files.

No re-encoding of the stream to preserve maximum quality, I frame to P frame accurate, clean video/audio editing, timestamp and table correction.

You can cut commercials from captured sources, detect commercials using format changes, scan for errors, import TSPE and M2R logs, fix broken streams including timestamp wraparounds, continuity errors, timestamp gaps, realign to start codes and resync audio.

Stream stripping with table correction, PCR remapping, Demuxing, Core Audio from HD audio extraction.

Video preview is much more accurate, uses your own directshow filters, test and full edits are previewable with TSPE's own internal viewer, multi-part ts file joining with autosorting EDL (Edit decision List). Marker List to enable bookmarking of positions.

Packet Header Decode for analysis. Resizable interface.

Plenty more too, visit the website for details.

TSPE is now shareware, free trial period is 45 days. I wanted to release TSPE once I was sure it performed well enough to justify payment. It is still in beta testing, but most of the major bugs have been ironed out (I hope!). Still plenty more work to be done, eg. m2ts and vc1 support is still sketchy, but it does do the job just about.

Please let me know how you get along; of course I would like to know your feedback good or bad and especially if you have some streams which don't work properly.

Have fun,

Regards,

Vent

[Atak_Snajpera]

Those two ts samples do not work (no preview)
http://www.mediafire.com/?sharekey=fc577931a2d88b41d2db6fb9a8902bda

One more thing. Instead of small 'loading' window I would use window with graphic (splash screen)

[Ventolin]

Those two ts samples do not work (no preview)
http://www.mediafire.com/?sharekey=fc577931a2d88b41d2db6fb9a8902bda

One more thing. Instead of small 'loading' window I would use window with graphic (splash screen)
Hi,

This is a snippit of the status log when you try to load either of the two .ts files:

Opening: X:\tmp\Batman-5.TSSplit.1-57.ts
Opening File...
Auto Finding PIDs
Unable To Find PAT

Both streams do not have a PAT. The PAT is essential for a Transport stream and there should be one every half second or so. A transport stream should have at least 1 PAT with PID 0, 1 PMT and 1 stream (which can be video or audio).

Try loading the full source into TSPE instead.

Thanks for the note about the splash screen, I haven't gotten around to doing one yet, same with an icon. These will be added in due course, once I've learnt a bit about graphic design :)

Cheers,

Vent

[mediator]

many real life streams don't come with PAT (e.g. captures from DVB, seems often the PAT is just stripped during the recording process). Maybe you should consider supporting them anyway, even if they don't strictly comply with ISO.

[Ventolin]

many real life streams don't come with PAT (e.g. captures from DVB, seems often the PAT is just stripped during the recording process). Maybe you should consider supporting them anyway, even if they don't strictly comply with ISO.

PAT and PMT really are essential to a Transport stream. If a capture program strips them out of a stream, then the capture program is at fault. TSPE would then need to repair the stream to fix the poor design of the capture program. What real life streams have you come across without PAT and PMTs? If a specific capture program is stripping out the PAT and PMT perhaps the authors could fix their software?

The PAT and PMT tell the decoder what is in the transport stream. The PAT gives a list of the Programmes available in the form of PIDs for PMTs. Once the PID of the PMT is known, the PMT can be read to find out the what PIDs, types and timing information (PCR) are of the streams that make up that programme. When capturing a complete mux you may have *hundreds* of streams in the transport stream. How is anything supposed to know what streams belong to what programme if you dont have the PAT and PMTs?

Nevertheless, in the most basic mode TSPE *still* works even without the PAT and PMT. If you are able to find a working directshow filter setup (i.e. one that works in graphedit) you can then have video preview. TSPE will still allow you to set start and end points and make an edit if you turn the fix options off with or without the video preview.

Since I couldn't get any filter combination to work in graphedit, I couldn't test what happens with TSPE completely, but I did edit a section out successfully nonetheless.

Full table editing (for PAT and PMT) is already on the TODO list however. Once that is implemented it will allow TSPE to repair streams without PAT or PMTs. Perhaps I could get TSPE to launch an external viewer like MPC instead of it's internal viewer for test edits in the mean time. I could code that in if there was enough demand...

To edit these streams "blind" switch to the decode window (using the View) button. Then use the decoded PTS/DTS timestamps as a guide to the real timecode of the stream. Then you can navigate to test out positions. Make sure you turn off Fix Tables and Fix Timecodes and turn off the internal viewer from the menu options. You can then make test and full edits and preview the filename.test.ts or filename.edit.ts files generated with something like MPC. Scanning the file revealed a timestamp gap of 49 seconds or so at the start in the Batman sample; the scan dropped a marker at the end of the gap so you can navigate there easily.

Hope that helps,

Regards,

Vent

[canTsTop]

hello,
i have this sample http://www.megaupload.com/?d=5A68THW9 is it possible to fix it so i have audio / video synch after encode, and how?

[Atak_Snajpera]

once I've learnt a bit about graphic design
I can create something for you for free.

Another thoughts:
1) Main form should appear in center of the screen
2) Movie slider should start at first decodable key frame instead of P or B frame

I forgot to say. Excellent job!

i have this sample http://www.megaupload.com/?d=5A68THW9
Are you joking ? :) This sample is like wooden wall destroyed by Minigun. Too many holes :) ...

[Ventolin]

hello,
i have this sample http://www.megaupload.com/?d=5A68THW9 is it possible to fix it so i have audio / video synch after encode, and how?

Woah, this is a glitchy sample!

After loading into TSPE, do a scan. This will reveal:

Continuity Errors: 764, Transport Error Indicator Errors: 0
Timestamp gaps: 16, PCR gaps: 7
Total number of separate discontinuous regions: 655

The Marker List will now contain alot of entries. Looking through there are 4 points which we can use:

A. First entry is at offset 1186092 (5 seconds in marker list)
B. Last entry is at offset 5670456 (24 seconds in marker list)
Gap in the middle between:
C. offset 2977168 (13 secs)
D. offset 5000236 (22 secs)

Every other marker entry is too close to each other to be of use, the biggest gap being about 3 seconds, with not much useable inbetween. Note that the marker list time values aren't quite accurate, once the marker is double clicked a time based on the nearest timestamp in the stream will be used (assuming timestamp mode is running which it will be by default). The byte position is the only thing we should be concerned with.

So, we need to make 3 cuts:
from start to point A.
from point C to point D.
from point B to the end

So:
1. double click the first entry, then click "Set" under End Point
2. Click on EDL tab to switch to EDL view
3. Click Add under EDL view (first entry will appear)
5. Click on Markers tab
6. Scroll to point C in Markers (look for offset 2977168 (13 secs)
7. Double click entry - this moves current position to that offset
8. Click Set under Start Point
9. Scroll to point D (offset 5000236 (22 secs))
10. Click Set under End Point
11. Click on EDL Tab to switch to EDL view
12. Click Add under EDL view (second entry will appear)
13. Click on Markers tab
14. Scroll down to the bottom of the list to the last entry (offset 5670456 (24 seconds in marker list) )
15. Double click last entry
16. Click on Set under the Start Point
17. Scroll to the very end of the video with the trackbar and using >> to make sure you're at the very end
18. Click on Set under End Point
19. Click on EDL tab
20. Click on Add under EDL view - 3rd entry should appear.

Lots of steps to create a 3 entry EDL, but it's simpler once you do it.

The EDL should now have something like this:
00:00:00.00 0 00:00:03.80 1186280
00:00:15.08 2977168 00:00:21.96 5000424
00:00:29.79 5670456 00:00:46.01 10440580

If you click Edit now, you will generate a playable, non-crashing clip. The audio cuts out a bit too much though but stays in sync.

To try and improve the audio, Uncheck Menu Edit->Clean Edit Point->Clean Audio Frames. This cuts audio at byte positions instead of looking at timestamp positions from the video stream at the expense of possible broken audio packets.

Now clicking edit produces a playable stream with better audio and we're done!

I hope the rest of the stream is clean, if you have that many errors all over the stream, it's really too damaged to be of much use.

If you're going to be re-encoding from this source, you'll have to do some more experimentation, but from a playback point of view, the edited clip is in sync.

Good Luck!

Regards,

Vent

[Flaarn]

Thanks for this ventolin, checking it out on a couple of h264 transport streams now, bbc hd and channel 4 hd.
The big issue for me has always been frame accurate cutting so that no extra 2.0 audio frames are present if the main broadcast is 5.1 but adverts are 2.0, many conversion tools tend to throw a wobbly when there are audio changes, more so when the transport stream contains some 2.0 ac3 frames at the start, sure it can be patched when parsing, but it adds another step in the process which in my opinion is a nuisance, with this new release I'm hoping that I can bypass this step and have a clean ts to play with, thanks again for your ongoing work.

[Ventolin]

I can create something for you for free.
Seriously? Thanks! I can create one for free too :P but if you really want to have a go, we'd be happy to look at it :)

Another thoughts:
1) Main form should appear in center of the screen
2) Movie slider should start at first decodable key frame instead of P or B frame
Thanks for your suggestions, I'll think about 1) but might be confusing with multiple instances. 2) Slider should be at the very start of the stream as this is a *packet* level editor and analyser - the PAT usually is at the first packet!

I forgot to say. Excellent job!
Thank you kindly!

Are you joking ? :) This sample is like wooden wall destroyed by Minigun. Too many holes :) ...
LOL, but still possible to get something out, see above :)

Cheers,

Vent

[Ventolin]

Thanks for this ventolin, checking it out on a couple of h264 transport streams now, bbc hd and channel 4 hd.
The big issue for me has always been frame accurate cutting so that no extra 2.0 audio frames are present if the main broadcast is 5.1 but adverts are 2.0, many conversion tools tend to throw a wobbly when there are audio changes, more so when the transport stream contains some 2.0 ac3 frames at the start, sure it can be patched when parsing, but it adds another step in the process which in my opinion is a nuisance, with this new release I'm hoping that I can bypass this step and have a clean ts to play with, thanks again for your ongoing work.
Hi,
I've found the audio switchover point to be less than accurate on the broadcaster's end which means that you might chop off the start of the audio at the start of a segment. I hate this alot as my offboard processor will mute for a second at the switchover point. I did complain to the BBC some time ago, they tried to fix it but screwed things up alot and promptly gave up.

Hope you find TSPE useful though, look forward to your feedback.

Regards,

Vent

[Atak_Snajpera]

2) Slider should be at the very start of the stream as this is a *packet* level editor and analyser - the PAT usually is at the first packet!
I'm not saying about discarding everything before first keyframe.
Slider should be rewinded to first key-frame by default. See screenshot
http://img210.imageshack.us/img210/4282/new1ep5.th.png (http://img210.imageshack.us/my.php?image=new1ep5.png)

I see no reason to cut stream at P or B frame. Only cutting at key frame maintains audio synchronization if we want to transcode file further. This is main reason why I use this tool.

[canTsTop]

Woah, this is a glitchy sample!

After loading into TSPE, do a scan. This will reveal:

Continuity Errors: 764, Transport Error Indicator Errors: 0
Timestamp gaps: 16, PCR gaps: 7
Total number of separate discontinuous regions: 655

The Marker List will now contain alot of entries. Looking through there are 4 points which we can use:

A. First entry is at offset 1186092 (5 seconds in marker list)
B. Last entry is at offset 5670456 (24 seconds in marker list)
Gap in the middle between:
C. offset 2977168 (13 secs)
D. offset 5000236 (22 secs)

Every other marker entry is too close to each other to be of use, the biggest gap being about 3 seconds, with not much useable inbetween. Note that the marker list time values aren't quite accurate, once the marker is double clicked a time based on the nearest timestamp in the stream will be used (assuming timestamp mode is running which it will be by default). The byte position is the only thing we should be concerned with.

So, we need to make 3 cuts:
from start to point A.
from point C to point D.
from point B to the end

So:
1. double click the first entry, then click "Set" under End Point
2. Click on EDL tab to switch to EDL view
3. Click Add under EDL view (first entry will appear)
5. Click on Markers tab
6. Scroll to point C in Markers (look for offset 2977168 (13 secs)
7. Double click entry - this moves current position to that offset
8. Click Set under Start Point
9. Scroll to point D (offset 5000236 (22 secs))
10. Click Set under End Point
11. Click on EDL Tab to switch to EDL view
12. Click Add under EDL view (second entry will appear)
13. Click on Markers tab
14. Scroll down to the bottom of the list to the last entry (offset 5670456 (24 seconds in marker list) )
15. Double click last entry
16. Click on Set under the Start Point
17. Scroll to the very end of the video with the trackbar and using >> to make sure you're at the very end
18. Click on Set under End Point
19. Click on EDL tab
20. Click on Add under EDL view - 3rd entry should appear.

Lots of steps to create a 3 entry EDL, but it's simpler once you do it.

The EDL should now have something like this:
00:00:00.00 0 00:00:03.80 1186280
00:00:15.08 2977168 00:00:21.96 5000424
00:00:29.79 5670456 00:00:46.01 10440580

If you click Edit now, you will generate a playable, non-crashing clip. The audio cuts out a bit too much though but stays in sync.

To try and improve the audio, Uncheck Menu Edit->Clean Edit Point->Clean Audio Frames. This cuts audio at byte positions instead of looking at timestamp positions from the video stream at the expense of possible broken audio packets.

Now clicking edit produces a playable stream with better audio and we're done!

I hope the rest of the stream is clean, if you have that many errors all over the stream, it's really too damaged to be of much use.

If you're going to be re-encoding from this source, you'll have to do some more experimentation, but from a playback point of view, the edited clip is in sync.

Good Luck!

Regards,

Vent

Thank You for answer.

This sample intentionally so damaged :)

If i correctly understand this method cuts out damaged parts off file. I did exactly as you described (with Clean Audio Frames unchecked) and it kind off worked (DGAVCIndex didn't crash). But if i again open already edited (fixed) TS file in TSPE, scan, and there is Continuity Errors: 17. It would be good to have some kind of automation of this method :) if for example there is 3 hours recording where we need to do ~20 cuts :)


Is this is only way to fix damaged TS files? Maybe something like ProjectX, it replaces damaged frames with empty frames?

[Ventolin]

I'm not saying about discarding everything before first keyframe.
Slider should be rewinded to first key-frame by default. See screenshot
http://img210.imageshack.us/img210/4282/new1ep5.th.png (http://img210.imageshack.us/my.php?image=new1ep5.png)

I see no reason to cut stream at P or B frame. Only cutting at key frame maintains audio synchronization if we want to transcode file further. This is main reason why I use this tool.
Lets take the following two scenarios:

TV capture:
When I capture a TV programme I like to start capturing a few minutes before the start of the actual programme itself - that way I won't miss the very start if it's broadcast early. Then when I load the captured .ts file into TSPE I look for the start of the programme. This almost never is at the first I frame, but usually a minute or two into the file. So I dont really need to look at the first I frame. However, I might be interested at looking at the PAT and PMTs which could be at the very start of the stream.

BD analysis:
Suppose I want to look at a stream from a BD. Again the first few packets are the PAT/PMT, followed by an I-Frame for the start of the video. In this situation, I am already at the point I want to edit with.

So, I think in most situations starting from the very beginning of the stream is correct?

I guess the only time I'd want the very first I frame is if I have missed the start of the captured programme, in which case it's a simple case of navigating to the first I frame.

In any event, it's certainly possible to make sure the edit always starts with an I frame so you don't have to look for it manually. I'll look into that possiblity further, thanks :)

Regards,

Vent

[Ventolin]

Thank You for answer.

This sample intentionally so damaged :)

If i correctly understand this method cuts out damaged parts off file. I did exactly as you described (with Clean Audio Frames unchecked) and it kind off worked (DGAVCIndex didn't crash). But if i again open already edited (fixed) TS file in TSPE, scan, and there is Continuity Errors: 17. It would be good to have some kind of automation of this method :) if for example there is 3 hours recording where we need to do ~20 cuts :)


Is this is only way to fix damaged TS files? Maybe something like ProjectX, it replaces damaged frames with empty frames?
I think you are looking for similar functionality to Mpeg2Repair where bad mpeg2 frames are repaired. Whilst this is possible, it's certainly going to take some time to implement, especially as M2R is such an awesome tool and I'm still developing my c0dering skill0rz :) It would require a detailed knowledge of mpeg2/h264/vc1 encoding.

For playback purposes, having minor glitches in place is preferable in my opinion to missing a chunk of the programme.

TSPE actually does do some repair automatically eg. timestamp gaps, so for some glitches it is automated. It all depends on the type of glitch and severity.

Cheers,

Vent

[JonRead]

Thanks, I will test out with UK satellite broadcast HD h.264 streams and let you know if I find any bugs.

[robinsj]

In your editor is there a way to manually enter a time you want to go to in a video?

[Ventolin]

Thanks, I will test out with UK satellite broadcast HD h.264 streams and let you know if I find any bugs.

Hi, Please do let me know how you are getting on!

Vent

[Ventolin]

In your editor is there a way to manually enter a time you want to go to in a video?

I'm just going to add that in right now, this very minute! Well OK, maybe the next minute or two as I'm still typing :)

Cheers,

Vent

[robinsj]

Okay another question.... I am trying to edit a few movies that I have that my wife really likes, but she really hates hearing the F-word in movies.... So I am trying to use your software to edit them out. (not really interested in getting into an arguement about editing a movie to take out language, so please leave that alone to whomever is interested in arguing)

So what is the best way to do it..... Here is what I have tried to do... I try playing the movie in the preview window, and then I just hit pause right before he says it and mark that as the end, then add that to the edits. Then I play it again until it has passed that he said it, usually only is like half a second or so... and then mark that as the beginning, and then go all the way to the end of the movie and mark that as the end and then add that to the edits. So basically I have a very small portion that is maybe 15-25 frames long that is being cut out. Well, then when I edit it and go to play it, it seems it totally misses the swear word and cuts something else out within frames from where i was trying to cut. Does this have to do with the PID's? Or the I-Frames? I tried unselecting Fix tables and timecodes, and that seemed to work, but then all I got was digital snow instead of just a clean cut...??

Any Help?

[Ventolin]

Okay another question.... I am trying to edit a few movies that I have that my wife really likes, but she really hates hearing the F-word in movies.... So I am trying to use your software to edit them out. (not really interested in getting into an arguement about editing a movie to take out language, so please leave that alone to whomever is interested in arguing)

So what is the best way to do it..... Here is what I have tried to do... I try playing the movie in the preview window, and then I just hit pause right before he says it and mark that as the end, then add that to the edits. Then I play it again until it has passed that he said it, usually only is like half a second or so... and then mark that as the beginning, and then go all the way to the end of the movie and mark that as the end and then add that to the edits. So basically I have a very small portion that is maybe 15-25 frames long that is being cut out. Well, then when I edit it and go to play it, it seems it totally misses the swear word and cuts something else out within frames from where i was trying to cut. Does this have to do with the PID's? Or the I-Frames? I tried unselecting Fix tables and timecodes, and that seemed to work, but then all I got was digital snow instead of just a clean cut...??

Any Help?

Small edits are hard to deal with as what you really want is frame accurate video. TSPE can only do I frame for the start point and P frame for the end point, so you may find your edits do not line up exactly where you want them.

However, as TSPE can process at the packet level, there is a way to edit just the audio out but leave the video intact.

This isn't so easy to do however, and will take some experimentation to get to know how to do it easily. Level of difficulty for this is 5/10.

Lets take 1 swear word as an example. You will have 3 edits:

1. The start of the movie to the beginning of the swear point (section A)
2. The swear point (section B)
3. The end of the swear point to the end of the movie (section C)

To do audio editing, set the "Frame Step" mode to be audio instead of Video (bottom left hand corner). This makes the F+ and F- buttons jump to audio frames instead of video frames.

Now do the following:
Setup section A:
1. Set the start point to the start of the movie.
2.Set the end point to the start of the swear word.
3. Click Add to add this section to the EDL.

Setup section B (the swear word):
4. Set the start point to be the start of the swear word.
5. Set the end point to the be end of the swear word.
6. In the PID list, uncheck the audio stream. (This strips out the audio stream for section B)
7. Click Add to add this section to the EDL.

Setup section C (from the end of the swear word to the end of the file)
8. Set the start point to the end of the swear word.
9. Set the end point to the end of the file.
10. In the PID list, re-check the audio stream.
11. Click Add to add this section to the EDL.

To complete the edit:
12. Turn off "Clean Video" and "Clean Audio" under "Clean Edits" from the menus.
13. Click the Edit button.

This should now copy everything up to the swear word intact, then strip the audio only for the swear word itself, then copy everything after the swear word intact.

As long as you play

[rebkell]

I'm trying to understand the PTS stamps, I have a common problem I'm encountering with the Hauppauge HDPVR unit, it has audio dropouts and then it resets and basically just starts a new recording at that spot in the stream, unfortunately the video usually continues for a half a dozen frames or so, before it restarts. I can pretty much cut out the bad place and maintain sync, I know exactly what I want to cut out, but I can't seem to match up the timestamps(PTS) in xport(which are exact matches for what are exactly in the ts file) with the ones you show.

For example the first PTS of the ts file is always
43843, but according to TSPE the first PTS is:
00:00:00:14 (29) 487 .. how to I match these two PTS stamps up? I'm not seeing a correlation between the two, I tried changing from PCR to PTS based in the Options->Timestamp Mode, but I didn't see any difference.

Upon looking I do see how they correlate, 487 is the ms value of 43843, is there anyway I can get the raw 43843 to show in TSPE

[Ventolin]

I'm trying to understand the PTS stamps, I have a common problem I'm encountering with the Hauppauge HDPVR unit, it has audio dropouts and then it resets and basically just starts a new recording at that spot in the stream, unfortunately the video usually continues for a half a dozen frames or so, before it restarts. I can pretty much cut out the bad place and maintain sync, I know exactly what I want to cut out, but I can't seem to match up the timestamps(PTS) in xport(which are exact matches for what are exactly in the ts file) with the ones you show.

For example the first PTS of the ts file is always
43843, but according to TSPE the first PTS is:
00:00:00:14 (29) 487 .. how to I match these two PTS stamps up? I'm not seeing a correlation between the two, I tried changing from PCR to PTS based in the Options->Timestamp Mode, but I didn't see any difference.

Upon looking I do see how they correlate, 487 is the ms value of 43843, is there anyway I can get the raw 43843 to show in TSPE

Hi,

TSPE shows the PTS in a number of formats:

In the decode window you have:

PID: 0x0203
Continuity Counter: 4
PTS: 13:02:29:07 (15) 46949.304s
DTS: 13:02:29:04 (09) 46949.184s
MPEG2 Timestamp: 08:54:15:09
...
PES Header:
Stream ID: 224: Video Stream: 0
...
PTS Ticks: 4225437376
DTS Ticks: 4225426576

Here you can see the actual PTS and DTS "ticks" of the frame. To get the actual seconds this represents, divide by 90000 (90KHz)

eg: 4225437376 / 90000 = 46949.304 seconds
Converting from seconds to actual timestamp gives you the timestamp shown at the start of the packet.

I'm not sure how xport calculates it, but I think TSPE does all xport does anyway as far as I know (but I haven't used xport in a while).

Hope that helps,

Vent

[rebkell]

Hi,
...
PES Header:
Stream ID: 224: Video Stream: 0
...
PTS Ticks: 4225437376
DTS Ticks: 4225426576[/CODE]

Here you can see the actual PTS and DTS "ticks" of the frame. To get the actual seconds this represents, divide by 90000 (90KHz)

eg: 4225437376 / 90000 = 46949.304 seconds
Converting from seconds to actual timestamp gives you the timestamp shown at the start of the packet.

I'm not sure how xport calculates it, but I think TSPE does all xport does anyway as far as I know (but I haven't used xport in a while).

Hope that helps,

Vent

Got it, it's the ticks, is there anyway to make the font bigger in the PES Header view(Packet Decode Window)? It's awfully small and difficult to read.

[Ventolin]

Got it, it's the ticks, is there anyway to make the font bigger in the PES Header view(Packet Decode Window)? It's awfully small and difficult to read.

Funny, I was thinking exactly the same thing when I wrote that example just now!

Added to the list!

Cheers,

Vent

[rebkell]

Funny, I was thinking exactly the same thing when I wrote that example just now!

Added to the list!

Cheers,

Vent

Another request, you have an NPTS button, which will jump to the next PTS no matter the PID, would it be possible to add a PPTS button, that will jump to the previous PTS no matter the PID.

I was writing out an edited file, that was 29.97FPS and it kept reporting : Detected framerate 12.500 fps. The edited file seems to be fine, but just wondering where it got the 12.5 FPS number from, or am I totally misunderstanding what it means, like I said it's no biggie because the edited file was just fine.

Another question about cutting/editing. When I set an end point, if for instance I set it on the first packet of an audio frame, will that frame be included or excluded from the edit. You also said it has I-Frame and P-Frame accuracy, so if I'm trying to fine tune my cuts, would I position the end cut on a P-Frame or just prior to the P-Frame? Sorry for all the questions, but I think I can work very well with this program, but I just want to understand how the cut points function, exactly.

Edit: Ignore the end point question, I found the option to include it or not include it in the EDL file.

[Ventolin]

Another request, you have an NPTS button, which will jump to the next PTS no matter the PID, would it be possible to add a PPTS button, that will jump to the previous PTS no matter the PID.
There used to a button that did that, but the usefulness of the NPTS / PPTS buttons are a bit questionable. It's trivial to add this functionality back in, but I think that will have to wait for user customisable toolbars which is for a later release.

I was writing out an edited file, that was 29.97FPS and it kept reporting : Detected framerate 12.500 fps. The edited file seems to be fine, but just wondering where it got the 12.5 FPS number from, or am I totally misunderstanding what it means, like I said it's no biggie because the edited file was just fine.
This looks to be a bug, can you send me a small sample file please (10MB should be enough).

Another question about cutting/editing. When I set an end point, if for instance I set it on the first packet of an audio frame, will that frame be included or excluded from the edit. You also said it has I-Frame and P-Frame accuracy, so if I'm trying to fine tune my cuts, would I position the end cut on a P-Frame or just prior to the P-Frame? Sorry for all the questions, but I think I can work very well with this program, but I just want to understand how the cut points function, exactly.
With "Clean Edits" on, the exact edit location is automatically determined so you don't have to worry. You can do a "Test Edit Point" by right click 2 selected EDL entries. This will make a short test file with the last few seconds of the first entry together with the first few seconds of the second entry so you can see how the final edit will look. From that you can fine tune the start/end points of the edit. Note that most of the video codecs themselves have trouble moving to a B frame directly so finding the exact edit point is really a bit of trail and error.

Edit: Ignore the end point question, I found the option to include it or not include it in the EDL file.
The "Include packet at end point" option is useful for packet accurate editing with "clean edits" off. This lets you decide if you want the current packet to be included at the end point or not. Usually if you are at an I Frame you should have this off but have it on to include the very last packet of the file. This should be on if you are merging multi-part ts files in the edl for byte-perfect merging. With "clean edits" on however, this option has no effect, which is the default behaviour.

Hope that helps,

Regards,

Vent

[rebkell]

Ahh, clean edits, ok.

Also what is the significance of Auto Find Sensitivity? I lowered it and it found some discontinuities that were giving me trouble, I lowered it to 10 and it seems to find all the bad places in this file I'm working with. What do the numbers represent? The default is 30 I believe.

I'll try to upload a piece of this file for you to look at the 12.5 FPS thing, it's just one of the standard 1080i, Hauppauge recordings.

[Ventolin]

Ahh, clean edits, ok.

Also what is the significance of Auto Find Sensitivity? I lowered it and it found some discontinuities that were giving me trouble, I lowered it to 10 and it seems to find all the bad places in this file I'm working with. What do the numbers represent? The default is 30 I believe.

I'll try to upload a piece of this file for you to look at the 12.5 FPS thing, it's just one of the standard 1080i, Hauppauge recordings.
If you could upload the glitch part with a bit on either side, I can have a look at the glitches you are getting at the same time. +/- 10 seconds from a glitch would be great, thanks.

The Auto Find Sensitivity adjusts the depth of the search it does when you first load a file. Higher is better but slower, though in your particular file it just so happens to pick up the glitches you have.

Cheers,

Vent

[rebkell]

Ok, I gave you bad info, I actually don't find anything with Auto Find, the one it does when it loads a new file. It was actually Scan that finds all the problems, I think I was overlooking some bad places it was putting in the Markers tab section. I just cut out the file and demuxed it and then remuxed it and the a/v sync was pretty much dead on.

I'm sorry for the bad info I was giving, but I did learn a lot and I know my problem areas pretty well and this looks like I can fix them, I'm going to have to cut out anywhere from 6 to 8 frames to maintain sync, but it's a bad place in the file anyway, so it's really no big loss considering it was a bad place anyway. The good thing is that everytime I have an audio dropout, when the unit recovers it always starts afresh with an IDR frame and the audio is exactly 100 ms ahead of the audio.

We/you really need some of this stuff documented, I know it's a pain, but it's hard to determine some of this stuff even with trial and error, I probably would have never figured out the clean edits thing, since it seems to basically override other things.

[rebkell]

One more feature request, when you display the Frame type in the bottom right status bar, it would be convenient if you signified if the I Slice was an IDR frame.

Another thing I've noticed, it may be by design, is that when you are navigating on the track bar and you click to the right or left of the slider, it sometimes jumps two minutes and if held down it will jump numerous times, which is all well and good, but it doesn't display any of the frames when it jumps, only the last one when it stops.

All in all, this program is really nice, it has a lot of controls and it's just a matter of learning how to use them for my needs.

[Ventolin]

One more feature request, when you display the Frame type in the bottom right status bar, it would be convenient if you signified if the I Slice was an IDR frame.

Another thing I've noticed, it may be by design, is that when you are navigating on the track bar and you click to the right or left of the slider, it sometimes jumps two minutes and if held down it will jump numerous times, which is all well and good, but it doesn't display any of the frames when it jumps, only the last one when it stops.

All in all, this program is really nice, it has a lot of controls and it's just a matter of learning how to use them for my needs.
Hi,
Glad to hear you are able to get your glitches sorted! I got your sample file thanks, will look into the framerate bug.

IDR frame now identified in the bottom right status corner as requested, also IF+/- jumps to IDR frames if they are present in the stream. Need to do some more testing with that though.

Documentation is important, I know, trouble is the program is in quite a state of flux at the moment, so some of the options are changing and a whole lot of others are being added. Alot of the options are there so you can turn them off if something goes wrong though, normally the option defaults are the ones you should always use. I do feel though that you shouldn't really have to RTFM if you want to know how to do something, the program should tell you or it should be obvious, so that's where I would like to head towards.

Trackbar behaviour is by design, requests are actually made for frames, but the video decoders just can't handle that level of "scrubbing" - it takes alot to decode the video let alone deal with hundreds of requests in a split second.

To aid navigation, quite a few keyboard shortcuts and mouse controls have been added. Enough to please both keyboard and mouse jockeys alike!

That with Directshow filter control will be in the next release due shortly.

Regards,

Vent

[rebkell]

Are there any keyboard shortcuts for navigating the preview window? The trackbar control may be by design, but it jumps 2 minutes and sometimes 3 minutes, no matter how fast I click and release the mouse button. It can actually jump over an entire commercial with this behavior.

Keep up the good work, is there a new version available, you said the IDR shows up in the window, is a new beta version available for download?

[DreckSoft]

PAT and PMT really are essential to a Transport stream. If a capture program strips them out of a stream, then the capture program is at fault. TSPE would then need to repair the stream to fix the poor design of the capture program. What real life streams have you come across without PAT and PMTs? If a specific capture program is stripping out the PAT and PMT perhaps the authors could fix their software?


Offline Descrambler produces strange output. Till now I haven't found a program which can handle the output. The output of OD seems to be ok for the few pre-defined CAIDs but decrypting for example ORF1HD with Cryptoworks (0x0D05) creates a completely unusable file.

So a "fix my stream" would be a nice function.

[Ventolin]

Are there any keyboard shortcuts for navigating the preview window? The trackbar control may be by design, but it jumps 2 minutes and sometimes 3 minutes, no matter how fast I click and release the mouse button. It can actually jump over an entire commercial with this behavior.

Keep up the good work, is there a new version available, you said the IDR shows up in the window, is a new beta version available for download?
Hi, thanks, I'm looking into the trackbar sensitivity. Almost all the keyboard and mouse shortcuts are for navigation.

New public version isn't available yet, though fully paid users may have a chance to test before the next public beta is released.

Regards,

Vent

[Ventolin]

Offline Descrambler produces strange output. Till now I haven't found a program which can handle the output. The output of OD seems to be ok for the few pre-defined CAIDs but decrypting for example ORF1HD with Cryptoworks (0x0D05) creates a completely unusable file.

So a "fix my stream" would be a nice function.

Hi,

If the output is still scrambled, there isn't anything TSPE will do to descramble the output. However, if it's a case of missing tables, there is a possiblity. Table editing support is high on the priority list, though I'm not sure it will make it to the next public beta.

If you can send me a sample, I'll see if there's anything that can be done.

Regards,

Vent

[robinsj]

As long as you play

Did you mean to say something after this? Seems like an incomplete statement. Thanks for those directions for editing out audio, that should work perfectly. Next thing I know, my wife will be complaining that she can't hear it, but see can see them mouth it... haha... So just thought I would ask, is there anyway out there, to blur the mouth or anything like that. I mean, I think for the most part just cutting the audio would work great, but some videos are really close up on a face or something. I know this is out of taste for some people, and like I said before, not looking to get into arguements about what I want to edit. I seriously doubt its possible, but let me know if you have any ideas....

[Ventolin]

Did you mean to say something after this? Seems like an incomplete statement. Thanks for those directions for editing out audio, that should work perfectly. Next thing I know, my wife will be complaining that she can't hear it, but see can see them mouth it... haha... So just thought I would ask, is there anyway out there, to blur the mouth or anything like that. I mean, I think for the most part just cutting the audio would work great, but some videos are really close up on a face or something. I know this is out of taste for some people, and like I said before, not looking to get into arguements about what I want to edit. I seriously doubt its possible, but let me know if you have any ideas....

I did have some text after that line, I tried to edit the post to add it in, but looks like it didn't stick. Here's the rest of the text...

"As long as you play the file as a transport stream you will maintain sync, if you demux it you will loose sync at the edit point. To fix this, just demux in separate parts for processing then reassemble afterwards."

As for blurring of mouth, a quick hand infront of the eyes works well, as does changing the video input on the TV with a quick flick of the remote. You should know exactly where the swear words are if you cut them out!

If you want to do it in the video, you'll have to re-encode with a blur filter (using a digital compositor or AVS script for the blur), or just cut the whole GOP out (from I/P frame to I frame).

Cheers,

Vent

[robinsj]

Yeah, i figured that was gonna be the answer but thought I would ask. I honestly don't care, cutting the audio is good enough, and I absolutely don't want to re-encode. What I have been doing, is just ripping the sub titles and doing a search through those for the f words, and it gives me the timestamp right there which is within pretty much a second of when it plays... so then I just use your fine piece of software to pinpoint it and snip it out. Great detailed instructions you gave by the way. This is the reason why I was wondering if you can just enter in a time that you wanted to goto in the video? Did you add this feature, I noticed you said you were going to. Was wondering if it has been added to the downloadable file yet?

[Renzz]

I downloaded TSPE yesterday along with the trial license and managed to do a few edits ok. However, today it will no longer edit. I can open files, move through them, select edit points etc, but when I click on "Edit" to do the final edit, it always immediately fails with

Error:
Index was outside the bounds of the array. at FmX1GV7Hfyex.3rSrxSTS.3I9TgLAHWw6yjYty6cd()
PID: 0, iPii: 0, ProgIndex: 0
Warning: No identified video PID found in this section.
Total time taken 0.06 seconds, write speed 0.0 MB/s

Edit complete: successfully written file with 376 bytes.

It is doing this on the file I successfully edited yesterday as well as others. Not sure what I am doing wrong.

(Oh and BTW, the email address under Help - betatest@bitstreamtools.com - is bouncing as unknown).

[blubberbirne]

Same Problem here......

Maybe the Beta Key Expired?

[Ventolin]

Thanks for that notification, looking into it.

Vent

EDIT: I have updated the licence on the website, please redownload the licence! This will only affect trial licences, people with fully paid licences will not have an issue.

Please let me know how you get on with the new licence.

Newer version should be out very soon btw.

[Renzz]

Nope, still failing for me:

Error:
Index was outside the bounds of the array. at FmX1GV7Hfyex.3rSrxSTS.3I9TgLAHWw6yjYty6cd()
PID: 0, iPii: 0, ProgIndex: 0
Warning: No identified video PID found in this section.
Total time taken 0.09 seconds, write speed 0.0 MB/s

Edit complete: successfully written file with 188 bytes.

[Ventolin]

Nope, still failing for me:

Error:
Index was outside the bounds of the array. at FmX1GV7Hfyex.3rSrxSTS.3I9TgLAHWw6yjYty6cd()
PID: 0, iPii: 0, ProgIndex: 0
Warning: No identified video PID found in this section.
Total time taken 0.09 seconds, write speed 0.0 MB/s

Edit complete: successfully written file with 188 bytes.

Thanks, please stay tuned...

Vent

[Ventolin]

New Version 0.740!

Hi folks, as the 0.700 licence has now expired, please try the NEW version 0.740!

Please note, this version is unstable and needs testing, please do let me know how you get on. Check the website for the latest updates.

You will need both the new 0.740 (01/01/09) .rar file and the beta test licence (01/01/09) file.

Whats New:

Directshow Filter Control. - Choose your directshow video and audio decoder filters (under options menu).
Keyboard and Mouse shortcuts - see under Help - Keyboard Shortcuts.
GUI now remembers positions (including preview window)
Video Preview should now be more accurate - two modes added, Preview Mode and Auto-Preview - see below.
H.264 IDR frames now clearly identified in bottom corner status, IF+/- snaps to IDR frames if present (needs some testing)
Internal Viewer now stays active, can be full screen via keyboard shortcut.
Decode Packet window font size increased slightly.
Jump To Timecode - just type in a new timecode in the Current Time box above the preview window.


Preview Mode:
Pressing CTRL-K will do a very fast test and show the first frame in the main viewer window. This lets you quickly see if your start point is accurate (jump to an IFrame first!).

Autopreview mode (under options - timestamp mode)
This disables realtime preview and instead creates a fast test edit and displays the first frame from that. This is much slower than realtime preview and is only recommended for use if you really have inaccurate preview. (If you do, please send me a sample file!)

In almost all cases, preview is accurate with CoreAVC, Cyberlink and FFDShow filters. However, you can adjust the preview window by tweaking the "Preview Adjust" option (under options - timestamp mode)

Directshow Control:
This needs to be enabled if you want to use your own filter choices instead of the automatic filter choice. Note, incorrectly setting the filters may leave TSPE unstable!

Right clicking the video display will show the filters in use. If verbose logging is enabled (under options), you can get a dump of the directshow filter chain in use in the status window.

Hope this release works OK for you, feedback is always appreciated!


Regards,

Vent

[Renzz]

Thanks, new version working ok again for me.

A couple of things I have noticed as I use it today:

1. If I open a stream with several audio tracks (BBC HD stream for example which has both AC3 and MPEG soundtracks), it detects both when it scans the file, but only lists the AC3 one under the PID tree, and only retains this track in the edited file. How do you get it to preserve all audio tracks?

2. Opening a stream from ITV HD, it detects it as 50fps, when it is 25 - possibly due to the MBAFF encoding they use?

[Renzz]

A quick note about editing commercials out and keeping sync.

I have captured a stream from ITV-HD that I wanted to edit the commercials from, demux and encode to make it smaller, then remux the original AC3 soundtrack back in. However, on the default settings in TSPE, I could successfully do the edits and the resulting TS file played fine and in sync. I then used DGAVCDECNV to index the stream and demux the AC3 out. I then encoded the video using MEGUI to something smaller and then remuxed the AC3 back. The start of the resulting MKV was fine and in sync, but at each cut point it would go slightly out of sync, so by the end it was out by quite a lot.

I have since discovered that if I go to Edit/Clean Edit Point and untick Clean Audio Frames, it fixes this and keeps everything in sync. I do get the occasional squeak at the edit point where I guess the AC3 is slightly corrupt, but at least sync is now good.

Is this the only way I can achieve good sync when cutting h264 streams at an IDR frame?

[rebkell]

A quick note about editing commercials out and keeping sync.

I have captured a stream from ITV-HD that I wanted to edit the commercials from, demux and encode to make it smaller, then remux the original AC3 soundtrack back in. However, on the default settings in TSPE, I could successfully do the edits and the resulting TS file played fine and in sync. I then used DGAVCDECNV to index the stream and demux the AC3 out. I then encoded the video using MEGUI to something smaller and then remuxed the AC3 back. The start of the resulting MKV was fine and in sync, but at each cut point it would go slightly out of sync, so by the end it was out by quite a lot.

I have since discovered that if I go to Edit/Clean Edit Point and untick Clean Audio Frames, it fixes this and keeps everything in sync. I do get the occasional squeak at the edit point where I guess the AC3 is slightly corrupt, but at least sync is now good.

Is this the only way I can achieve good sync when cutting h264 streams at an IDR frame?

Commercials are a pain for sure, I've pretty much just started encoding the whole thing before I edit it, I still have problems with sync sometimes, but I've decided I spend less time getting a good finished product, than if I try to cut commercials before encoding, The extra time it takes to encode I figure I would spend trying to edit it out and clean it up anyway. Of course in my case I'm re-encoding back to h264 and then muxing back to a ts file, then I use either TSPE or h264ts_cutter to cut out commercials.

[Ventolin]

A quick note about editing commercials out and keeping sync.

I have captured a stream from ITV-HD that I wanted to edit the commercials from, demux and encode to make it smaller, then remux the original AC3 soundtrack back in. However, on the default settings in TSPE, I could successfully do the edits and the resulting TS file played fine and in sync. I then used DGAVCDECNV to index the stream and demux the AC3 out. I then encoded the video using MEGUI to something smaller and then remuxed the AC3 back. The start of the resulting MKV was fine and in sync, but at each cut point it would go slightly out of sync, so by the end it was out by quite a lot.

I have since discovered that if I go to Edit/Clean Edit Point and untick Clean Audio Frames, it fixes this and keeps everything in sync. I do get the occasional squeak at the edit point where I guess the AC3 is slightly corrupt, but at least sync is now good.

Is this the only way I can achieve good sync when cutting h264 streams at an IDR frame?

Hi,

Did you try to demux from within TSPE? You can edit and demux in the same operation.

Regarding the ITV-HD mpeg audio stream, it might be that the stream is there but perhaps not identified clearly. Do you have any other streams in the PID tree?

Vent

[Renzz]

Hi,

Did you try to demux from within TSPE? You can edit and demux in the same operation.

Regarding the ITV-HD mpeg audio stream, it might be that the stream is there but perhaps not identified clearly. Do you have any other streams in the PID tree?

Vent

No, I haven't tried editing and demuxing at the same time - I'll give that a go.

Regarding the MPEG audio stream (actually this is from BBC HD not ITV), the only streams in the PID tree are are the AVC video and AC3 audio. However, when you open the file initially, it does detect it:

Opening: G:\Recordings\Doctor Who at the Proms (BBC HD) 2009-01-01 13-47-02.ts
Opening File...
Video Loaded.
Auto Finding PIDs
Debug: PMT Version Change Detected near 0x6A414
Debug: PMT Version Change Detected near 0x1DF8C
Debug: PMT Version Change Detected near 0x6A414
PID Attribute List:
PID: 0x0918: H.264 1440x1088 (1920x1080) @ 25.000 fps
PID: 0x0919: AC3: 3/0.1: L, C, R, LFE 44.1KHz 256Kbit/s
PID: 0x091A: MPEG1 Audio Layer 2, 64Kbit/s @ 44100Hz

Scanning For Format Changes
File size: 8707612356 Bytes
Video Resolution from Codec: 1440 x 1080 @ 25.00 fps
Duration from Codec: 4077.44
Duration from Timestamps: 4077.56
Average System Bitrate: 17083966.60
Done.

[Ventolin]

No, I haven't tried editing and demuxing at the same time - I'll give that a go.

Regarding the MPEG audio stream (actually this is from BBC HD not ITV), the only streams in the PID tree are are the AVC video and AC3 audio. However, when you open the file initially, it does detect it:

Opening: G:\Recordings\Doctor Who at the Proms (BBC HD) 2009-01-01 13-47-02.ts
Opening File...
Video Loaded.
Auto Finding PIDs
Debug: PMT Version Change Detected near 0x6A414
Debug: PMT Version Change Detected near 0x1DF8C
Debug: PMT Version Change Detected near 0x6A414
PID Attribute List:
PID: 0x0918: H.264 1440x1088 (1920x1080) @ 25.000 fps
PID: 0x0919: AC3: 3/0.1: L, C, R, LFE 44.1KHz 256Kbit/s
PID: 0x091A: MPEG1 Audio Layer 2, 64Kbit/s @ 44100Hz

Scanning For Format Changes
File size: 8707612356 Bytes
Video Resolution from Codec: 1440 x 1080 @ 25.00 fps
Duration from Codec: 4077.44
Duration from Timestamps: 4077.56
Average System Bitrate: 17083966.60
Done.

Thanks for that, looks like there are still some bugs in that then - not sure BBC would be outputting 3/0.1 @44.1KHz AC3.

Can you send me a 10MB (or more) sample from the start please?

Also, a screenshot of the PID tree would be good so I can see whats there (if you want a faster response that is, if you send me a sample, I'll know for sure).

Cheers,

Vent

[Renzz]

I've put a 100MB sample here - http://www.mediafire.com/?sharekey=014635695f2c1f3bd2db6fb9a8902bda

Cutting smaller chunks seems to have unpredictable results with TSPE sometimes detecting streams and sometimes not. Looks like the longer the better.

[Ventolin]

I've put a 100MB sample here - http://www.mediafire.com/?sharekey=014635695f2c1f3bd2db6fb9a8902bda

Cutting smaller chunks seems to have unpredictable results with TSPE sometimes detecting streams and sometimes not. Looks like the longer the better.

Thanks for that, my bad I should have mentioned earlier, can you send me the sample with both Fix Tables and Fix Timecodes turned off (if you haven't already)?

The sample you sent is good though as it has picked up a different bug, thanks!

The "unpredictable behaviour" may be due to the scan that happens when you open a file isn't a full scan - that only happens when you press the "scan" button. When you open a file, it's mostly looking for the PIDs, scanning for ad breaks (format changes) and doing a few other housekeeping tasks.

To make sure every PID is included in the output, make sure every PID in the PID tree is *unchecked*, then set start/stop points and add to the EDL. If you notice, when you add something to the EDL normally, you'll see a PID list as the last entry for that row. If that PID list is empty (when you turn off all the PIDs in the PID tree), TSPE will not strip any PID from the stream.

Cheers,

Vent

PS. If anyone has the latest coreavc codec and wants to use that with directshow control, there's a bug that prevents it being used. I've already fixed that bug and will be available in the next release. Sorry bout that!

[Renzz]

Thanks for that, my bad I should have mentioned earlier, can you send me the sample with both Fix Tables and Fix Timecodes turned off (if you haven't already)?

Not sure what you mean by this - the sample is just a part of the raw captured TS, not the output from TSPE.

To make sure every PID is included in the output, make sure every PID in the PID tree is *unchecked*, then set start/stop points and add to the EDL. If you notice, when you add something to the EDL normally, you'll see a PID list as the last entry for that row. If that PID list is empty (when you turn off all the PIDs in the PID tree), TSPE will not strip any PID from the stream.

Yes this works, but is a little messy - you have to remember to do this for everything you edit. Would be much better if it could just detect all the streams when it opens the file.

Thanks.

[Ventolin]

Not sure what you mean by this - the sample is just a part of the raw captured TS, not the output from TSPE.



Yes this works, but is a little messy - you have to remember to do this for everything you edit. Would be much better if it could just detect all the streams when it opens the file.

Thanks.

OK, I think I may understand what you mean.

In the sample you provided, when you open the file, the PID tree lists:

PAT PID: 0x0000 Ver: 1 TS ID: 0x0802
0x0102: PMT PID: 0x0102 Ver: 2 Program : 6940
0x0918: H.264/AVC Video (as defined in ITU-T Rec. H.264 | ISO/IEC 14496-10 Video)
0x091A: MPEG1 Audio (ISO/IEC 11172-3 Audio)
0x091E: Unknown
0x0919: AC-3 Audio (DVB) Stream

Every PID in the list is checked, which means it will be included in the output.

Note: the Unknown stream there is actually a Subtitle stream which isn't decoded yet.

The status log shows:

PID Attribute List:
PID: 0x0918: H.264 1440x1088 (1920x1080) @ 25.000 fps
PID: 0x091A: MPEG1 Audio Layer 2, 64Kbit/s @ 44100Hz
PID: 0x0919: AC3: 3/2: L, C, R, SL, SR 48KHz 96Kbit/s

which also shows a bug in the packet decode section (I'm working on that).

So with the sample you sent, I don't see the problem you mentioned.

However, in your previous post, you said:
Auto Finding PIDs
Debug: PMT Version Change Detected near 0x6A414
Debug: PMT Version Change Detected near 0x1DF8C
Debug: PMT Version Change Detected near 0x6A414

PMT version changes can mean PIDs and streams can change. At the moment the first PMT is used as a base for the whole stream as I haven't come across a file with PMT version changes that do change the PIDs mid-stream. So it could be the tree is based on a PMT that doesn't include all the streams in the programme that a later PMT might have.

So, I would really need the file you were working on to get that PMT version change to figure out how to handle multi-version PMT streams.

So, can you send me a sample of the Dr who proms .ts that shows the PMT version changes? From the byte positions it looks like it's near the start of the file, but if you can send the whole stream that would be very useful (though might not be possible as I know it's 9GB but PM me if you can send it). Please turn off the two checkboxes "Fix Timecodes" and "Fix Tables" next to the edit button.

Note, in your case, as it's BBCHD there are no adverts, so you don't need to even have "Fix Timecodes" or "Fix Tables" turned on. This will just trim the file to the start/end points you have set. Those checkboxes are remembered when you restart TSPE so if you are only editing BBCHD you can turn them off once.

Quick explanation of the two checkboxes:

Fix Timecodes - turns on all timecode fixing options. This is required if you are doing anything more than just a simple trim.

Fix Tables - turns on PAT and PMT fixing. This is required if you want to strip streams out and/or fix missing streams in the PAT and PMT (left overs from other apps that didn't fix this).

In the meantime, I'll get working on fixing the decode and identification bugs.

Cheers,

Vent

[Renzz]

Hi Vent,

Now that's interesting - that 100MB sample definitely showed the problem of the mpg stream not appearing in the PID tree - I wonder why it worked for you?

Anyway, I have done it again, using a different file cutter. I have again cut the first 100MB and this time it definitely shows the problem, and also shows the PMT changes as well. Just so you don't think I'm making it up (:)), here's a screen shot of what I get when I open this file:

http://img.photobucket.com/albums/v357/lauried/tspe1.jpg

The 100MB sample has been uploaded to http://www.mediafire.com/?xm52mm2ahjm

See if you now get the same thing with that.

Cheers

[Ventolin]

Hi Vent,

Now that's interesting - that 100MB sample definitely showed the problem of the mpg stream not appearing in the PID tree - I wonder why it worked for you?

Anyway, I have done it again, using a different file cutter. I have again cut the first 100MB and this time it definitely shows the problem, and also shows the PMT changes as well. Just so you don't think I'm making it up (:)), here's a screen shot of what I get when I open this file:

...

The 100MB sample has been uploaded to http://www.mediafire.com/?xm52mm2ahjm

See if you now get the same thing with that.

Cheers

Awesome, I can reproduce the issue here now...

The PMT version change is exactly what is causing the issue - you can see this if you jump to the first PMT where the version change is detected - it will be Version 2 and if you look at the PMT decode it will have all the streams listed, but if you jump back 1 PMT (to the other byte location indicated with a version change) you can see there are only the H.264 and AC3 streams.

Infact, there is only 1 PMT with version 1, the other PMTs in the stream are all version 2.

I notice that the PMT version change happens in the middle of the programme, which is not such a great thing from the BBC's side of things (if they are responsible), they should do version changes inbetween programmes, not during the middle of one!

So, either the BBC is being "naughty" or the capture programme or tools used to create the sample have changed
the first PMT incorrectly.

Since there's only 1 PMT with version 1, I think it could be the latter causing the issue. Is this at the very start of the recording? if you have any other streams that have a PMT version change, check if there is only 1 PMT with the lower version - if that is the case, then I suspect the Capture Software or other utils used is at fault. Otherwise, if there is more than 1 PMT with a lower version, a sample would be most useful!

Nonetheless, since PMT version changes are within spec and so is changing the PIDs mid stream, TSPE should be able to handle it, so I'll get working on that, but this is definitely non-trivial!

The workaround is to simply remove this first PMT from the stream and re-process. You can do this by removing this packet with the EDL - set start/end points around the offending PMT and make an edit. Or you could just chop off the start with the PMT included if you are not in the middle of the programme you want, just make sure to turn off "Fix Tables" and uncheck the menu option "Edit-EDL Options-Insert Tables At Start Only?"

Thanks again for the sample files, they are really useful!

Cheers,

Vent

[Renzz]

It could well be the capture software - I use DVBViewer, and that has an option in the recording setting called "Adjust PAT/PMT" which I have set. I find I need this to make the ITV HD streams play on my Popcorn Hour. The documentation for that option says:

"If activated this option causes the DVBviewer to generate a PAT and PMT in TS recordings, which does reflect the present streams.
The PAT only contains the recorded service and the PMT only the recorded ES Streams.
This option may help with playback hardware/software which is not able to cope with the unchanged Tables of a TS."

[Ventolin]

It could well be the capture software - I use DVBViewer, and that has an option in the recording setting called "Adjust PAT/PMT" which I have set. I find I need this to make the ITV HD streams play on my Popcorn Hour. The documentation for that option says:

"If activated this option causes the DVBviewer to generate a PAT and PMT in TS recordings, which does reflect the present streams.
The PAT only contains the recorded service and the PMT only the recorded ES Streams.
This option may help with playback hardware/software which is not able to cope with the unchanged Tables of a TS."
Ahh, DVBViewer, now it makes sense!

Try turning that option off for everything apart from ITVHD. I'm already working on getting ITVHD to work from streams created by DVBViewer without a PAT or PMT, so you'll be able to process those completely inside TSPE.

Let me know how you get on with BBCHD without the DVBViewer PAT/PMT option set.

Cheers,

Vent

[Cela]

I really appreciate TSPE.

But navigation can be quite confusing with certain DVB-S2 captures for a newbie like myself. The behaviour can be quite unexpected, non-linear, anisotropic at certain points in the TS.

Please refer to the example which I uploaded to ArteHD_Test1 (http://rapidshare.com/files/181313167/ArteHD_Test1.rar) (~100MB). As all my captures, this example was captured with DVBViewer (with "Adjust PAT/PMT" set) and follows Ventolin's scheme from post #14 starting before the start of the actual programme.

Navigation to the Start Point at 00:00:36.14 gets somewhat confusing when trying a step by step approach. Especially just before the actual start of the programme navigation becomes really weird: very non-linear and anisotropic. -IF/IF+, -F/F+ are unsymmetric, etc., and there is a considerable difference between what I see in the Play and Preview window.

If I pause at 36.08 and then press PLay, I wait for sound and video until about 01:02.00 and end up far off the Start Point of the film. In these occasions I cannot rely on the Play window since it does not show what I will get! I have to navigate practically blindly by trial and error.
The Preview window is helpful, though, especially the fast preview by CTRL-K. But it should not be necessary. As a simple user I suppose to see what I will get.

The Play window must be as reliable as the view through the front screen of my car! If the controls of my car behaved in such a strage way, it would be a lot more difficult to drive savely. Why don't I see what I will get in the Play window? Why can't "-IF" after "IF+" get me back to where I started? Very confusing for a NOOB.

Just as before and around the beginning of the film, also at and around the end of the film, some GigaBytes later in the TS, the strange behaviour again makes it difficult to navigate exactly to the proper end of the film. Some times IF+ navigates a large step backwards ( ! ) from somewhere near the end of the TS towards the middle of the TS, and the next IF+ even further towards the middle or even the beginning of the TS... Very strange, indeed.

Maybe a change in strategy could help: reliable (mpeg2-) programmes (Mpeg2Schnitt, Cuttermaran,...) and DGAVCDec/DGAVCDecNV (with DGAVCIndex/DGAVCIndexNV) make an initial indexing step to support reliable navigation.
I would be glad to wait a minute or two till the necessary indices are established, when this would lead to real WYSIWYG behaviour for all streams and all parts of a stream.

I hope this matter can be sorted out before expiration of the free trial. Then IMHO TSPE would be the best of the programmes for TS I-frame cutting I was able to test.

[Ventolin]

Hi Cela,

Thank you for your sample, and your feedback!

I see 2 main issues here:

1. Accurate Preview
2. Navigation issues

1. Accurate Preview.
The major cause of inaccurate preview is the Direct Show Splitter that is in use. It's the job of the splitter to provide frames to the Video Decoder which then has the duty to reassemble the frames and decode them in the right order to provide the requested frame.

a. The splitter:
We have found Haali splitter to do a pretty good job for seeking. Please try installing or re-installing Haali and seeing if that helps. You should get the Haali icon appear in your system tray. You can download Haali from:

http://haali.csu/mkv/ (http://haali.cs.msu.ru/mkv/)

b. The Video Decoder:
We have had good results with CoreAVC, Cyberlink Video Decoder and FFDShow direct show filters. The latest FFDShow tryout has a few issues with seeking to the exact frame in a GOP, but this can be compensated for from within TSPE using the Preview Adjust option under Options->Timestamp Mode. It is possible to get the exact frame with the filters recommended which can be seen by quickly using the Preview Mode Ctrl-K as you have been trying.

So, let me know how you get on with the preview with Haali installed/re-installed.

2. Navigation issues.
Your sample is very good indeed as it has shown a bug in the decoding and navigation systems. Thankfully, this is already fixed in the dev build I'm working on at the moment. The next release will have the fixes in place.

However, even with version 0.740 I hope that when you ensure Haali is being used, you will still have accurate or pretty close to accurate video preview. The IF+/- and F+/- buttons still move to valid locations, but the packet decode may not decode the I frames correctly.

Thanks again for your feedback. Please do let me know how you get on.

Regards,

Vent

[Cela]

Hi Vent,

Thank you for your quick answer! :)

These are my current settings (http://rapidshare.com/files/181551937/ArteHD_test_TSPE_current_settings.jpg). What is the complete set of suggested settings?
I am looking forward to the next beta version.

Here I have another task for TSPE concerning I-frame editing and merging.

Last year I recorded two different broadcasts of Mahler's 7th Symphony from the nightly Astra HD demo broadcasts of Unitel Classica HD.

Both captures are incomplete: Capture A stopped too early, capture B started too late.
Now I want to combine both captures at the overlap at the beginning of the 5th part of the symphony, one -IF before the text "V.SATZ RONDO-FINALE" can be seen in the preview window.
For the test I used TSPlayer 2.0.0 to crudely cut two 1 minute chunks around the intended assembly point:

AstraHDTest_A.ts represents the first part of the symphony up to the intended assembly point, just before the start of part 5 of the symphony.
AstraHDTest_B.ts represents the second part including the assembly point.

CutIn of AstraHDTest_A.ts shall be one IF+ after the beginning of that TS.
CutOut of AstraHDTest_A.ts shall be one -IF before the text "V.SATZ RONDO-FINALE" can be seen in the preview window.

CutIn of AstraHDTest_B.ts shall be also one -IF before the text "V.SATZ RONDO-FINALE" can be seen in the preview window.
CutOut of AstraHDTest_B.ts shall be one IF before the end of that TS.
(If I drag the slider too quickly to the end, or is it beyond the end, it does not remain at the end of the TS, instead it jumps to somewhere near the middle! It remains at the end, when I do it very carefully, though.)

The test material can be found in AstraHDTest_AB.rar (http://rapidshare.com/files/181537290/AstraHDTest_AB.rar) (145002 KB ):
AstraHDTest_A.ts (82964 KB) ... input part A
AstraHDTest_B.ts (67042 KB) ... input part B
AstraHDTest_AB.edl (1 KB) ... EDL of the two cuts
AstraHDTest.log.txt (1 KB) ... TSPE log with some comments

Question 1:
According to tsMuxeR 1.68.4b there are the following intrinsic Audio Delays (AD) in the input TS:
AstraHDTest_A.ts: AD=-426ms
AstraHDTest_B.ts: AD=-404ms
How to get rid of these two individualADs?

Question 2:
The TSPE log comments the addition of the second part as follows:
Processing EDL Item: 2 of 2
New PAT Generated.
New PMT Generated.
Detected framerate 50,000 fps
Video PTS gap of 65998,0 seconds fixed at 00:01:15 in output file, 0x17060C4.
Info: Audio PTS gap of -3190ms found in PID 0x0503, video PTS gap is -60ms, temporarily not synced
...
Where does this additional Audio Delay of -3190 ms for the second line in the EDL come from?
How can it be avoided ???
It is best seen at the moment where the drum starts part 5 of the symphony near 00:01:28.20 of the resulting AstraHDTest_AB.ts and heard approx. 3590 ms later.
The ADs definitly has to be removed!

Question 3:
During experimenting with TSPE I got the impression that changes of the Edit Options are not saved.
Changes like
Edit > Timecode Options > Zero Timecodes? ON
Edit > Timecode Options > Sync Offset xxx
Edit > EDL Options > Insert Tables At Start of Each EDL Item? ON
get lost after ending TSPE and are reset to default at the next start of TSPE.
Is that correct? I would prefer that the changes are saved.

Question 4:
I did not notice any effect of setting a sync offset
Edit > Timecode Options > Sync Offset 'value'
Atually there should be an individual AD for each input TS!

I hope, this example may help to test and validate TSPE's I-frame editing and merging capabilities.

Regards,

Cela

[Ventolin]

Hi Cela,

Have you get accurate preview now?
What is the complete set of suggested settings?
You need to turn Clean Edits -> Clean Audio ON (checked). You can also reset all the options to the defaults under the Options menu for the recommended settings.

(If I drag the slider too quickly to the end, or is it beyond the end, it does not remain at the end of the TS, instead it jumps to somewhere near the middle! It remains at the end, when I do it very carefully, though.)

Yeah, I have noted that issue already.

Question 1:
According to tsMuxeR 1.68.4b there are the following intrinsic Audio Delays (AD) in the input TS:
AstraHDTest_A.ts: AD=-426ms
AstraHDTest_B.ts: AD=-404ms
How to get rid of these two individualADs?

You only need to worry about these if you are demuxing and are not correcting the sync at the same time (by default TSPE automatically makes the sync offset 0).

Question 2:
...
Where does this additional Audio Delay of -3190 ms for the second line in the EDL come from?
How can it be avoided ???
It is best seen at the moment where the drum starts part 5 of the symphony near 00:01:28.20 of the resulting AstraHDTest_AB.ts and heard approx. 3590 ms later.
The ADs definitly has to be removed!
Turn on Clean Audio to fix. Note, the first drum beats in part B is out of sync in the *source*.

Question 3:
During experimenting with TSPE I got the impression that changes of the Edit Options are not saved.
Changes like
Edit > Timecode Options > Zero Timecodes? ON
Edit > Timecode Options > Sync Offset xxx
Edit > EDL Options > Insert Tables At Start of Each EDL Item? ON
get lost after ending TSPE and are reset to default at the next start of TSPE.
Is that correct? I would prefer that the changes are saved.
These options are not necessary for correct operation.

Question 4:
I did not notice any effect of setting a sync offset
Edit > Timecode Options > Sync Offset 'value'
Atually there should be an individual AD for each input TS!

(Re)Sync Offset appears to be broken! Thanks, will fix it.

Try your edit with clean edits on - you should find you get a perfect edit!

Regards,

Vent

[Cela]

...I see 2 main issues here:
1. Accurate Preview
2. Navigation issues
...
a. The splitter:
We have found Haali splitter to do a pretty good job for seeking.
...
b. The Video Decoder:
We have had good results with CoreAVC, Cyberlink Video Decoder and FFDShow direct show filters.
...
In all tests I use the Haali Media Splitter 1.8.122.18
ffdshow Audio Decoder (from ffdshow_beta4_rev1685_20071206_clsid.exe)
CoreAVC Video Decoder (from CoreAVC PE 1.8.5.0)
AC3Filter (from ac3filter_1_51a.exe)
as documented in ArteHD_test_TSPE_current_settings.jpg stored in AstraHDTest_AB.rar.
And I still get the navigation issues reported in post #60.
You should be able to reproduce theses issues with my test material in ArteHD_Test1.rar

...
2. Navigation issues.
...However, even with version 0.740 I hope that when you ensure Haali is being used, you will still have accurate or pretty close to accurate video preview. The IF+/- and F+/- buttons still move to valid locations, but the packet decode may not decode the I frames correctly.
...As stated before, with version 0.740 and the recommended newest Haali splitter and newest CoreAVC codec I get the issues that I reported in my post #60. I still experience these issues with practically all my captures from Astra HD, Arte HD, Anixe HD and ORF1 HD. Unfortunately, the largest and most disturbing deviations between Play and Preview occure near the points of interest around the beginning and ending of programmes. Same applies for the IF+/i issues.

...
Have you get accurate preview now? No.
...You need to turn Clean Edits -> Clean Audio ON (checked). You can also reset all the options to the defaults under the Options menu for the recommended settings...
Turn on Clean Audio to fix.OK. Done.
No influence on accurate preview, though.
Fixed the additional Audio Delay of -3190 ms.

...
Note, the first drum beats in part B is out of sync in the *source*. Yes, that is what I said:
According to tsMuxeR 1.68.4b there are the following intrinsic Audio Delays (AD) in the input TS:
AstraHDTest_A.ts: AD=-426ms
AstraHDTest_B.ts: AD=-404ms
But, you are right, I took care of that AD with tsMuxeR. Visual inspection showed that there still remains an AD of approx. -500ms in the *source*.
Thus the following total AD must be taken care of:
AstraHDTest_A.ts: AD=-426ms-500ms = -926ms
AstraHDTest_B.ts: AD=-404ms-500ms = -904ms
Again, I applied these total ADs with tsMuxeR. Then i used the AD corrected TS from tsMuxeR in TSPE with "Clean Audio to fix" turned on. Now the additional Audio Delay of -3190 ms is also gone. Now I get perfect sync with excellently synced drum beats in the second part. And that is not only true for the small test segments but also for the full 6GB symphony! :)
TSPE is a charm!

...
Try your edit with clean edits on - you should find you get a perfect edit!True. Confirmed. :)

Using "Restore Default Options" I found an issue.
That is what I did:
> Start TSPE
> Open video file
> Options > Restore Default Options? OK
> Continue navigating video file
=> some error messages:
All Options Reset to Defaults. <------------------------------------------------------------
Could not find Timestamp.
Warning, Timestamp not found

Error: Decoding PAT
Die arithmetische Operation hat einen Überlauf verursacht. <---- arithmetic overflow

Error: Displaying PAT
Der Objektverweis wurde nicht auf eine Objektinstanz festgelegt. <----- another error (use Google to translate; I can't)
Could not find Timestamp.
Warning, Timestamp not found

What helps?
===> After "Restore Default Options" restart TSPE!
Please, issue a warning to the user.

Conclusion of my current tests:
1. + TSPE is on a very good way.
2. + Perfect edit.
3. Please, improve navigation and WYSIWYG !
4. + Preview with CTRL-K and preview with "Test" is very helpful and ergonomic to handle.
5. Please, allow AD corrections for each individual input TS
6. Please, add a "Remark"-field/column to the Markers-list (for user input of text)
7. Please let the user add Markers with text input and load/save the contents of the marker-list from/to a textfile.
(I am used to thos luxury from Mpeg2Schnitt !! )

===> I am looking forward to the improvements of the next beta version (as I already stated in post #62).

Thank you for your quick and kind help.

Regards,

Cela

[Ventolin]

Hi Cela,

Inaccurate Preview is a directshow filter issue. Please try uninstalling Haali and then re-installing Haali. You might also like to try the very latest version of Haali, 1.9.42.1.

You should also update your FFDShow version to the latest tryout. I have tryouts version 1920, March 2008 and tryouts version 2527 Dec 17 2008.

As I have said before, I have accurate preview here with your samples with haali and CoreAVC / Cyberlink / FFDShow filters.

I will try and see if I can offer more functionality to the Directshow Control system in TSPE to enable you to further troubleshoot your directshow issue, though I don't know if this will be available for the next release. I could also have a look at your setup through VNC and fix your issue that way, but I normally only offer this for paying customers.

In the meantime, there is another way to get preview which is to use "Auto Preview" which is under Menu Options -> Timestamp Mode -> Auto Preview. This is a non-realtime system as it makes a short test file and then shows the first frame. This means you can not "scrub" by moving the slider and seeing video move in realtime. You will get video when you release the slider however, and with any navigation control, but the update will take about a second. This is only intended for troubleshooting purposes however as normal preview is very accurate once directshow issues are resolved.

Regarding your Audio Delays, you do not need to adjust these to get good edits. Just make sure clean audio edits is on. However, the bug in Sync Offset is now fixed so you can resync your 500ms delay all within TSPE.

Note, setting your delays to 0 for normal transport streams may actually add desync to your playback! This is because the player needs to buffer audio and video separately and may not be able to present the audio at the same time as the video. Therefore you only need to use the Sync Offset option if you have desync in the source files as with the drums in the sample you sent.

Thanks for the bug report regarding Restore Defaults, I'll look into that.

Marker List comments is already on the todo list as is importing from other editors and utilities. It's not high on the priority list though so may not appear in the next release. The EDL and Marker list is due for an update when the Timeline system is incorporated in a future version.

Let me know how you get on after you have uninstalled/reinstalled Haali.

Regards,

Vent

[canTsTop]

TSPE cant scan this TS video http://www.mediafire.com/?kynimztnjwe

Scan performed on: Mon, 12 Jan 2009 20:05:00 GMT

Start Code Error Found at offset 0x17FE50
Aborting
Filter Info:
Video Resolution from Codec: 544 x 576 @ 25.00 fps
Approximate System Bitrate: 241633971827624,00 Mbits/s
Approximate Duration: 00:03:35.20

Video Duration by PTS: 00:00:04.400
Filesize: 00,06GB Average bitrate 118,18 Mbps

PID: 0x0321: H.264 544x576 (792x576) @ 25,000 fps
111 Total Video Frames found
19 Total Audio Frames found

Continuity Errors: 0, Transport Error Indicator Errors: 0
Timestamp gaps: 0, PCR gaps: 0
Total number of separate discontinuous regions: 0

End Of Log.

[Ventolin]

TSPE cant scan this TS video http://www.mediafire.com/?kynimztnjwe

Scan performed on: Mon, 12 Jan 2009 20:05:00 GMT

Start Code Error Found at offset 0x17FE50
Aborting
Filter Info:
Video Resolution from Codec: 544 x 576 @ 25.00 fps
Approximate System Bitrate: 241633971827624,00 Mbits/s
Approximate Duration: 00:03:35.20

Video Duration by PTS: 00:00:04.400
Filesize: 00,06GB Average bitrate 118,18 Mbps

PID: 0x0321: H.264 544x576 (792x576) @ 25,000 fps
111 Total Video Frames found
19 Total Audio Frames found

Continuity Errors: 0, Transport Error Indicator Errors: 0
Timestamp gaps: 0, PCR gaps: 0
Total number of separate discontinuous regions: 0

End Of Log.
Hi,

Start Code Error Found at offset 0x17FE50

Your stream is damaged! Fix the start code errors first by doing:

1. Select Menu Edit -> Align To Start Codes?
2. Select the source file
3. Select the output destination (will have extension ".fixed.ts" added to the filename)
4. Wait for fixed output file to be created.
5. Load fixed file into TSPE again.

You should now be able to further process the stream.

Hope that helps,

Regards,

Vent.

[canTsTop]

1. Select Menu Edit -> Align To Start Codes?
2. Select the source file
3. Select the output destination (will have extension ".fixed.ts" added to the filename)
4. Wait for fixed output file to be created.
5. Load fixed file into TSPE again.


after that then i open ".fixed.ts" i get:
Opening File...
Unknown File Type, aborting

[Ventolin]

after that then i open ".fixed.ts" i get:
Opening File...
Unknown File Type, aborting

This can happen if a stream is too badly damaged to fix, but I'm looking at it now.

Vent.

[Ventolin]

Hi, canTsTop,

I think I can fix your stream but need to tweak some code here, hopefully will have it working by the next version. Hope you can wait till then.

Cheers,

Vent

[Cela]

... Inaccurate Preview is a directshow filter issue. Please try uninstalling Haali and then re-installing Haali. You might also like to try the very latest version of Haali, 1.9.42.1.

You should also update your FFDShow version to the latest tryout. I have tryouts version 1920, March 2008 and tryouts version 2527 Dec 17 2008.

As I have said before, I have accurate preview here with your samples with haali and CoreAVC / Cyberlink / FFDShow filters.

Let me know how you get on after you have uninstalled/reinstalled Haali.I strictly followed your recommendations and de-installed Haali and re-installed Haali 1.9.42.1. Then I de-installed FFDShow and re-installed FFDShow tryouts version 2527 Dec 17 2008 (there is a newer one available, though, ffdshow_rev2602_20090109_clsid.exe, which I did not install).
Preview accuracy and navigation has improved, indeed. :)
But preview still is slightly inaccurate in some positions and navigation sometimes ambiguous (where you get to may depend an where you were before the move).

Examples with test file "ArteHD_Test1.ts":
> copy test string "00:00:36.08" (without the " ) by CTRL-C/CTRL-V into the Current Time field and then press RETURN -> 00:00:36.06 ???
> manually correct "00:00:36.06" to "00:00:36.08" RETURN -> 00:00:36.14 ???
=> two different results for two different ways of putting "00:00:36.08" into field Current Time.
> another example: drag slider to start and get "00:00:00.00", then copy/paste "00:00:42.24" to get "00:00:42.24", then manually change 24 to 26 to have 00:00:42.26 in field current time and press RETURN -> 00:00:42.32
But playing from the new position works instantly (withot waiting for many seconds as before). That is a good improvement.
> Drag slider to the end to get position "00:02:19.00"; then copy/paste "00:00:42.24" into current time field, press return -> 00:00:42.26 ! What you get can depend on where you were before the move?

> enter text-string "00:00:52.46" into Current Time by CTRL-V, RETURN -> 00:00:52.44
> then pressing a series of F- -> 00:00:52.42, 00:00:52.46 ?, 00:00:52.38, 00:00:52.36, 00:00:52.40 ?, 00:00:52.32, 00:00:52.30, 00:00:52.34 ?, ...
> similar up-and-down behaviour with F+ sequence ...

> drag the slider just below Start Point Set, then click below End Point Set -> the slider jumps too far right; click 1 cm to the right of slider position -> slider jumps to the end, ...

... I could also have a look at your setup through VNC and fix your issue that way, but I normally only offer this for paying customers.I am starting to like that idea. I am ready to buy a license first, if its that what you mean with paying customer?

... Regarding your Audio Delays, you do not need to adjust these to get good edits. Just make sure clean audio edits is on. However, the bug in Sync Offset is now fixed so you can resync your 500ms delay all within TSPE.Is now fixed in your version, but not in mine. When will you release the next beta?
If part A and B had two different AD, i.e -400 ms part A and -500ms part B, can I resync these individual ADs for the two input ts in the same TSPE session, or do I need to resync each ts first and then go on with the adding part B to part A?

... Note, setting your delays to 0 for normal transport streams may actually add desync to your playback! This is because the player needs to buffer audio and video separately and may not be able to present the audio at the same time as the video. Therefore you only need to use the Sync Offset option if you have desync in the source files as with the drums in the sample you sent. Is there a pragmatic "best" value of AD for good replay? If yes, shouldn't TSPE resync all outputs to this value?
Should video or audio be early?

What is the best strategy to find the real AD. As with the drums, I am not so good in "seeing" the faulty AD. I transcode a small part to mpeg2 and use Mpeg2Schnitt to test with different AD values. Is there a quicker strategy?

... ...next version...ETA?

Thank you very much for your kind help!

Regards,

Cela

P.S.
Please provide an "Undo" against erranously pressing "Clear" instead of "Del" a single EDL line. Just happend to me. :(

[Ventolin]

Hi gallis,

from your other thread;

when I open with TSpes I give this when I browse
Could not find Timestamp.
Could not find Timestamp.
Could not find Timestamp.
Could not find Timestamp.
Could not find Timestamp.

This is a non-critical error message, TSPE should still work fine, though it should be fixed in the next release.

There is no need to remux in TSMuxer. You can demux in TSPE if you want also.

Regards,

Vent

[Ventolin]

Hi Cela,

Very happy to hear you have accurate preview now!

The navigation behaviour is correct, check what happens to the byte and packet positions when you frame step.

You will find that frames are non-sequential in the stream, i.e. they are out of order. This is because Video compression uses prediction by references frames in the future and in the past. These frames are called B frames (Bi-Directional).

TSPE is the only editor / analyser available that lets you see the frames in their true positions in the stream!

When you edit, TSPE automatically works out which frames to include and which frames to exclude at the edit point even though the frames are out of order. So you can have care free editing with Clean Edits!

With the random seek, TSPE is very fast at seeking to the timestamp though is sometimes slightly out. You can see how fast TSPE is by turning the video off and moving the slider. The screen update is actually limiting the speed at which TSPE seeks at. This is especially true for VBR streams (Variable Bitrate), where you can not just jump to a position based on average bitrate. Compare how long it takes for the Video to get updated when you seek a VBR stream compared to a CBR stream (Constant Bitrate) with the same codecs. The CBR stream will show you how long it takes for a video frame to be presented and difference with the VBR stream will show you how long the delay is to find the frame requested.

Sequential timestamp framestep will probably appear in a later version when video frame accurate editing is in place.

Audio Delays - once again, you do not need to fix the Audio Delays. TSPE will fix these for you when you edit. However, if you really want to change the delay, you will have to process part A and part B separately as the "Sync Offset" is a global offset applied to the whole edit.

There is no "best" delay to use though, the streams' original delay is fine.

You can test the effects of Sync Offset with the "Test" buttons and check the effects of an edit point by right clicking 2 entries in the EDL and selecting "Test Edit Point".

As for resyncing out-of-sync sources, that is a different topic altogether.

If you are doing this visually, this is what I do:

1. Set a start offset, say -100ms.
2. Create a Test Edit
3. Check sync
4. Adjust by 20ms
5. Check sync
6. Repeat step 4 and 5 until you reach +100 ms
7. You should hopefully find a small range of values which "look" right. Eg. -40ms to +20ms.
8. The correct sync will be the average, eg. -10ms.

Audio waveform display is planned for a future version, which will allow you to visually sync the waveform to the picture.

ETAs are hard to give, though we are trying to keep a regular update schedule. As Sync Offset is already fixed though, you will be happy to hear that it will be in the next release, which really won't be too far away!

Regards,

Vent

[eggtimer]

To resync, I do this (at least, I did when sync was working before - hope you get the next release out soon!)

1. Play original file in Zoom Player, use 'Shift +' and 'Shift -' keys to change audio sync on the fly (note: Zoom Player Internal Audio/DSP Filter must be enabled for this to work).

2. Note the sync which is needed to make the file perfect

3. Edit in TSPE, applying the sync to the file

(do a test edit to make sure I have the correct +/- as I can never remember which is which!)

[canTsTop]

then i load TS in DGAVCIndex, until commercial ends i count 23 steps (pressing > button), according to http://neuron2.net/dgavcdec/DGAVCIndexManual.html#Navigation
Use the navigation slider, and these two buttons for navigation:

< - Seek to previous GOP (IDR frame or recovery point frame).
> - Seek to next GOP frame.
in TSPE same TS until commercial ends i count 10 steps - pressing IF+

can i correctly cut video only at IDR frames, or GOP (like DGAVCIndex navigates) also good for correct cutting?

[neuron2]

DGAVCIndex now steps to IDR, I, or recovery point, so the inclusion of I may account for the difference.

[Cela]

...Very happy to hear you have accurate preview now!Yes, with a little practicing, it is fairly comfortable! :)

Last weekend, I successfully cut the highlights out of a ski-racing event from the weekend before, with 11 EDL-lines. Was fairly easy to do and worked out well.

...ETAs are hard to give, though we are trying to keep a regular update schedule. As Sync Offset is already fixed though, you will be happy to hear that it will be in the next release, which really won't be too far away!


Please allow me some more recommendations for the next release:
1. Please remember size and position of Preview window (floating TSPE internal viewer) between starts of TSPE. Now, for HD material, everytime the first "Set > Test" completely fills the monitor with the Preview window overlaying over everything else. Forces extra steps to get it back into the position and size which I established and liked last time.

2. Please put "Clear" far away from the other options.
3. Please provide an "Undo" for "Clear" and "Del" (and other actions). The more EDL there are the higher is the risk to loose a lot of work with one erraneous "Clear".

4. Please provide an easier way to pre-test the future "Test edit point". I do not like to cut and glue in the middle of a word, sentence or a loud tune. From the beloved Mpeg2Schnitt I am used to the following options "Previews" which actually serve more as Pre-"listen" than "view":

4.a Please provide one-click "Preview from Current Time" and one-click "Preview to Current Time" to see and hear ( ! ) in advance what preview "Test edit point" between two highlighted EDL-lines will bring.
(Now, in all my previews, audio in the Preview window always stops much earlier (approx. 60% - 75%) than video, and I never know what I will get!? )
Please keep existing "Test edit point" preview of connection between two highlighted EDLs as is.

5. Please make it easier, with fewer mouse clicks, to modify the end point of one EDL and/or the start point of the next. Again for getting an improved "Test edit point" audio result.
Maybe like this: when EDL line is highlighted and activated to current "Start-/End" then let the EDL-line inherit the modifications of the Start and End-points with "Set" and "Test" or a "Replace EDL" instead of "Add EDL" would be fine, too.
Then, in combination with 4.a, the next "Test edit point" will in most cases be already a success and the extra housekeeping of the EDLs would not be necessary.
Now I have to "Add" my guess for a better Start/End, put the resulting new EDL into correct position, "Del" (hopefully the correct) old EDL - and no erraneous "Clear", etc.
One or two of these improvements may be fairly easy to implement. #1 would be my #1 ;)

Regards,

Cela

[Ventolin]

Hi Cela,
Yes, with a little practicing, it is fairly comfortable! :)
Good to hear!
1. Please remember size and position of Preview window (floating TSPE internal viewer) between starts of TSPE.
Try this: Open Internal Viewer, adjust size and position, close internal viewer. Reopen internal viewer - it should remember the position and size.
2. Please put "Clear" far away from the other options.
Clear now asks for confirmation.
3. Please provide an "Undo" for "Clear" and "Del" (and other actions). The more EDL there are the higher is the risk to loose a lot of work with one erraneous "Clear".
Will include this in the EDL update in a future version.
4. Please provide an easier way to pre-test the future "Test edit point". I do not like to cut and glue in the middle of a word, sentence or a loud tune. From the beloved Mpeg2Schnitt I am used to the following options "Previews" which actually serve more as Pre-"listen" than "view":

4.a Please provide one-click "Preview from Current Time" and one-click "Preview to Current Time" to see and hear ( ! ) in advance what preview "Test edit point" between two highlighted EDL-lines will bring.
(Now, in all my previews, audio in the Preview window always stops much earlier (approx. 60% - 75%) than video, and I never know what I will get!? )
Please keep existing "Test edit point" preview of connection between two highlighted EDLs as is.
Will look at this with the GUI update planned for a future version.
5. Please make it easier, with fewer mouse clicks, to modify the end point of one EDL and/or the start point of the next. Again for getting an improved "Test edit point" audio result.
Maybe like this: when EDL line is highlighted and activated to current "Start-/End" then let the EDL-line inherit the modifications of the Start and End-points with "Set" and "Test" or a "Replace EDL" instead of "Add EDL" would be fine, too.
Then, in combination with 4.a, the next "Test edit point" will in most cases be already a success and the extra housekeeping of the EDLs would not be necessary.
Now I have to "Add" my guess for a better Start/End, put the resulting new EDL into correct position, "Del" (hopefully the correct) old EDL - and no erraneous "Clear", etc.
Will include this in the EDL update in a future version.
One or two of these improvements may be fairly easy to implement. #1 would be my #1 ;)
Hope #1 is already answered! I did have a look at the other suggestions, will take a bit more to do than I have time for to include with the next update, but certainly will look at including them in a future update.

Next release should be early Feb.

Regards,

Vent

[Cela]

...Try this: Open Internal Viewer, adjust size and position, close internal viewer. Reopen internal viewer - it should remember the position and size.
...Yes, it works!
Very tricky, though. Never got the idea to treat the test viewer window individually, always only in the context of "Test Start", "Test End". Never closed internal window by itself, never got the idea there would be a benefit of doing so. Always closed the main window and all was gone.

Please consider to remember the test viewer window position even when it gets closed implicitly by the main window.

But I have another request: please allow an option to have the test viewer window maintain the video's aspect ratio when being zoomed in or out.

Early Feb! Good to hear!

Will there be a rebate for a second license (i.e. notebook, weekend cottage, ...)?

Regards,

Cela

[Ventolin]

Hi Cela,
Yes, it works!
Very tricky, though. Never got the idea to treat the test viewer window individually, always only in the context of "Test Start", "Test End". Never closed internal window by itself, never got the idea there would be a benefit of doing so. Always closed the main window and all was gone.

Please consider to remember the test viewer window position even when it gets closed implicitly by the main window.
Already added!

But I have another request: please allow an option to have the test viewer window maintain the video's aspect ratio when being zoomed in or out.
You can set the aspect ratio in the directshow codec itself (ffdshow and coreavc amoungst others support this), but since a few people are asking for the option, I'll see about adding it to a future version. Don't think it will make the next update though.

Will there be a rebate for a second license (i.e. notebook, weekend cottage, ...)?
Yes, there may be for those that are especially helpful and give lots of feedback and do lots of testing etc, but only for a limited time and on a case by case basis. :)

Cheers,

Vent

[Borgqy]

Hi ,
I'm having some trouble editing commercials in my sky+ hd caps.
Sometimes the edited caps seem slightly desynced to me and when i scan them i receive those errors :

Scanning File...
PCR gap of -0,5 seconds in PID 0x0200 at offset 0x707BA7D4
PCR gap of -0,5 seconds in PID 0x0200 at offset 0xA06AC430

Those Errors happen exactly on the commercial cuts. Some of the caps that seem desynched to me have them on all cuts , some only on a few.

[canTsTop]

here is sample http://www.mediafire.com/?tyvm5km2nom
after i load it in TSPE and press F+ button it says: No More Frames Found.

[Ventolin]

Hi ,
I'm having some trouble editing commercials in my sky+ hd caps.
Sometimes the edited caps seem slightly desynced to me and when i scan them i receive those errors :

Scanning File...
PCR gap of -0,5 seconds in PID 0x0200 at offset 0x707BA7D4
PCR gap of -0,5 seconds in PID 0x0200 at offset 0xA06AC430

Those Errors happen exactly on the commercial cuts. Some of the caps that seem desynched to me have them on all cuts , some only on a few.
Hi,

Thanks for your report!

Can you please check:
1. Fix Timecodes and Fix Tables are checked.
2. Menu Edit -> Clean Edit Point -> Clean Video and Clean Audio are checked.

If that is the case, then please send me a sample, from -30 seconds before the ads to +30 seconds after the ads with all the ads intact. Turn off Fix Tables and Fix Timecodes when you make the sample.

Thanks,

Vent

[Ventolin]

here is sample http://www.mediafire.com/?tyvm5km2nom
after i load it in TSPE and press F+ button it says: No More Frames Found.

Hi,

Thanks for your excellent sample, I get the same behaviour here also. It seems the Primary Picture Type is not set in this stream which indicates the type of the frame. Will try and get a solution for this in the next release.

In the meantime, you can still navigate the stream by dragging the slider and using the PTS +/- buttons to move frames. If you look at the Decode Window, look for:

PID: 0x0212
Continuity Counter: 5
PTS: 00:00:42:21 (42) 42.840s (398400000) (Frame: 996)
DTS: 00:00:40:23 (47) 40.951s
Attributes: H.264 720x576 (786x576) @ 25.000 fps
Payload Unit Start Indicator: True
Transport Scrambling Control: 0: (no scrambling of TS packet payload)
Adaptation Field Control: 3
Adaptation Field Length: 1
Random Access Indicator: True
Stream Priority Indicator: True

When the Random Access Indicator is set, the frame is an I-Frame and a good point to cut at.

There is another issue with the audio after editing, to resolve this issue for this stream turn off Menu -> Edit -> Clean Edit Point -> Clean Audio Frames. This is because the stream has a PCR on both Video and Audio streams for the same programme. This is out of spec as "A program shall have one and only one PCR time base associated with it." (from ISO13818-1). I'll see what can be done to fix the stream in this case.

Thanks,

Regards,

Vent

[Borgqy]

Hi ,

heres the sample :
http://www.mediafire.com/?sharekey=ce4c8139a77c5e78d1014a7a667fa2b4e04e75f6e8ebb871

I kept all pids and disabled the fix tables and fix timecodes.
When i edit my caps i always remove the mpeg audio pid when there is one.

[xeniste]

Hi guys,

I'd like to buy a permanent license, but can you before explain some problems?

I record DVB emissions with my Topfield 7710 HD PVR, both in SD and in HD resolution. Then, with h.264 Cutter, I transform .rec files in .ts files.
Next step - cutting and trimming with your software. And then, I have never could do it properly - either the start point is before or after the indicated one, or the end point does the same thing.
I have tried all possibilities: on screen, everything goes smooth, but the final .ts file is every time improperly cut.

Where's the problem?

And every time, the preview window shows a black screen.
Besides, in your manual, you omit to indicate to press Add before trimming: if I don't do it, the process doesn't start.

Thank you in advance for your explanations.

[Ventolin]

Hi ,

heres the sample :
http://www.mediafire.com/?sharekey=ce4c8139a77c5e78d1014a7a667fa2b4e04e75f6e8ebb871

I kept all pids and disabled the fix tables and fix timecodes.
When i edit my caps i always remove the mpeg audio pid when there is one.
Hi Borgqy,

Thanks for the sample, hopefully will have it fixed in the next version which is due very soon. Already I get good sync at the edit points and hopefully will also fix the remaining errors.

Regards,

Vent

[Ventolin]

Hi guys,

I'd like to buy a permanent license, but can you before explain some problems?

I record DVB emissions with my Topfield 7710 HD PVR, both in SD and in HD resolution. Then, with h.264 Cutter, I transform .rec files in .ts files.
Next step - cutting and trimming with your software. And then, I have never could do it properly - either the start point is before or after the indicated one, or the end point does the same thing.
I have tried all possibilities: on screen, everything goes smooth, but the final .ts file is every time improperly cut.

Where's the problem?

And every time, the preview window shows a black screen.
Besides, in your manual, you omit to indicate to press Add before trimming: if I don't do it, the process doesn't start.

Thank you in advance for your explanations.
Hi xeniste,

Can you send me a sample of your .rec files and associated .ts files? I'll see if I can add support for those directly.

Regarding trimming, if there are no items in the EDL you can just set the start / end points and then hit Edit. This will automatically add the start / end points to the EDL then perform the edit.

As for preview:
Please make sure you have the latest Haali and FFDShow filters installed.

If you still have the same issues after installing / reinstalling / uninstalling-reinstalling Haali and FFDShow, then send me a sample large enough to exhibit the issue. When sending samples, please turn off Fix Tables and Fix Timecodes, thanks.

Regards,

Vent

[xeniste]

Hi Vent,

Thank you for your reply.
Sure I can send the - how much in K/M/sec, and where ?
I discovered your TSRepair program - no effect, same problems.

[Ventolin]

Hi Vent,

Thank you for your reply.
Sure I can send the - how much in K/M/sec, and where ?
I discovered your TSRepair program - no effect, same problems.
Hi,

I'm not sure how big you need to make your sample to exhibit the issue, perhaps 1 minute would be sufficient? Same goes for the .rec sample.

You can you post to a file sharing site such as zshare / mediafire / rapidshare.de / etc.

TSRepair is not my program, TSPE already has it's own repair options!

Regards,

Vent

[xeniste]

Here they are :
.rec beginning :https://download.yousendit.com/Y2ovTkF0dENwM2xFQlE9PQ

.rec end : https://download.yousendit.com/Y2ovTkFxbEpRR2MwTVE9PQ

.ts beginning: https://download.yousendit.com/Y2ovRE9uTmFRWUkwTVE9PQ

.ts end : https://download.yousendit.com/Y2ovRE90UnEzMW14dnc9PQ

The problem arises near the very beginning of the film and near its end. Probably something is added (or missing) in the flow. Because when I rewind or go fast forward the record on the receiver, it's exactly the same thing : the image freezes, then rewinds or goes forward after or before the desired point.

Hope it can help you.

Many thanx once again.

[CocoaCrispy]

Is there a way to use TSPE to join files sequentially, combining 2 or more files into one file?

Thanks

[CocoaCrispy]

I sent / rcvd a PM with Vent. I am posting the response here since it is for any who have the same question.
-----
(from Ventolin)

Hi, yes, you can merge 2 files as long as they have the same PIDs.

Just drag/drop the files into the EDL window. You can double click the edl entry to open the file or you can load each file individually and add to the EDL.

Click Edit once you are ready, (make sure no item in the EDL is selected).

The drag/drop EDL feature is really for multi-part archives and will be byte accurate for this purpose (saves you from having to use another program to combine the files)

----

My (new) response - Great. I understand now that each EDL entry not only has the start and stop points for the cuts but the input filename too. So I can combine cuts from different files into one output file, all in one operation.

A snag that I've found is that TSPE is not working properly using .m2ts input files that I made with tsremux.

If I take the same file and remux it into .ts format (using tsremux again) then TSPE likes it.

When the same file is in m2ts format, and I press the edit button to generate the output file, I get an error message in the log at the bottom left that states, "Error: Index was outside the bounds of the array. PID o, ipii: 0, ProgIndex: 0 Warning: No identified PID found in this section, Total time taken 0.10 seconds, write speed 0.0 MB/s"

It looks like TSPE has a problem accepting the m2ts files that come out of tsremux during the output compilation phase. I have hundreds of these and would prefer to not have to remux them into .ts format. Can TSPE be improved for this?

Aside from this, so far TSPE looks great. It's the only software I've found that will allow cutting commercials from h264 HD content. The ability to reassemble h264 HD segments together without reencoding is another unique feature. The results are not as nice as what Videoredo does with mpeg2, since there is no smooth transition at the cuts though.

Now after that nice pat on the back, please look into my requested improvement.

CC

[t12345]

Ventolin,
My computer AMD64@3000+ is not enough powerful to play 1080i source without problems. I have bought ati 4650 vga card for use hardware acceleration , but I do not see any big progress. I can play the material in sense quality in MediaPlayerClassic but stream editors can not use this support I afraid. (TSPE needs 100% CPU power but playing file is much beter than in VLC)

My source of HDTV stream is dreambox8000. I can not use my PC for comfortable player , but I can use dm8000 for it. I can jumping every 3 minutes and check if is something interesting to cut (like good clip on MTVHD). If something interesting is I can write time from the start of file for find this time on PC.

TSPE looks brilliant for its extra quick working. The one thing what is not possible to find in it is key managing. I can use mouse and buttons to change position in opened file. OK. But could you add keyboard shortcuts to it ? Arrows left/right for -F/+F, ctrl+arrow for IF, alt+arrow for second, shift+arrow for 10s, ctrl+shift+arrow for 1minute scroll, and maybe alt+ctrl+arrow for 10 minutes scroll . (or other key layout). It will very help to manipulate the program using the keyboard.
The 1minute and 10minutes scroll do not exist now, but will be very helpful if you will add it.

My slow computer needs some time when I am pressing button like +IF for see result of action. Maybe TPSE can switch on a red point/flash in the window for show the user about all time working. I am pressing second time the button because I am not sure if TPSE has finished previous pressing button or not.
If I will have intel quad CPU will I be able to work online when I will press navigation buttons and will I see moved current frame in preview at once ?

I do not know which way it is doing. The 26G file TPSE is opening in 30-40 seconds checking it for errors (if I have corrected understood) in each packet. The TPSE is running in my PC. The source stream is stored in dlink323 , where is my storage center used by dreambox to store files. I have 1Gbit connection between dlink and PC. Max link utilization on dlink is 150Mbit/s . I can not imagine, how TPSE is able to analyze all file during this short time. But I love this speed. AVI demux needs hours for it.

[Ventolin]

Here they are :
.rec beginning :https://download.yousendit.com/Y2ovTkF0dENwM2xFQlE9PQ

.rec end : https://download.yousendit.com/Y2ovTkFxbEpRR2MwTVE9PQ

.ts beginning: https://download.yousendit.com/Y2ovRE9uTmFRWUkwTVE9PQ

.ts end : https://download.yousendit.com/Y2ovRE90UnEzMW14dnc9PQ

The problem arises near the very beginning of the film and near its end. Probably something is added (or missing) in the flow. Because when I rewind or go fast forward the record on the receiver, it's exactly the same thing : the image freezes, then rewinds or goes forward after or before the desired point.

Hope it can help you.

Many thanx once again.

Hi,

Thanks for the samples, looking in to it.

Regards,

Vent

[Ventolin]

A snag that I've found is that TSPE is not working properly using .m2ts input files that I made with tsremux.

If I take the same file and remux it into .ts format (using tsremux again) then TSPE likes it.

When the same file is in m2ts format, and I press the edit button to generate the output file, I get an error message in the log at the bottom left that states, "Error: Index was outside the bounds of the array. PID o, ipii: 0, ProgIndex: 0 Warning: No identified PID found in this section, Total time taken 0.10 seconds, write speed 0.0 MB/s"

It looks like TSPE has a problem accepting the m2ts files that come out of tsremux during the output compilation phase. I have hundreds of these and would prefer to not have to remux them into .ts format. Can TSPE be improved for this?
Hi,

If you send me some sample files and a description on how to reproduce the error I'll look into it.

TSPE has the option of stripping the m2ts headers on edit.

Aside from this, so far TSPE looks great. It's the only software I've found that will allow cutting commercials from h264 HD content. The ability to reassemble h264 HD segments together without reencoding is another unique feature. The results are not as nice as what Videoredo does with mpeg2, since there is no smooth transition at the cuts though.

Now after that nice pat on the back, please look into my requested improvement.

CC
The transition point should be smooth, again, please provide a sample with some description on how to reproduce the behaviour and I'll look into it.

Thanks for the pat on the back I'll need some more once I fix your issues though!

Regards,

Vent

[Ventolin]

Ventolin,
My computer AMD64@3000+ is not enough powerful to play 1080i source without problems. I have bought ati 4650 vga card for use hardware acceleration , but I do not see any big progress. I can play the material in sense quality in MediaPlayerClassic but stream editors can not use this support I afraid. (TSPE needs 100% CPU power but playing file is much beter than in VLC)
I have an Athlon XP 3200+ based machine with an ATI 2600 AGP card inside. If I play through PowerDVD I get very good results with BluRay discs. TSPE lets you choose which directshow filter to use for the preview, you may find a filter that is able to take advantage of the ATI 4650 hardware acceleration.

My source of HDTV stream is dreambox8000. I can not use my PC for comfortable player , but I can use dm8000 for it. I can jumping every 3 minutes and check if is something interesting to cut (like good clip on MTVHD). If something interesting is I can write time from the start of file for find this time on PC.
You can jump directly to a time by typing in the time in the Current Time box above the Video Preview.

TSPE looks brilliant for its extra quick working. The one thing what is not possible to find in it is key managing. I can use mouse and buttons to change position in opened file. OK. But could you add keyboard shortcuts to it ? Arrows left/right for -F/+F, ctrl+arrow for IF, alt+arrow for second, shift+arrow for 10s, ctrl+shift+arrow for 1minute scroll, and maybe alt+ctrl+arrow for 10 minutes scroll . (or other key layout). It will very help to manipulate the program using the keyboard.
The 1minute and 10minutes scroll do not exist now, but will be very helpful if you will add it.
Menu -> Help -> Keyboard Shortcuts. Enjoy!

My slow computer needs some time when I am pressing button like +IF for see result of action. Maybe TPSE can switch on a red point/flash in the window for show the user about all time working. I am pressing second time the button because I am not sure if TPSE has finished previous pressing button or not.
If I will have intel quad CPU will I be able to work online when I will press navigation buttons and will I see moved current frame in preview at once ?
A Quad Core is indeed very good for instant response, as is a Dual Core. On my Athlon XP 3200+ I get reasonable response times, so perhaps it is your DirectShow filter setup that is causing slow responses?

I do not know which way it is doing. The 26G file TPSE is opening in 30-40 seconds checking it for errors (if I have corrected understood) in each packet. The TPSE is running in my PC. The source stream is stored in dlink323 , where is my storage center used by dreambox to store files. I have 1Gbit connection between dlink and PC. Max link utilization on dlink is 150Mbit/s . I can not imagine, how TPSE is able to analyze all file during this short time. But I love this speed. AVI demux needs hours for it.
When you open a file, TSPE quickly scans the file to detect the PIDs in the stream and to find Commercial Points by looking for attribute changes, eg. going from 5.1 AC3 to 2.0 AC3.
To perform a full scan, press the Scan button, I expect the scan time to take a good few minutes for a 26GB file across a network!

Regards,

Vent

[Borgqy]

Hi Borgqy,

Thanks for the sample, hopefully will have it fixed in the next version which is due very soon. Already I get good sync at the edit points and hopefully will also fix the remaining errors.

Regards,

Vent

Hi ,
Sounds great already looking forward to it. If u need more Samples i can always upload more.

[Cela]

Hi Ventolin,

after some testing with my 2 licenced TSPE copies I have found a possible issue (which may be already solved in your coming soon new version) and two additional requests:

Requests:
1. I need a nice TSPE Icon for my desktop link to TSPE.
2. I need a TSPE-like version of your programme for WMV (VC1) streams which I record with Total Recorder Video Add-On v7.1. And/or can you tell me how to convert ".wmv" <-> ".TS(WMV)"

Issue:
3. I recently captured, as usual with DVBViewer, a HD broadcast of a sports event: Total Size 15179,5 MB (15916829744 Bytes). But unlike other captures, this one could not be processed with TSPE. DVBViewer's TSPlayer_v2.0.0 and H264_TS_Cutter_v1.1.1 can process the TS. I uploaded a small part which I cut out of the original TS with TSPlayer to Requests&Issue (http://rapidshare.com/files/196465898/TSPE_Icon_and_WMSPE_needed.rar). Then I cut out a small part from this section with H264_TS_Cutter_v1.1.1. TSPE refused to process any of these sections. I guess it did not find the video PID 0x0780 (1920)?

This is one of the TSPE logs:
Version: 0.740
Buffer Size: 1540096
Command Line Parameter Found, Opening: F:\HD Val d_Isere 090207 WM Downhill Clip_cut.ts
Opening File...
Auto Finding PIDs
PID Attribute List:
PID: 0x0782: MPEG1 Audio Layer 3, 56Kbit/s @ 44100Hz
PID: 0x0783: AC3: 2/0: L,R 48KHz 448Kbit/s

Scanning For Format Changes
File size: 18191820 Bytes
Duration from Timestamps: 11.26
Average System Bitrate: 12920313.96
Done.

and this is the original capturing report from the DVBViewer:

[General]
0=PAT - Program Association Table
[61920]
Name=T.TV
920=PMT
1920=Video
1921=Audio
1922=Audio
1923=AC3
1809=PMT
61=Video
62=Audio


Regards,

Cela

[Ventolin]

Hi Ventolin,

after some testing with my 2 licenced TSPE copies I have found a possible issue (which may be already solved in your coming soon new version) and two additional requests:

Requests:
1. I need a nice TSPE Icon for my desktop link to TSPE.
2. I need a TSPE-like version of your programme for WMV (VC1) streams which I record with Total Recorder Video Add-On v7.1. And/or can you tell me how to convert ".wmv" <-> ".TS(WMV)"

Issue:
3. I recently captured, as usual with DVBViewer, a HD broadcast of a sports event: Total Size 15179,5 MB (15916829744 Bytes). But unlike other captures, this one could not be processed with TSPE. DVBViewer's TSPlayer_v2.0.0 and H264_TS_Cutter_v1.1.1 can process the TS. I uploaded a small part which I cut out of the original TS with TSPlayer to Requests&Issue (http://rapidshare.com/files/196465898/TSPE_Icon_and_WMSPE_needed.rar). Then I cut out a small part from this section with H264_TS_Cutter_v1.1.1. TSPE refused to process any of these sections. I guess it did not find the video PID 0x0780 (1920)?

Regards,

Cela
Hi,

Thanks for the samples, yes they both open fine in my dev version here and edit successfully, so whatever it was is fixed!

You'll have to figure out how to create BD compliant VC-1 streams instead of WMV and then use a muxer to create a VC-1 TS file before TSPE will be able to edit it.

Haven't had much time to spend on the icon, will have some time after the next version is released I hope. Feel free to offer some suggestions!

Regards,

Vent

[t12345]

Ventolin
thanks for your feedback and suggestions.
I have found possibility to change the codec in TSPE configuration. :) I have used multimedia without understanding so far. I have installed a packet of codecs and all have been worked good and quick ... maybe because SD mpg2 and avi need much smaller power than my computer has. Now, HDTV needs rapidly much more power.
TSPE has used fddshow as standard codec. I have changed it to cyberlink's one , and playing video in TSPE needs only 12% of CPU power. The same in PowerDVD needs 45%. Maybe you are able to use it better then authors of this codec. Congratulation for good job.

I have used videoredo so far for my SDTV captures in ts files. I have used it because of its extra key management and very quick working. I can press a key for frame/second/10 seconds/minute scrolling and the video is scrolled smooth , continuous like normal playing but in bigger speed. It makes checking what is stored very easy (and FAST!!!). It is working enough good that the very bad working of videoredo with corrupted (not descrambled) source is not enough to use something else. But videoredo is only for mpg2.
I hope, TPSE will be in future in the same ergonomic level for hdtv like now videoredo is for SDTV. Today with accelereated codec TSPE is not able to play the source smooth when I pressing scrolling by frame/iframe/second/10 seconds (or is it possible but I do not have good configuration on my PC ?). It can be result of not enough good TSPE, used codec or slow PC. I hope it will be improved in future , only improved PC can done by me , but it will not help I afraid.

How am I using the videoredo now?
1. normal playing in normal speed
2. scrolling by frame using arrows (the same in TSPE)
3. scrolling by 1 second using ctrl + arrow (page up/down)
4. scrolling by 10 seconds using shift + arrow (ctrl + page up/down)
5. scrolling by 1 minute using shift+ctrl+arrow (pressing mouse on scroll bar)
The scrolling by 1 minute is enough smooth/quick and scrolling with bigger speed is not needed.

I understand, own settings of keys in TSPE will make TSPE will be unique. But maybe better will be using more ergonomic key management.
The scrolling of material is doing in horizontal direction : previous in the left and next in the right. The using of page up and page down in vertical direction is not natural.
Maybe as long as smooth scrolling in TSPE will not be possible it does not have big importance but why not project it good at once?
Why this what I am using current is fine ? I am using both hands or only one !!! One finger on arrow second on ctrl , shift or on shift and ctrl in the same time. If I want to slower the scrolling I am only releasing one finger or changing pressing the key and I do not have to look on it because the key is next to used current and no possibility to press wrong key.
Do you think, will you be able to change key mapping to suggested by me?

Scrolling over iframes . I thing it is not helpful during checking material for finding something interesting. It is helpful when I will find what I want to edit. When it will occure, I will pay attention for correct selecting part for edit and only in this moment I will need scrolling by iframe. I am trying to get to like you to use arrow and control for 1 second scrolling , not for iframe. The combination for alt+arrow can be good for iframe or other.

The pressing on graphic scroll bar is moving current position + or - 1 minute. My suggestion is to add button for scroll 1 minute. If user will click on scroll bar the current position should be changed to apropriate position on scroll bar. Of course if the calculation of position will be not problem. I hope will not be, because dragging scroll pointer is fast.

I do not want to suggest, the current set of key management is wrong. No. It is good but only for use your application for short time. If someone will want open 20 hours recording for checking what is inside to edit, the absence of ergonomic key management will be a nightmare. I hope you will understand my intention, I do not want to criticize you but only help making your application better.

[Ventolin]

Hi t12345,

...Now, HDTV needs rapidly much more power.
TSPE has used fddshow as standard codec. I have changed it to cyberlink's one , and playing video in TSPE needs only 12% of CPU power. The same in PowerDVD needs 45%. Maybe you are able to use it better then authors of this codec. Congratulation for good job.
Thanks for the kind words regarding performance here, it's interesting that TSPE is faster than PowerDVD using Cyberlink's own Video Decoder, but then the splitter in use may be different, so the lower CPU load may be due to the splitter. You are right, HD needs alot more power, even with a Quad Core and 4GB ram it's not enough for fast scrubbing!
I have used videoredo so far for my SDTV captures in ts files. I have used it because of its extra key management and very quick working. I can press a key for frame/second/10 seconds/minute scrolling and the video is scrolled smooth , continuous like normal playing but in bigger speed. It makes checking what is stored very easy (and FAST!!!). It is working enough good that the very bad working of videoredo with corrupted (not descrambled) source is not enough to use something else. But videoredo is only for mpg2.
I hope, TPSE will be in future in the same ergonomic level for hdtv like now videoredo is for SDTV. Today with accelereated codec TSPE is not able to play the source smooth when I pressing scrolling by frame/iframe/second/10 seconds (or is it possible but I do not have good configuration on my PC ?). It can be result of not enough good TSPE, used codec or slow PC. I hope it will be improved in future , only improved PC can done by me , but it will not help I afraid.
Most of the CPU time is taken up by the direct show filters. If you want to see the raw speed of TSPE alone, turn off the video preview and scrub the video with the scroll bar. You will see the data in the decode windows appear instantly in most cases. TSPE has not even been optimised for this yet!

How am I using the videoredo now?
1. normal playing in normal speed
2. scrolling by frame using arrows (the same in TSPE)
3. scrolling by 1 second using ctrl + arrow (page up/down)
4. scrolling by 10 seconds using shift + arrow (ctrl + page up/down)
5. scrolling by 1 minute using shift+ctrl+arrow (pressing mouse on scroll bar)
The scrolling by 1 minute is enough smooth/quick and scrolling with bigger speed is not needed.

...

I do not want to suggest, the current set of key management is wrong. No. It is good but only for use your application for short time. If someone will want open 20 hours recording for checking what is inside to edit, the absence of ergonomic key management will be a nightmare. I hope you will understand my intention, I do not want to criticize you but only help making your application better.
Thank you for your feed back regarding ergonomic keyboard controls. They made me think about it enough for me to work on the keyboard controls ahead of schedule!

I have implemented two ways of 60 second jumps with keyboard combinations, and have improved the default key settings also. I am unable to make the key settings the same as VideoRedo, but I do hope you will find the keyboard settings alot more to your liking.

Although you are obviously used to using the cursor keys for navigation, I would suggest trying the main keyboard shortcuts instead (with improvements in the next version). These are:

J, K, L, I, O, M.

Have a look at where these keys are on the keyboard. You'll find them right under your fingertips if you are used to touch typing (on a UK/US keyboard). Not only can you navigate fully (with Shift and Ctrl modifiers), but you can set and test start and end points and set markers all with one hand!
Furthermore, if you are used to using other professional editors, you'll already be familiar with the keyboard combinations. Try them out and see how you get on.

In a future version, the keyboard shortcuts will be user definable, but for now, the updated keyboard shortcuts will be included the next released version, which I really hope will be out very very soon!

Best Regards,

Vent

[jeffnoone]

Was going to buy a license for this, but eventually on purchase page, I got to the need to insert Activation Key Strings 1&2

Can I get some more detail on conditions of use for a licence?

I use a couple of computers to video-edit at home, but never two computers at one time. These are also the computers I upgrade, i.e. if I build new computer, the new one is used for video-editing, the older one goes to one of the kids

So, some questions

Can this be licensed on several computers at the same time (same owner, no simultaneous usage)?

Can I transfer license to another computer WITHOUT input from the author?

On same lines, if auther goes out of business, can I move licence?

Can someone let me know before purchase

Thanks
Jeff

[Ventolin]

Was going to buy a license for this, but eventually on purchase page, I got to the need to insert Activation Key Strings 1&2

Can I get some more detail on conditions of use for a licence?

I use a couple of computers to video-edit at home, but never two computers at one time. These are also the computers I upgrade, i.e. if I build new computer, the new one is used for video-editing, the older one goes to one of the kids

So, some questions

Can this be licensed on several computers at the same time (same owner, no simultaneous usage)?

Can I transfer license to another computer WITHOUT input from the author?

On same lines, if auther goes out of business, can I move licence?

Can someone let me know before purchase

Thanks
Jeff
Hi,

Thanks for your interest!

The licence applies to 1 PC only and is hardware locked, similar to Windows OEM OS licences.

As TSPE is hardware locked, you can not move licence yourself from one PC to another and if you change your hardware significantly you will have to get a replacement licence.

At the moment, we are issuing Free replacement licences if you upgrade or change your PC. Just send the new activation details with an explanation and a new licence will be sent to you (as long as the requests are reasonable ie, not requesting a new one every week!)

If you want to use 2 PCs simultaneously, you will need 2 licences.

Major version upgrades will be chargable, if you buy now however, your licence will last until Version 2.0 is available. For the price I think that's a very good deal!

It'll be "Acts of God" that will take the authors "out of business" and even then I'm sure TSPE will still live on!

I hope that answers your questions, if you do have any further queries, please feel free to give me a PM.

Regards,

Vent

[jeffnoone]

Seems fair - just purchased
Keep up the good work
Jeff

[rebkell]

I'm having problems, I've edited dozens of files with TSPE with no problems, but now all of a sudden it won't let me create an edited file, even ones it has done previously. I can open the file and create all the edit points, but when I click on edit or create a test edit, I get a message like this:

Testing Edit Point
Copying Data...
Creating Test Edit
New PAT Generated.
New PMT Generated.

Error:
Index was outside the bounds of the array.
PID: 0, iPii: 0, ProgIndex: 0
Launching Internal Viewer

Error: Auto Creating Filter Graph
Error HRESULT E_FAIL has been returned from a call to a COM component.
Error Loading Video, Disabling.


I used H264ts_cutter to edit the file and it worked flawlessly, I had a previously edited file, and loaded up the edl file and got the same errors. I can play all these files, I can edit them in TSPE, but when it comes time to create edit points, I get the above error, every time: Help.

[Ventolin]

Hi rebkell,

I have sent you a PM regarding your issue.

Cheers,

Vent

Edit: Now Sorted :)

[guppyman]

I'm having problems, I've edited dozens of files with TSPE with no problems, but now all of a sudden it won't let me create an edited file, even ones it has done previously. I can open the file and create all the edit points, but when I click on edit or create a test edit, I get a message like this:

Testing Edit Point
Copying Data...
Creating Test Edit
New PAT Generated.
New PMT Generated.

Error:
Index was outside the bounds of the array.
PID: 0, iPii: 0, ProgIndex: 0
Launching Internal Viewer

Error: Auto Creating Filter Graph
Error HRESULT E_FAIL has been returned from a call to a COM component.
Error Loading Video, Disabling.


I used H264ts_cutter to edit the file and it worked flawlessly, I had a previously edited file, and loaded up the edl file and got the same errors. I can play all these files, I can edit them in TSPE, but when it comes time to create edit points, I get the above error, every time: Help.
The same thing is happening to me - TSPE has been working great and now it won't. Here's the latest error message:
Error:
Index was outside the bounds of the array.
PID: 0, iPii: 0, ProgIndex: 0
Warning: No identified video PID found in this section.
Total time taken 0.11 seconds, write speed 0.0 MB/s

Edit complete: successfully written file with 376 bytes.

Error: Auto Creating Filter Graph
Exception from HRESULT: 0x80040265
Error Loading Video, Disabling.
Internal Viewer Closed.

[Jackda]

And the same here, too.

Same error, same message, different language ;) .

TSPE has been working great here, too. Did many, many files with the trial,
and a couple ones since I purchased it.

[Ventolin]

Hi folks,

Yeah, there is an issue with the licence system which shows up in Vista which gives this error.

If you have a full licence and get the error, just send in your activation details again and I'll regenerate a new licence for you.

This shouldn't affect temp licences, if you do get this error with a temp licence, which hasn't expired, then let me know and I'll issue an extension for you.

Apologies for the inconvenience caused, hopefully this issue is fixed in the next release!

Next release has "Table Editor" which lets you re-order, auto-sort and auto-select streams in the PID tree, handles streams with broken or missing PAT/PMTs (including Topfield, and other PVRs) and a good few other minor additions.

Haven't released that yet though as there is a bug in the output that needs to be ironed out, which is proving to be hard to fix.

Regards,

Vent

[Lovok]

I'm testing TSPE 0.740 on Vista x64, 25 days of trial left, but suddenly 'Error: Index was outside the bounds of the array.' instead of Editing :(

[Renzz]

Same failure here on the trial licence on Vista 32.

[satnews]

Hi folks,

This shouldn't affect temp licences...

same on my side too, but 1 day left of 45 only

windows xp/x64 sp2

thank

[Ventolin]

Hi folks,

Thanks for all the reports and sorry for the inconvenience.

Just to let you know I'm working on it, will give an update soon.

Does anyone with a full licence have trouble?

Regards,

Vent

[Jackda]

Yes, as posted above, I do.
I have sent an email with my activation details
to your activation email-adress this morning.
Should I have used the support one?

Don't get me wrong, I don't want to complain about anything,
I'm totally aware of TSPE being beta and that things like this can happen.

I just don't want "the problem" to be a wrong way of contacting you.

[Ventolin]

Yes, as posted above, I do.
I have sent an email with my activation details
to your activation email-adress this morning.
Should I have used the support one?

Don't get me wrong, I don't want to complain about anything,
I'm totally aware of TSPE being beta and that things like this can happen.

I just don't want "the problem" to be a wrong way of contacting you.

Hi Jackda,

I did get your activation details, but am trying hard to fix this major bug. There is no "wrong" way of contacting me, just don't turn up at my house!

I'm hoping that the new version of TSPE I'm working on will resolve the issues, for both full and temp users. Hopefully will have this ready in the next 24 hours or less.

Regards,

Vent

[Jackda]

Wow, I didn't expect the new version to be (at least hopefully) ready so soon.

Keep up the great work.

[Ventolin]

Hi folks,

New version 0.755 now up at website!

http://www.bitstreamtools.com/download

Hopefully this will fix the issues you have had with edits.

If you have a full licence you might have to regenerate your activation details and email them in again.

New 30 day temp licence, everyone's temp licences get reset to 30 days!

Quite a few new features added to 0.755, the most notable is the Table Editor.

This auto-sorts and auto-selects video and audio PIDs, can create missing PAT / PMT files for Topfield PVRs and others, supports ITVHD, AC3 stream now prefered instead of the mp2 audio description stream for BBCHD, fixes alot of broken tables from capture software outputs, you can drag/drop PIDs in the tree to re-order them and a whole host of other features too.

One issue is still remaining however, audio resync offset is not working, this will be fixed in the next update.

Please let me know how you get on with this release. I haven't had time to test this a much as I would like as the licence issue has pushed this release forward.

Apologies again for the inconvenience caused and thank you for your understanding!

Looking forward to your feedback,

Regards,

Vent

[Jackda]

Great news,
no error here with the new trial license,
no difference though with my old full-license.
Still get the same error.

I will generate the activation details again with the new version,
and send them over to you.

Because of the working trial you have 30 days to send me a new license ;)
I think it's great that you renewed the temp licenses by the way.

[dhnj]

My full license is working with 0.755 & Windows XP Pro SP3.
Just noticed that the output size of same file is little less than
in 0.740. Would like more documentation regarding new features in 0.755

[dhnj]

More testing of 0.755 has reveal the following:
I can author a AVCHD/BDMV DVD disc with Arcsoft Total Media Extreme without re-encoding with a Hauppauge HD PVR capture file edited with TSPE 0.740. If I edit the same file using EDL with TSPE 0.755 then Arcsoft Total Media Extreme will re-encode the input file doubling its file size.

[Ventolin]

More testing of 0.755 has reveal the following:
I can author a AVCHD/BDMV DVD disc with Arcsoft Total Media Extreme without re-encoding with a Hauppauge HD PVR capture file edited with TSPE 0.740. If I edit the same file using EDL with TSPE 0.755 then Arcsoft Total Media Extreme will re-encode the input file doubling its file size.

Hi dhnj,

Try making a new EDL without using the one from 0.740. You can also try disabling the Tables when you make the edit.

It is very strange that TME is re-encoding though but the only real difference is the table processing in 0.755 vs 0.740.

I will add some more documentation for sure on the Table Editor a bit later.

Let me know how you get on!

Regards,

Vent

[dhnj]

Hi dhnj,

Try making a new EDL without using the one from 0.740. You can also try disabling the Tables when you make the edit.

It is very strange that TME is re-encoding though but the only real difference is the table processing in 0.755 vs 0.740.

I will add some more documentation for sure on the Table Editor a bit later.

Let me know how you get on!

Regards,

Vent
Hi Vent

TME has always been very fussy with the edit points used in the input files. I have edit using the same edit points both in 0.740 & 0.755 & the results are as follows:
0.740 edited file size 128360 KB TME no re-encode.
0.755 edited file size 134475 KB TME will re-encode.

Regards

[Borgqy]

Hi ,
the new version looks very nice , especially the table editor is great.
I still get these errors on edits at the cutpoint:
PCR gap of 0,5 seconds in PID 0x0200 at offset 0x9F7CE224
However the audio seems to be in sync now as far as i could test it atm.
What are those Errors exactly , and what consequences could they bring if i keep the edits with the errors? (like problems with playback on devices and stuff)

[Jackda]

Just did some testing yesterday, and I had no problems at all.
The new version, in combination with a new license file is working flawlessly.

And the table editor solved an otherwise time consuming problem.

Great work, Vent.

[joerg]

I am seeing problems editing (cutting) M2TS files I captured with my iCord HD. If I try to cut such a file, I get the following error:

----------
Version: 0.755
Buffer Size: 1540096

Opening: D:\Cap\iCord\***.ts
Opening File...
Video Loaded.
Auto Finding PIDs

Warning: PID 0x0010 Not Found in Scan

Warning: PID 0x189C Not Found in Scan

Warning: PID 0x1900 Not Found in Scan
PID Attribute List:
PID: 0x1842: H.264 1280x720 (1280x720) @ 50.000 fps
PID: 0x184D: MPEG1 Audio Layer 2, 64Kbit/s @ 44100Hz
PID: 0x184E: MPEG1 Audio Layer 2, 64Kbit/s @ 44100Hz

Scanning For Format Changes
File size: 8164254336 Bytes
Video Resolution from Codec: 1280 x 720 @ 50.00 fps
Duration from Codec: 5688.64
Duration from Timestamps: 5688.70
Average System Bitrate: 11481363.88
Done.
Copying Data...
Processing Selected EDL Item: 1 of 1
New PAT Created.
New PMT Created.

Error:
Index was outside the bounds of the array.
PID: 0, iPii: 0, ProgIndex: 0
Error: Edit failed, only 384 bytes written.
----------
In the preview area the picture displays properly.

Also if I try to scan the input stream I get lots of messages like the following:
M2TS Arrival Time Stamp Gap of 536868367 found at offset 0x180
M2TS Arrival Time Stamp Gap of 268431534 found at offset 0x300
M2TS Arrival Time Stamp Gap of 536866989 found at offset 0x480

I know the stream is basically fine, since I could cut it successfully with TS Doctor.

If I try to remux the stream from m2ts to ts using TsRemux prior to editing, these problems do not occur. But obviously I would like to omit that additional step. btw: When the remuxed stream is loaded, TSPE sets the video mode to MPEG instead of h264 by default for some reason, and I have to change that manually. Also the Audio streams are now detected as MPEG2 while they are really MPEG1 Layer2.

Any hints what might be wrong would be much appreciated!

P.S.: Is it possible what kind of frame (I, P, B) one is currently at in the Video Window? I understand that for cut-in I should only cut at I-frames. Can I cut at P-frames, too, for cut-out or are there any reasons why I shouldn't do so?

[Ventolin]

Hi folks,

Sorry for the delay in replying I have been quite busy!

joerg - I can confirm that there is a bug in 0.755 with Temp licences and .m2ts files. Normal .ts files should be fine however, as should Full Licences. However, if you do have a full licence and find .m2ts editing not working, please let me know!

I did want to post with a fix by now, but haven't been able to yet; I will of course give an update as soon as it's ready.

In the meantime, please do continue to remux - TSPE's built in M2TS stripping function is also affected by the temp licence M2TS bug unfortunately.

As for the detection being incorrect, I dont know what TSMuxer produces but TSPE reads the stream type information from the PAT/PMT (tables). If they are incorrect or missing then TSPE tries to "guess" the streams type (more work needs to be done here obviously!). The table editor can indeed be used to correct this. If you can send a sample of the remux (turn off Fix Tables and Fix Timecodes before you press the edit button) I can see what the cause is and perhaps find a better solution. If you do send a sample, please post to somewhere like Mediafire / rapidshare / zshare, etc.

When you do a scan of M2TS files, you can get alot of M2TS header errors. This feature is still in development and definitely needs more work! In the meantime, there's a new option in 0.755 to turn off M2TS header checking under Menu Scan -> Disable M2TS Header checking.

Most of the time, the Frame Type is shown in the bottom right hand corner as you navigate, so you can see what type of frame you have in the video window. You can also use the Decode Window to see the frame type. There are few instances where this does not happen for instance when a MPEG2 stream does not have a GOP header. Support for this situation is very limited at the moment!

Borgqy - The PCR error you mention is a minor non-critical bug. It was picked up in 0.740 and is carried forward to 0.755 unfortunately. The bug is still being worked on as a high priority though and hopefully will be resolved soon. The bug should not affect the playback of the edit in any way. Hopefully the next version will be able to eliminate the bug and also possibly fix the error in previous edits.

dhnj - Very puzzled about TME being so picky with re-encoding. I will try to figure out what TME looks for when deciding when to re-encode or not, but it would be great if you can ask the authors of TME also if you haven't already. I hope you don't mind if I fix the "big" bugs first though! I will also work on the documentation as time permits.

Finally, thank you to all that have supported TSPE! It really means alot!

Regards,

Vent

[joerg]

joerg - I can confirm that there is a bug in 0.755 with Temp licences and .m2ts files. Normal .ts files should be fine however, as should Full Licences. However, if you do have a full licence and find .m2ts editing not working, please let me know!

Thank you for your kind reply. As I have just started evaluating tools to handle my iCord HD TV captures, I am in fact using the trial license. If your program would just cost 10 EUR or so I'd probably have bought it immediately, but with the price you are asking I definitely want to try this out thoroughly before deciding to buy it.


As for the detection being incorrect, I dont know what TSMuxer produces but TSPE reads the stream type information from the PAT/PMT (tables). If they are incorrect or missing then TSPE tries to "guess" the streams type (more work needs to be done here obviously!). The table editor can indeed be used to correct this. If you can send a sample of the remux (turn off Fix Tables and Fix Timecodes before you press the edit button) I can see what the cause is and perhaps find a better solution. If you do send a sample, please post to somewhere like Mediafire / rapidshare / zshare, etc.

I am using TsRemux, not TSMuxer. I'll send you a PM with a link to the sample. With the sample both the Audio and Video stream type are set to MPEG2.


When you do a scan of M2TS files, you can get alot of M2TS header errors. This feature is still in development and definitely needs more work! In the meantime, there's a new option in 0.755 to turn off M2TS header checking under Menu Scan -> Disable M2TS Header checking.

With header checking disabled the m2ts stream scans without errors.


Most of the time, the Frame Type is shown in the bottom right hand corner as you navigate, so you can see what type of frame you have in the video window. You can also use the Decode Window to see the frame type. There are few instances where this does not happen for instance when a MPEG2 stream does not have a GOP header. Support for this situation is very limited at the moment!

Oh, I overlooked that in the corner - thanks for the hint! It displays fine. :D
I would be happy if you could answer this question from my last posting: "I understand that for cut-in I should only cut at I-frames. Can I cut at P-frames, too, for cut-out or are there any reasons why I shouldn't do so?"

[Ventolin]

Hi Joerg,
Thank you for your kind reply. As I have just started evaluating tools to handle my iCord HD TV captures, I am in fact using the trial license. If your program would just cost 10 EUR or so I'd probably have bought it immediately, but with the price you are asking I definitely want to try this out thoroughly before deciding to buy it.
I hope that you'll find TSPE is worth the asking price! Of course it helps if there weren't bugs in the first place, but then it wouldn't be a beta and the price wouldn't be the same!

I am using TsRemux, not TSMuxer. I'll send you a PM with a link to the sample. With the sample both the Audio and Video stream type are set to MPEG2.
Thanks I had a look at the sample. If you look at the Decode of the PMT (Programme Map Table, PID 0x100 in your case) you can see that TsRemux has set the stream PIDs and stream types to BD specifications (almost). It has also set the audio streams to MPEG2 Audio instead of MPEG1. TSPE uses the PMT to identify the streams but as the PMT is incorrect TSPE's identification is affected. The good news is that you can change the stream type using TSPE's Table Editor.

1. Goto Edit -> Table Options -> Table Editor
2. Select the first MPEG2 Audio stream
3. Change the Stream Type to MPEG1 Audio in the dropdown list
4. Click Update
5. Select the 2nd MPEG2 Audio stream and repeat steps 3 and 4
6. Click Save
7. Make a full edit of the stream
8. Reload fixed edit back into TSPE.

You should now find MPEG1 Audio is in the tables and TSPE's mode is set accordingly.


With header checking disabled the m2ts stream scans without errors.
Great! (well not so great until m2ts header checking is improved!)


Oh, I overlooked that in the corner - thanks for the hint! It displays fine. :D
Yeah, it's tucked away in the corner, Office style, though it is easy to miss.


I would be happy if you could answer this question from my last posting: "I understand that for cut-in I should only cut at I-frames. Can I cut at P-frames, too, for cut-out or are there any reasons why I shouldn't do so?"
TSPE has "Clean Edit" intelligence which means that it will automatically attempt to find the best edit points given the set start & end points that you specify. Internally it finds the last P frame for an End point and the next I frame for the start point. Historically with GOP based editors, edit points are set at I frames for the cleanest results. Personally, I set the end points manually to the nearest P frame and the start point to the nearest I frame. Clean edits is still not perfected but is quite usable for most.

Hope that helps!

Regards,

Vent

[dhnj]

Hi folks,

dhnj - Very puzzled about TME being so picky with re-encoding. I will try to figure out what TME looks for when deciding when to re-encode or not, but it would be great if you can ask the authors of TME also if you haven't already. I hope you don't mind if I fix the "big" bugs first though! I will also work on the documentation as time permits.

Finally, thank you to all that have supported TSPE! It really means alot!

Regards,

Vent


Hi Vent

Thanks for your feedback.
Since I wrote to you last I have done many edits and all edits with 740 are Ok with TME, but not with the same edit with 755. I have uploaded a small edit(11.5 Mb) with 740 & 755
to rapidshare as follows:

1. Download Link: Click here to download file

http://rapidshare.com/files/211697415/HPOE1_740.edit.TS.html
MD5: 267150B041F5D2DFA48738AB1A139425

2. Download Link: Click here to download file

http://rapidshare.com/files/211697416/HPOE1_755.edit.TS.html
MD5: BB8EB9E52B5FA4B19D4D0B60479AB7B5

I think when you load a file into TME, it first looks at the begining of the input file to see if it is COMPLIANT or NON COMPLIANT according to their rules what is COMPLIANT or NON COMPLIANT.

I hope that the the 2 uploaded files will show you the difference.

Regards

dhnj

[joerg]

Hi Vent,

thank you for your detailed answers. I hope I'm not bugging you too much with my questions, but I've got more ;)


If you look at the Decode of the PMT (Programme Map Table, PID 0x100 in your case) you can see that TsRemux has set the stream PIDs and stream types to BD specifications (almost). It has also set the audio streams to MPEG2 Audio instead of MPEG1. TSPE uses the PMT to identify the streams but as the PMT is incorrect TSPE's identification is affected. The good news is that you can change the stream type using TSPE's Table Editor.

1. Goto Edit -> Table Options -> Table Editor
2. Select the first MPEG2 Audio stream
3. Change the Stream Type to MPEG1 Audio in the dropdown list
4. Click Update
5. Select the 2nd MPEG2 Audio stream and repeat steps 3 and 4
6. Click Save
7. Make a full edit of the stream
8. Reload fixed edit back into TSPE.

You should now find MPEG1 Audio is in the tables and TSPE's mode is set accordingly.

Even though the decode of the (broken) PMT shows the video as H.264/AVC TSPE sets the video mode it is operating with to MPEG2 (and I have to set that manually to H.264 via Options --> Video Mode --> H.264). Why?
Fixing the audio stream types worked as you described and afterwards also the video mode was set correctly when loading the stream.

The audio streams (in the original m2ts stream as well as in the remuxed and fixed stream) are displayed as being 64 kbit and 44100Hz:
PID: 0x184D: MPEG1 Audio Layer 2, 64Kbit/s @ 44100Hz
PID: 0x184E: MPEG1 Audio Layer 2, 64Kbit/s @ 44100Hz
In fact the streams are 256 Kbit/s @ 48000Hz.
So it looks like they are being mis-detected (doesn't look like this has any negative effect as far as I can tell, but I thought I'd mention it).

Another strange thing I found is that the +F button for advancing a frame doesn't seem to work properly. For example I am at position 01:18:33:28 (I,P,B slice). I click +F and I jump to 01:18:33:36 (I,P slice). I click +F again and now the video jumps back to 01:18:33:32 (I,P,B slice). Next steps are 01:18:33:34 (I,P,B slice), 01:18:33:42 (I slice), 01:18:33:38 (I,P,B slice), 01:18:33:40 (I,P,B slice). I would expect the +F button to always advance the video in time instead of jumping back and forth in time. Almost at the end there's even one point where it displays 01:18:33:54/01:18:33:52 (i.e. it says the frame being displayed is after the end of the stream, which doesn't seem to make sense).
Is there anything I am not understanding or overlooking or is this a bug?

Update: I found another likely bug: I loaded my video and picked an I slice for cut-in and cut-out. But when I load the cut-result, it turns out the cut-out-point is not the I-slice I selected, but instead the previous I,P slice. Therefore I'm losing a few frames while cutting. One of the main reasons I am looking into TSPE for editing my HDTV captures is that I want to cut my video as exact as possible, so while this is only very few frames that are lost it is an important issue for me. I hope that I'm either doing something wrong or that you can fix this.

Last but not least one more question: have you thought about adding an option to write the edit result to an mkv file instead of ts or would that be out of scope for TSPE?

[Ventolin]

Hi Vent

Thanks for your feedback.
Since I wrote to you last I have done many edits and all edits with 740 are Ok with TME, but not with the same edit with 755. I have uploaded a small edit(11.5 Mb) with 740 & 755
to rapidshare as follows:

1. Download Link: Click here to download file

http://rapidshare.com/files/211697415/HPOE1_740.edit.TS.html
MD5: 267150B041F5D2DFA48738AB1A139425

2. Download Link: Click here to download file

http://rapidshare.com/files/211697416/HPOE1_755.edit.TS.html
MD5: BB8EB9E52B5FA4B19D4D0B60479AB7B5

I think when you load a file into TME, it first looks at the begining of the input file to see if it is COMPLIANT or NON COMPLIANT according to their rules what is COMPLIANT or NON COMPLIANT.

I hope that the the 2 uploaded files will show you the difference.

Regards

dhnj
Hi dhnj,

The difference between the two files is the PAT (PID=0).

In 0.740 sample, the PAT includes a reference to a non-existent NIT (in red in the PID tree). In the 0.755 sample, the PAT does not reference the NIT as it is not present in the stream and must have been corrected while making the edit (as it should be).

So perhaps TME is looking for a PAT with a NIT? To test this, from your source file, simply check the NIT in the PID tree before you hit edit in 0.755. The NIT has a PID of 0x001F and is red in the PID tree.

Hopefully that will now make TME handle the file without re-encoding.

If that is the case, then this is a TME bug! The NIT is not mandatory for Transport Streams.

Please do let me know how you get on.

Regards,

Vent

[Ventolin]

Hi joerg,
Hi Vent,

thank you for your detailed answers. I hope I'm not bugging you too much with my questions, but I've got more ;)

Even though the decode of the (broken) PMT shows the video as H.264/AVC TSPE sets the video mode it is operating with to MPEG2 (and I have to set that manually to H.264 via Options --> Video Mode --> H.264). Why?
As mentioned before, the PMT generated by TsRemux is not correct, which causes TSPE to get things wrong. However, I have added a small modification to correct the mode setting in TSPE in such cases.

Fixing the audio stream types worked as you described and afterwards also the video mode was set correctly when loading the stream.
Great!

The audio streams (in the original m2ts stream as well as in the remuxed and fixed stream) are displayed as being 64 kbit and 44100Hz:
PID: 0x184D: MPEG1 Audio Layer 2, 64Kbit/s @ 44100Hz
PID: 0x184E: MPEG1 Audio Layer 2, 64Kbit/s @ 44100Hz
In fact the streams are 256 Kbit/s @ 48000Hz.
So it looks like they are being mis-detected (doesn't look like this has any negative effect as far as I can tell, but I thought I'd mention it).
Yes, this does look like a minor bug, it will only affect the display side of things and will not affect the output. Hopefully will have this fixed soon.

Another strange thing I found is that the +F button for advancing a frame doesn't seem to work properly. For example I am at position 01:18:33:28 (I,P,B slice). I click +F and I jump to 01:18:33:36 (I,P slice). I click +F again and now the video jumps back to 01:18:33:32 (I,P,B slice). Next steps are 01:18:33:34 (I,P,B slice), 01:18:33:42 (I slice), 01:18:33:38 (I,P,B slice), 01:18:33:40 (I,P,B slice). I would expect the +F button to always advance the video in time instead of jumping back and forth in time. Almost at the end there's even one point where it displays 01:18:33:54/01:18:33:52 (i.e. it says the frame being displayed is after the end of the stream, which doesn't seem to make sense).
Is there anything I am not understanding or overlooking or is this a bug?
I should really add this to the FAQ on the website :)

The frame step issue you are seeing is because TSPE works at very low level (right down to the Packet) - you are moving through the stream exactly how frames are stored in the stream. Have a look at the byte positions as you step forwards a frame, you'll notice they will be going up every time. Now look at the PTS/DTS timestamps in the Decode window as you frame step. You'll notice the PTS jumps backwards and forwards as you describe. The PCR timestamp will always go upwards however.

So you are able to see the frames in the stream exactly where they are in the stream!

The Frames are indeed stored in stream out of order. This is because there are different types of frames:

I - Frames (Index Frames)
P - Frames (Predicted Frames)
B - Frames (Bi-Directional Predicted Frames)

The I frame is a full frame like a jpg picture, the P frames use prediction from the previous frame(s) to create the frame and the B frames use prediction from previous and future frames to create the frame.

As frames can use previous and future frames they are stored out of order so the decoder always has enough information to create the frame without having to wait.

H.264 extends the idea from frames to slices of a picture where a frame can be made of I, P and B slices or combinations thereof.

While this is useful for the advanced users amoung us, I'll add in a mode where it works at the PTS (Presentation Time Stamp) level so frame +/- works as you would expect.

Update: I found another likely bug: I loaded my video and picked an I slice for cut-in and cut-out. But when I load the cut-result, it turns out the cut-out-point is not the I-slice I selected, but instead the previous I,P slice. Therefore I'm losing a few frames while cutting. One of the main reasons I am looking into TSPE for editing my HDTV captures is that I want to cut my video as exact as possible, so while this is only very few frames that are lost it is an important issue for me. I hope that I'm either doing something wrong or that you can fix this.
There could be a few reasons for this.

1. The end frame that you are seeing is the start of the frame in the stream. If you set the end point at this frame, then everything upto that point will be included, but not past that point. So try moving to the next frame (most likely will be a B frame) and then set the end point.

2. The Directshow Filter and TSPE are slightly out of sync. This can be tweaked a little with Menu->Options->Timestamp Mode->Preview Adjust. Try changing this from -12 to +12 and seeing how that affects start and end points.

3. Test the edit point between 2 edits by right clicking 2 entries in the EDL. Check the frames there. It might be that the Test start/end points miss a few frames, but a Test Edit Point should show all the frames as it would be in the final edit.

4. TSPE might have a bug in the Clean Edits logic, certainly this is under continual development.

Last but not least one more question: have you thought about adding an option to write the edit result to an mkv file instead of ts or would that be out of scope for TSPE?
Yes, I have thought about it alot, though MKV itself is quite a complex container, easily comparable to .TS.

I would love to add support for MKV but there are a number of issues that need to be resolved first, and it may be that MKV support is provided in a different product.

I do have some mkv streams that I do want to edit however, and I have not found a satisfactory editing solution, so MKV is definitely on my mind!

Hope that answers your questions, you are welcome to ask more :)

Regards,

Vent

[dhnj]

Hi dhnj,

The difference between the two files is the PAT (PID=0).

In 0.740 sample, the PAT includes a reference to a non-existent NIT (in red in the PID tree). In the 0.755 sample, the PAT does not reference the NIT as it is not present in the stream and must have been corrected while making the edit (as it should be).

So perhaps TME is looking for a PAT with a NIT? To test this, from your source file, simply check the NIT in the PID tree before you hit edit in 0.755. The NIT has a PID of 0x001F and is red in the PID tree.

Hopefully that will now make TME handle the file without re-encoding.

If that is the case, then this is a TME bug! The NIT is not mandatory for Transport Streams.

Please do let me know how you get on.

Regards,

Vent
Hi Vent

Many thanks for your quick reply.
I have tested your solution with 0.755 and it works great with TME(no re-encoding). Only outstanding point now is the documentation regarding Table Editor.

Regards

dhnj

[joerg]

Hi Vent,

thank you for your fast and very helpful answers.


There could be a few reasons for this.

1. The end frame that you are seeing is the start of the frame in the stream. If you set the end point at this frame, then everything upto that point will be included, but not past that point. So try moving to the next frame (most likely will be a B frame) and then set the end point.


This is what the stream I want to cut looks like around the cut-point (according to TSPE, i.e. using the +F button)

I,P
I,P,B
I,P,B
I,P
I,P,B
I,P,B
I
I,P,B
I,P,B
I,P
I,P,B
I,P,B
I,P
I,P,B
I,P,B
I <-- my cut-out-point

If I open the resulting stream the last frames look exactly like the sequence above (and look identical in the hex display), except that the final I-frame is not in the stream according to TSPE. Is that what you meant? With your explanation about the order frames are stored in this actually makes sense - in order to ever reach this last I-frame for displaying, the B-frames stored after the I-frame in the stream would be required, so it makes no sense to actually include this I-frame in the stream. So for me this basically means that I have to look for the I- or P-frame I'd like to cut at and then move ahead in the stream using the +F button until I reach the next I or P frame.

What I'd like to see is that this is handled automatically once you implement the mode for advancing frames at the PTS level (i.e. if I select the I-frame to cut at, automatically add in the frames following this one in the stream, but which are located before that I-frame looking at them at the PTS level). For the not-so-advanced users this would probably be very helpful. I am really looking forward to such a mode since my main goal is easy and accurate cutting.

btw: I noticed that the Jump button sometimes does not jump to the frame I set the cut-out at, but instead to the next frame in the stream. Maybe a minor bug?

[Ventolin]

Hi Joerg,

I'm not quite sure what you are doing still but perhaps it is easier to describe with a sample of what you want to do. Please send an edl with the sample so we both have a reference point.

Also, what video decoder are you using? FFDShow does have issues when requesting particular frames. The best directshow video filters to use for h264 are CoreAVC and from very initial tests, the MS h264 video filter from win7.

The most accurate way to see what is going on is when you make an edit point joining two video sections together, as the directshow codecs themselves need a fair bit of data to buffer in order to show what you want. So try and have 2 items in the edl and adjust the end point of the first to see what's really going on.

A final note is that TSPE isn't frame accurate yet, so might be a few frames out. This is defintely a work in progress though and I'm working hard on getting closer to that goal.

Regards,

Vent

[Ventolin]

BTW, editing your post while I'm replying to your previous post is cheating!

Vent :)

[joerg]

BTW, editing your post while I'm replying to your previous post is cheating!

Vent :)

:D
Sorry for editing - but I couldn't let that nonsense I wrote sit there and also couldn't know you were writing a reply.
So are things clearer with my edited post?

[rebkell]

:D
Sorry for editing - but I couldn't let that nonsense I wrote sit there and also couldn't know you were writing a reply.
So are things clearer with my edited post?

I think it's easier to see what is going on, if you had left the timestamps in there, because the sequence of frames were not in the sequence they would be displayed.

[Ventolin]

:D
Sorry for editing - but I couldn't let that nonsense I wrote sit there and also couldn't know you were writing a reply.
So are things clearer with my edited post?

A little bit yes, but send a sample with an edl so we can have a look at it :)

Vent

[joerg]

ok, here are the timestamps as requested by rebkell (added a few frames after my desired cut-I-frame as well)

01. I,P (01:23:59.64)
02. I,P,B (01:23:59.60)
03. I,P,B (01:23:59.62)
04. I,P (01:23:59.70)
05. I,P,B (01:23:59.66)
06. I,P,B (01:23:59.68)
07. I (01:23:59.76)
08. I,P,B (01:23:59.72)
09. I,P,B (01:23:59.74)
10. I,P (01:23:59.82)
11. I,P,B (01:23:59.78)
12. I,P,B (01:23:59.80)
13. I,P (01:23:59.88)
14. I,P,B (01:23:59.84)
15. I,P,B (01:23:59.86)
16. I (01:23:59.94) <-- my intended cut-out-point (this should be the last frame displayed by the player)
17. I,P,B (01:23:59.90)
18. I,P,B (01:23:59.92)
19. I (01:24:00.00)

If I try to set the cut-out at 16. all frames up to 15. are included in the resulting stream.
If I set the cut-point to 17., 18. or 19. all frames up to 18. are included in the resulting stream.

Vent: will try to send a sample tomorrow

[rebkell]

ok, here are the timestamps as requested by rebkell (added a few frames after my desired cut-I-frame as well)

01. I,P (01:23:59.64)
02. I,P,B (01:23:59.60)
03. I,P,B (01:23:59.62)
04. I,P (01:23:59.70)
05. I,P,B (01:23:59.66)
06. I,P,B (01:23:59.68)
07. I (01:23:59.76)
08. I,P,B (01:23:59.72)
09. I,P,B (01:23:59.74)
10. I,P (01:23:59.82)
11. I,P,B (01:23:59.78)
12. I,P,B (01:23:59.80)
13. I,P (01:23:59.88)
14. I,P,B (01:23:59.84)
15. I,P,B (01:23:59.86)
16. I (01:23:59.94) <-- my intended cut-out-point (this should be the last frame displayed by the player)
17. I,P,B (01:23:59.90)
18. I,P,B (01:23:59.92)
19. I (01:24:00.00)

If I try to set the cut-out at 16. all frames up to 15. are included in the resulting stream.
If I set the cut-point to 17., 18. or 19. all frames up to 18. are included in the resulting stream.

Vent: will try to send a sample tomorrow

Well, that makes sense, you can play with the Edit->EDL Option->Include Packet at End Point? , but since you want to cut out at 16, then you have to display 17 and 18 because they actually come before 16 according to the timestamps, the timestamps are the sequence they are displayed in, the actual frame display order is 17, 18, 16 ... but since 17 and 18 depend on the information in 16, that's about the best you can get is the three frame resolution, without flickering and corruption, at least that's been my experience.

[rebkell]

I've noticed when I drag an drop a file, that if I try to save the edl file, that it will default to the last location I saved an edl file in. I would think it should save it in the current directory of the ts file I'm currently editing.

[Chumbo]

I've noticed when I drag an drop a file, that if I try to save the edl file, that it will default to the last location I saved an edl file in. I would think it should save it in the current directory of the ts file I'm currently editing.
That's the default behavior of this standard Windows dialog.

[rebkell]

That's the default behavior of this standard Windows dialog.

I can understand that, but it's not really a good fit, you can change that default to the current directory of the current file you are editing, I would assume. I haven't programmed in C++ in a while, but it's pretty simple in C#, and I would assume it's pretty similar in C++.

[Chumbo]

I can understand that, but it's not really a good fit, you can change that default to the current directory of the current file you are editing, I would assume. I haven't programmed in C++ in a while, but it's pretty simple in C#, and I would assume it's pretty similar in C++.
Yes, it is easy in c++ too. The developer just has to wire it up that way.

[rebkell]

Yes, it is easy in c++ too. The developer just has to wire it up that way.

We'll call it a feature request. ;)

[Ventolin]

Hi rebkell, Chumbo

I've wired it up that way now just for you :)

Cheers,

Vent

[rebkell]

Hi rebkell, Chumbo

I've wired it up that way now just for you :)

Cheers,

Vent

Thanks.
:thanks:

[Chumbo]

Hi rebkell, Chumbo

I've wired it up that way now just for you :)

Cheers,

Vent
Very cool. Thank you. :)

[rebkell]

I have another feature addition request :o Would it be possible to increase the time for test edits. Sometimes I have I frames farther apart than your test window time and I really can't see the test edit. It will be just green screen with audio until the test/cut point hits. It would be nice if we could adjust the test edit total time.

The ideal option IMO would be the ability to set the test point time in a variable, like it was in h264ts_cutter.

[Ventolin]

Hi rebkell,

How about the larger of 3 seconds or +/- 2 GOPs? Will add user definable duration later.

Vent

[rebkell]

Hi rebkell,

How about the larger of 3 seconds or +/- 2 GOPs? Will add user definable duration later.

Vent

GOP's would probably be fine, it should assure a good start point, some of the GOPs can be 5 to 10 seconds apart on my videos, as I would figure on a lot. I think x264 has a default of 240 frames. Which depending on FPS could be 10 seconds or so.

[joerg]

Vent: will try to send a sample tomorrow
Well - I got distraced by other stuff and put this aside for a while - sorry for that!
When I wanted to start working on this again now, I noticed that due to the long pause in my tests my trial of TSPE ran out. Currently I am not ready to buy it, so unfortunately I am not able to provide the sample you requested - sorry!

[Ventolin]

Well - I got distraced by other stuff and put this aside for a while - sorry for that!
When I wanted to start working on this again now, I noticed that due to the long pause in my tests my trial of TSPE ran out. Currently I am not ready to buy it, so unfortunately I am not able to provide the sample you requested - sorry!
Hi joerg,

I can send you a temp licence extension if you like, send me a pm if you want one.

The next upcoming version will have timelines, with frame re-ordering so you can now frame step in PTS order.

This should make it easier to understand what is going on.

Regards,

Vent

[joerg]

Hi Vent,

thanks for your very positive reply and kind offer for an extended trial license. Based on this reply and your previous support in this thread I decided to go ahead and buy the software rightaway.

There are two things that made me hesitate buying and I want to let you know about them:
1. I hate hardware-locked software. Not that I'd want to use it on more than one machine, but it keeps me dependent on you if I change my system. While you issue free replacement licenses today, you may decide that replacement licenses will no longer be issued for free at any time you want or you may go out of business. That gives me a pretty bad feeling. I'll have to change my system fundamentally in the next few months (it is more than 4 years old and not capable of playing the HD material that I capture), so I hope getting the replacement license will be painless.
2. I still think the price is a bit high. But your responsiveness and helpfulness on this forum convinced me ;)

I'm looking forward to receiving the license key, which I just purchased.

To keep this at least a bit on topic: Do you have any (approximate) idea when the new version you talked about may be ready? Is it days, weeks or months away? Before posting a sample etc. I'll wait for the new version with frame-stepping in PTS order. Maybe this will already solve all my "problems". Will this new mode also allow me to see which frames I can actually (really) cut at? It would be really helpful if it would allow to a) step through the video looking for possible cut-in-points and b) step through the video looking for possible cut-out-points. I envision two buttons in the PTS-order mode for this: a) jump to next possible cut-in-frame and b) jump to next possible cut-out-frame. This would make exact cutting that much easier and make TSPE the perfect tool (at least for my purposes). With MPEG2 video this was very easy: cut in at I frame, cut-out at I-frame or P-frame. With the HD video formats things got much more complicated. :(

[Ventolin]

Hi Joerg,

Thank you very much for your support!

I'm not a fan of hardware locks either, but I would also hate to see pirate copies readily available especially when people have paid good money. I doubt also TSPE would be where it is today without the support of people such as yourself! Every sale gives a very welcome boost in determination to make TSPE the best I can!

I'm working on the Timelines at the moment, I hope to have that finished by the end of the month. I don't know if I'll add in some of the forthcoming Muxer functions before the next release is available however, but if I did, the timeline would be more useful.

Having a P frame button is a good idea, will see if I can include something like that in the next release or so.

Editing with TSPE should still be as easy as selecting an I frame for the start point and a I or P frame as an end point. "Clean Edits" is supposed to do this! Hopefully with Frame Accuracy it will be even easier.

As for the sample, please do send it to me! I will have a look at the issue you are having and if need be, change the code for the next release.

Regards,

Vent

[Chumbo]

Hi Vent,

Any plans for allowing to save in a different format than being edited? Would be nice to open a .ts file but save as .m2ts or vice versa and so on. Thank you.

[prenz]

@ventolin,

I have problems editing ITVHD streams (sample: http://rapidshare.com/files/233207948/itvhd.ts.html) with TSPE .755

Error: Auto Creating Filter Graph
Beim Aufruf einer COM-Komponente wurde ein HRESULT E_FAIL-Fehler zurückgegeben.
Error Loading Video, Disabling.)

I can't set any start- or endpoints. When I try to save the "edit", it comes up with "Error: PID tree failed validation checks".
I recorded this on an UFS910 with Enigma2. BBCHD and C4HD from the same machine are fine.

[Ventolin]

@ventolin,

I have problems editing ITVHD streams (sample: http://rapidshare.com/files/233207948/itvhd.ts.html) with TSPE .755

I can't set any start- or endpoints. When I try to save the "edit", it comes up with "Error: PID tree failed validation checks".
I recorded this on an UFS910 with Enigma2. BBCHD and C4HD from the same machine are fine.
Hi Prenz,

Thank you very much for your sample!

Your ITVHD sample has no PMT which means TSPE can not correctly identify the streams and will not let you make an edit as all transport streams must have a PAT and a PMT. TSPE should be able to fix this however with it's Table Editor.

However, this is a bad news, good news, good news situation.

The bad news:
There is a bug with the Table Editor which means 0.755 can not fix this by itself. (This is good news really though as I like getting bugs like this as TSPE becomes better once the bug is fixed!)

The good news:
The next upcoming beta has this issue already fixed, though not completely but enough to fix the stream.

The good news 2:
It's possible to fix this with 0.755 with a little help from a working ITVHD cap. All you need to do is copy just the PAT and PMT from the working ITVHD cap to the cap you need to fix.

You can do this 2 ways:
First Way (the hard expert way):
a. Set the start/end packets around the PAT and PMT
b. Add to the EDL.
c. Uncheck Fix Tables and Fix Timecodes
d. Click edit.

(second way joins here)
You should end up with a 1K (2x 188 byte packet) edit.
e. Open this 2 packet .ts file back into TSPE and ignore the warning messages
f. Add to the EDL
g. Load your ITVHD ts file
h. Add to the EDL
i. Make sure Fix Tables and Fix Timecodes is unchecked
j. Click Edit
k. Proceed to Combined Step below

Second Way (Easier Way)
a. Download the pre-edited ITVHD PAT/ PMT file from this URL:
http://www.bitstreamtools.com/download/ITVHD PAT PMT.rar
b. Unrar and save in the same folder as your ITVHD.ts
c. Follow from step e above.

You should now have an edit that should play in a player and give you video preview in TSPE. However, the PIDs still need tidying up.

Combined Steps:
l. Menu->Edit->Table Options -> Table Editor
m. Delete all red entries (missing references in the PAT) if present.
n. Change the order of the H264 and AC3 PIDs by drag/dropping so the AC3 PID comes before the H264 PID. (This is really just to make sure the Table editor's PAT and PMT get used instead of the original PAT/PMT in the stream. You should now have:

PAT PID 0x000
-> PMT PID 0x0D48
--> 0x0D49 H.264 / AVC Video
--> 0x0D4A AC3 Audio

o. Clear the EDL
p. Add to the EDL
q. Click Edit

You should now have a working ITVHD .ts file.

So, we have now used TSPE in Packet Edit mode to fix this issue.

The above is quite complicated and the procedure is quite advanced. I would rate this at 8/10 difficulty.

If this is too much for you, then please be patient until the next beta is ready. You will be able to fix your ITVHD caps by just using the Table Editor.

Hope that helps,

Regards,

Vent

[prenz]

Cheers vent,

The "Hard Expert Way" failed due to the fact I don't have an ITVHD recording with intact PAT and PMT, so I went for the easy way, gone through it step by step and have been left with a 20 MB-file (out of 13 gigs) with no video or audio inside.

So I went for a third option which I would call "the lazy man's way": if the PAT and PMT is missing, why not just prepend ITV PAT PMT.TS right in front of my itvhd.ts - after doing so I could edit my recording without any problems whatsoever. After all, I use TSPE in this case only to edit out the commercials and then demux it right away, which worked beautifully.

So thanks again for being very helpful!

[joerg]

Hi Vent,

I have finally performed more testing and prepared the sample you asked. You can find it at http://rapidshare.com/files/233814814/sample20090517.zip.html. I have added an edl as well as the corresponding .ts file (actually it is an m2ts-file).
As you can see, the edl sets the cut-out at 0:06.70 (an I-Slice).

This is the list of frames around my cut-point (I have numbered them for easier reference):

01. I (0:06.52)
02. I,P,B (0:06.48)
03. I,P,B (0:06.50)
04. I,P (0:06.58)
05. I,P,B (0:06.54)
06. I,P,B (0:06.56)
07. I,P (0:06.64)
08. I,P,B (0:06.60)
09. I,P,B (0:06.62)
10. I (0:06.70) <-- my intended cut-out (last frame I want to see)
11. I,P,B (0:06.66)
12. I,P,B (0:06:68)
13. I (0:06.76)
14. I,P,B (0:06.72)
15. I,P,B (0:06.74)

1) I had to set a preview frame offset adjustment of 3 to get a correct preview. It the problem with my system or TSPE? (This is using CoreAVC). What definitely seems to be a problem with TSPE, however, is the fact that any offset adjustment setting I make is not persistent and re-set once I restart TSPE. Edit: The setting of the "Strip m2ts Extra Headers?" option is also not saved - each time I start TSPE it is not set. The "Save Settings on exit?" option is enabled.

2) If I click the "Jump" button for the End Point the time jumps to 0:06.66 (frame 11) instead of 0:06.70 (frame 10). If I then click "F+" once it stays at that frame/time code. It should obviously jump to 0:06.70 (frame 10) if I hit the "Jump" button.

3) If I cut with the EDL, all frames up to 10. are included in the resulting stream. What I would expect is all frames up to 12. being included (so that the selected frame is the last one being displayed). With the way this is being cut now, the result is that in the timeline frames 11 and 12 are missing. I have to cut at frame 12. explicitly to get a proper result. With further testing I found that apparently for m2ts-files clean edit does not work.

4) The audio in the cut result stops before the video does (i.e. there is a bit of audio missing at the end) - any idea what might be the reason? Is there anything I can do to avoid this? EDIT: I noticed that once I convert the resulting ts to mkv using GDSMux the "missing" audio is there - just when I play the ts it isn't - strange... So it doesn't really look like a problem in TSPE, but if you have any idea why this is I would be interested.

After noticing that the clean cut feature only seems to work with t2 files I converted the whole file to .ts by enabling the "strip m2ts extra headers" option and then cutting again from beginning to end to get a real clean cut. Since the original stream did not start with an I-frame, the timecodes for the frames listed above changed, so here is the list for the second sample, which you can download at http://rapidshare.com/files/233828890/sample20090517_2.zip.html:

01. I (0:06.24)
02. I,P,B (0:06.20)
03. I,P,B (0:06.22)
04. I,P (0:06.30)
05. I,P,B (0:06.26)
06. I,P,B (0:06.28)
07. I,P (0:06.36)
08. I,P,B (0:06.32)
09. I,P,B (0:06.34)
10. I (0:06.42) <-- my intended cut-out (last frame I want to see)
11. I,P,B (0:06.38)
12. I,P,B (0:06:40)
13. I (0:06.48)
14. I,P,B (0:06.44)
15. I,P,B (0:06.46)


5) If I set the cut-out at frame 10, the final frame that is included in the resulting stream is frame 9. The correct cut would be to include all frames till frame 12. If I set the cut-out at frame 11, 12 or 13, the final frame that is included in the resulting stream is frame 12. This seems correct for frames 11 and 12, but if frame 12 is selected, frame 15 should be the final frame that is included.

If the links do no longer work when you read this, please let me know - then I'll send you a new link via PM.

[joerg]

I think found another bug:
If I take an m2ts file, enable "Strip m2ts extra headers" and enable "Fix tables" at the same time, the resulting output file is broken - the video won't play, TSPE reports:

Error: Auto Creating Filter Graph
Error HRESULT E_FAIL has been returned from a call to a COM component.
Error Loading Video, Disabling.
Internal Viewer Closed.

And if I try to open the resulting file in TSPE, it reports:
Opening File...
Unknown File Type, aborting

If I do this in a two-step process (first remove m2ts headers without fixing tables, then load the resulting ts and use "fix tables" everything works fine. But obviously I'd like to avoid that extra step.

You can reproduce that with the first sample package from my last post.

[Ventolin]

Cheers vent,

The "Hard Expert Way" failed due to the fact I don't have an ITVHD recording with intact PAT and PMT, so I went for the easy way, gone through it step by step and have been left with a 20 MB-file (out of 13 gigs) with no video or audio inside.

So I went for a third option which I would call "the lazy man's way": if the PAT and PMT is missing, why not just prepend ITV PAT PMT.TS right in front of my itvhd.ts - after doing so I could edit my recording without any problems whatsoever. After all, I use TSPE in this case only to edit out the commercials and then demux it right away, which worked beautifully.

So thanks again for being very helpful!
Hi Prenz,

Glad to know you have your edit working!

Your Third Option was actually my 2nd option though I guess it was harder to write than it was to do :)

Cheers,

Vent.

[Ventolin]

Hi Joerg,

Thanks for your sample and your tests!

I've had a look at the issues you are seeing. (The numbers below aren't matching with your post)

1. With this sample you do need a preview offset of +3. The preview offset is dependant on the stream though which is why it is not remembered when you restart TSPE. I will have a look at ways to improve this situation.

2. When you set an end point, what you are actually setting the exact point at which to stop the edit. So if you are setting an I frame then everything upto that I frame is inlcuded but not the I frame itself. You would need to set the end point to the next frame in the PTS order.

This I agree is a bit confusing so I will have a look at changing this so the current frame you are looking at is also included.

However, if you do move past the I frame to the next B frame in PTS order, the resulting output is missing the last B frame before the target I frame (in PTS order) which is incorrect. This needs fixing.

3. M2TS vs TS - the sample has shown a few bugs which I need to fix before I can do any further analysis. Thanks for the sample again!

4. Strip M2TS headers with Fix Tables is indeed causing an issue. However, I don't have this issue with my dev build so this bug might already be fixed. I'll investigate further.

5. Jump time inaccuracy - this is a bug that I've known about but haven't got around to fixing yet.

6. Audio - There is an offset between video and audio as you can see from the timestamps of each. In your sample for instance it looks to be about 1.1 seconds. Video arrives in the decoder before the audio as it takes longer to decode the video than the audio. Since the video stream finishes before the audio stream, there are alot of audio packets at the end with no video frames physically in the stream (DTS order as opposed to PTS order). The decoder should output the audio frames, but it seems the decoder stops when the video stream stops as opposed to when the audio stream stops. This results in the audio cutting off before the video. I need to do some more analysis though to confirm this is the case but it looks like this is a design issue in the decoder.

The workaround to this would be to insert blank video frames at the very end of the edit, but it should be OK at an edit point as the next segment will have video first then audio to match.

Still quite a bit of work to do it but I'm working on it!

Thanks for your sample and your analysis :)

Regards,

Vent

[joerg]

Hi Vent,

thanks for your fast reply. I'm glad I could help with the sample. I have a few more questions regarding your reply:


1. With this sample you do need a preview offset of +3. The preview offset is dependant on the stream though which is why it is not remembered when you restart TSPE. I will have a look at ways to improve this situation.

If this is stream-dependent, is there a good and reliable way for me to find the correct offset number?
Also please note that the "strip m2ts extra headers" option is also not remembered.


However, if you do move past the I frame to the next B frame in PTS order, the resulting output is missing the last B frame before the target I frame (in PTS order) which is incorrect. This needs fixing.

I haven't seen that kind of problem. If I set the cut-out at frame 11 from my example and frame-list in my previous post, then frame 12 is also included. So to me it looks correct (unless I'm using m2ts streams).


6. Audio - There is an offset between video and audio as you can see from the timestamps of each.


Thank you for the explanation on the audio issue. Where can I see those timestamps you are referring to? (I can see video timestamps below the preview window, but where do I see the corresponding audio timestamps?).

[Ventolin]

Hi Vent,

If this is stream-dependent, is there a good and reliable way for me to find the correct offset number?
Also please note that the "strip m2ts extra headers" option is also not remembered.
I don't have a good way to do that apart from trial and error. TSPE asks the codecs for the frame and the codecs then displays it. It all depends on how the codecs numbers frames which I don't have access to. One day I'll write my own codecs but that's quite a way off at the moment.

I haven't seen that kind of problem. If I set the cut-out at frame 11 from my example and frame-list in my previous post, then frame 12 is also included. So to me it looks correct (unless I'm using m2ts streams).
I will look further into it, but there is something amiss!

Thank you for the explanation on the audio issue. Where can I see those timestamps you are referring to? (I can see video timestamps below the preview window, but where do I see the corresponding audio timestamps?).
The timestamps are in the Decode Window - look at the PTS times (use the View button to rotate the view to the Packet Decode Window).

I've now made the strip m2ts header remembered :)

Cheers,

Vent

[joerg]

Hi Vent,

I can see the PTS times - but this is for video frames, isn't it? How can I see the times for the audio frames to compare the two and see (calculate) a difference/delay? If I set the "Frame step" mode to audio and step to an audio frame I get something like this:
---------------
PID: 0x184D
Continuity Counter: 0
Attributes: MPEG1 Audio Layer 2, 64Kbit/s @ 44100Hz
Transport Scrambling Control: 0: (no scrambling of TS packet payload)
Adaptation Field Control: 1

MPEG1 Audio Header:
ID: True (MPEG1 Audio ISO/IEC 11172-3)
Layer: 2 (Layer Code: 2)
protection_bit: False (CRC Present)
Bitrate: 64 KBit/s (bitrate_index: 4)
Sampling Frequency: 44100 Hz (sampling_frequency code: 0)
padding_bit: False
private_bit: False
mode: 3: single_channel
mode_extension: 0: subbands 4-31 in intensity_stereo, bound==4
copyright: True
original / home: False (home)
emphasis: 3: CCITT J.17
crc_check: 733A
----------------

...and there is no PTS time displayed. I think there must be something that I do not (yet) understand.

[Ventolin]

Once you have arrived at the first audio packet, click PTS+

This will jump to the first audio PES frame with a PTS. It's then the difference between the Video Frame PTS and audio PES frame PTS.

Just some theory:

Video Frame -> PES frame -> TS Packet
Audio Frame -> PES frame -> TS Packet

PES frame has PTS, DTS, other info, TS Packet has PID, continuity counter, flags, PCR.

Video frames and audio frames can be split over multiple PES frames as necessary.

PES frames are split into many TS packets.

So there are 3 layers:

Elementary streams (Video, Audio, Subtitle, etc)
Packetised Elementary Streams (PES frames, PTS, DTS, etc)
Transport Packets (Continuity Counter, flags, PCR, etc)

All will be alot clearer with the next release as you will be able to see the PES frames in PTS sorted order or in physical order in the transport stream.

Hope that helps,

Regards,

Vent

[joerg]

Hi Vent,

thanks for the explanation - that helped a lot!
Playing around with other caps I made, I found another stream for which the preview doesn't work at all. I ran this stream twice through TSPE to remove the m2ts headers and to fix the tables before starting with the real cutting, so the stream should be clean.
The entire stream is about 105 minutes long. At around 5:22 minutes, the preview window shows the beginning of the movie. But if I cut at this position, the resulting video starts about 9 seconds too early.
To test this some more I cut a segment of the stream from the beginning of the stream till somewhere around 8 minutes. I then loaded this stream and again tried to find the point where the preview window shows the beginning of the movie. This time this was at around 5:29 (instead of 5:22). I made the cut from this position and looked at the result - still about 3 seconds off.

So something with the preview and this stream must be seriously wrong. Do you have any explanation for this? I can't really post a proper sample this time, since that would be much too big to upload for me (the 8 minutes segment is about 450MB in size) - but if there's anything else I can do to help diagnose this, I'd be happy.

[Ventolin]

Hi Vent,

thanks for the explanation - that helped a lot!
Playing around with other caps I made, I found another stream for which the preview doesn't work at all. I ran this stream twice through TSPE to remove the m2ts headers and to fix the tables before starting with the real cutting, so the stream should be clean.
The entire stream is about 105 minutes long. At around 5:22 minutes, the preview window shows the beginning of the movie. But if I cut at this position, the resulting video starts about 9 seconds too early.
To test this some more I cut a segment of the stream from the beginning of the stream till somewhere around 8 minutes. I then loaded this stream and again tried to find the point where the preview window shows the beginning of the movie. This time this was at around 5:29 (instead of 5:22). I made the cut from this position and looked at the result - still about 3 seconds off.

So something with the preview and this stream must be seriously wrong. Do you have any explanation for this? I can't really post a proper sample this time, since that would be much too big to upload for me (the 8 minutes segment is about 450MB in size) - but if there's anything else I can do to help diagnose this, I'd be happy.
Hi Joerg,

The only thing I can suggest is to try and cut the sample size down to something that still shows the issue and then send that to me. When you cut it down, turn off both Fix Tables and Fix Timecodes.

Cheers,

Vent

[joerg]

Hi Vent,

I can't cut it down to something that shows the problem and is still small enough for me to upload. I guess I'll just try to cut it using the "test" option for cut-in and cut-out. I'll keep the original stream so I can re-try with the next version of TSPE.

Update: I finally managed to cut the video, finding the right cut points using the "test" buttons. While trying to find the right cut-point I noticed something very strange. Around my cut-out-point somewhere between 1h20min and 1h30min into the stream, I had a discrepancy between the preview window and the actual cut result of about 4 minutes (i.e. the preview showed a frame 4 minutes ahead of the actual cut point) . I started stepping through the video in 10s steps and very close to my planned cut-out-point the displayed preview video suddenly "jumped" back in time, displaying the correct position in the preview window. Unfortunately (as I said before) I think it is not possible to generate a small enough sample to demonstrate this problem.
Scanning the ts file with TSPE shows no problems at all:

Continuity Errors: 0, Transport Error Indicator Errors: 0
Estimated Continuity Breaks: 0

Update2: The problem exists with CoreAVC and ffdshow as decoder - switching from CoreAVC to ffdshow didn't make any difference.

[Ventolin]

Hi Vent,

I can't cut it down to something that shows the problem and is still small enough for me to upload. I guess I'll just try to cut it using the "test" option for cut-in and cut-out. I'll keep the original stream so I can re-try with the next version of TSPE.

Update: I finally managed to cut the video, finding the right cut points using the "test" buttons. While trying to find the right cut-point I noticed something very strange. Around my cut-out-point somewhere between 1h20min and 1h30min into the stream, I had a discrepancy between the preview window and the actual cut result of about 4 minutes (i.e. the preview showed a frame 4 minutes ahead of the actual cut point) . I started stepping through the video in 10s steps and very close to my planned cut-out-point the displayed preview video suddenly "jumped" back in time, displaying the correct position in the preview window. Unfortunately (as I said before) I think it is not possible to generate a small enough sample to demonstrate this problem.
Scanning the ts file with TSPE shows no problems at all:

Continuity Errors: 0, Transport Error Indicator Errors: 0
Estimated Continuity Breaks: 0

Update2: The problem exists with CoreAVC and ffdshow as decoder - switching from CoreAVC to ffdshow didn't make any difference.

Hi Joerg,

It sounds like Haali is having trouble with the stream then.

The work around is to use Auto-Preview which actually makes a test edit in the background and then displays it in the main Video Window. You loose scrubbing ability as there is a bit of a delay but if you navigate with the buttons or with the keyboard you should be able to get somewhere. The delay would be about a second or two.

To turn on Auto-Preview, Menu-Options -> Timestamp Mode -> Auto-preview.

Hope that helps.

Vent

[rebkell]

Any estimates on when the next revision might come out Vent? No problems with this one, it's working pretty well.

[joerg]

To turn on Auto-Preview, Menu-Options -> Timestamp Mode -> Auto-preview.


Thanks for the hint. I'll try to use that option if I run into such a stream again. I have quickly tested it on a stream. It seemed to work basically ok, but it came back with a black frame instead of the actual frame quite often (I used the I+ and I- buttons to navigate).

One (not really important) request for convenience: Could you make TSPE remember the target directory for edits/cuts (or add an option to do this if not everybody wants this)? Currently it always resets it to the same directory where the source file lies as soon as a new file is loaded.
To improve performance for cutting I have two harddrives, so I can always have different source and target harddrives when doing cuts. I usually have a directory with all source files on one harddrive and put the editing results of all those source files into a fixed target directory on the other drive. So the output directory stays the same for several edits.

I'm looking forward to the next version - being able to cut my m2ts streams in one go will save a lot of time! :)

[Ventolin]

Any estimates on when the next revision might come out Vent? No problems with this one, it's working pretty well.

Hi rebkell,

Still working on the next release, was hoping to have it ready by the end of May but wanted to add in comskip / showanalyzer support. Am adding alot of code for batch processing also, I don't know if batch will be ready for the next release, but the way things are evolving at the moment, it could well be.

Scan and Edit engines will be the same, they will be updated in the release after this.

Major new features of the next version (0.80) will be:
Timelines (yes, there's more than 1!)
Import of different EDLs (comskip / showanalyser / others)
New TSPE EDL format.

Version 0.90 will have hopefully:
Mux Engine (replacing existing Edit Engine)
Timeline enhancements (tied to Mux Engine)
New Scan engine (hopefully)

And for 1.0:
Frame accurate video editing (for MPEG2 / h.264) - VC-1 frame accurate editing to follow.
GUI update

Tentative date for 0.80 end of 2nd Week of June, but don't hold me to that!

Regards,

Vent

[Ventolin]

Hi joerg,
Thanks for the hint. I'll try to use that option if I run into such a stream again. I have quickly tested it on a stream. It seemed to work basically ok, but it came back with a black frame instead of the actual frame quite often (I used the I+ and I- buttons to navigate).
Sorry to hear you are getting black frames with autopreview - it could be that the test size is too small. I still have that to fix!

One (not really important) request for convenience: Could you make TSPE remember the target directory for edits/cuts (or add an option to do this if not everybody wants this)? Currently it always resets it to the same directory where the source file lies as soon as a new file is loaded.
To improve performance for cutting I have two harddrives, so I can always have different source and target harddrives when doing cuts. I usually have a directory with all source files on one harddrive and put the editing results of all those source files into a fixed target directory on the other drive. So the output directory stays the same for several edits.
The new TSPE EDL format will support specifying the output filename but adding in the output folder would need some gui work to be able to offer the option. Not sure I'll be able to add it in for the next release though.

I'm looking forward to the next version - being able to cut my m2ts streams in one go will save a lot of time! :)
Hmmm, I don't know if it would be possible to correct the tables and edit in one operation. The main issue is that the directshow filters will be working on the sourcefile as opposed to the fixed output file which is yet to be created.

Keep bugging me about your issues though! (pun intended!)

Regards,

Vent

[dylmaster]

Hi Vent
I'm a paid up TSPE customer and wondering if you've got any ideas about the widespread problems with lip-sync on ITV HD on Freesat in UK. I record the TS from DVB Viewer and I can watch it back within DVB Viewer with good lip sync. I also get good lip sync with your preview in TSPE. I cannot play it in other players like MP Classic and WMP, or (more important) via ethernet to my network Media Player that feeds the TV in my lounge without a sync error of several seconds! I get the same (bad) results with your (edited) output and also after feeding the TS files through all the TS 'fixer' apps that I have found!
I am wondering if it is simply the rather extreme compression that they are using? BBC HD (also on Freesat) has a similar format but uses more data bandwidth, and gives me no problems with lip sync.
A quick surf of TV forums reveals many many people with all manner of PVR methods all suffering similar problems with recorded ITV HD. To me the crux is that it plays fine within DVBViewer (and TSPE). I have several H.264 decoders loaded.

Go easy on all the nuts and bolts for an old fart!

[Ventolin]

Hi Vent
I'm a paid up TSPE customer and wondering if you've got any ideas about the widespread problems with lip-sync on ITV HD on Freesat in UK. I record the TS from DVB Viewer and I can watch it back within DVB Viewer with good lip sync. I also get good lip sync with your preview in TSPE. I cannot play it in other players like MP Classic and WMP, or (more important) via ethernet to my network Media Player that feeds the TV in my lounge without a sync error of several seconds! I get the same (bad) results with your (edited) output and also after feeding the TS files through all the TS 'fixer' apps that I have found!
I am wondering if it is simply the rather extreme compression that they are using? BBC HD (also on Freesat) has a similar format but uses more data bandwidth, and gives me no problems with lip sync.
A quick surf of TV forums reveals many many people with all manner of PVR methods all suffering similar problems with recorded ITV HD. To me the crux is that it plays fine within DVBViewer (and TSPE). I have several H.264 decoders loaded.

Go easy on all the nuts and bolts for an old fart!
Hi dylmaster,

I don't have a sample long enough of ITVHD to be able to analyse what the issue might be. If there is a way you can upload a sample that is long enough to demonstrate the problem, I will gladly have a look. If not, perhaps you might be able to point me to somewhere that may have some samples large enough? Of course any sample would need to be "untouched" meaning they would have to have the adverts intact if there were any.

Outside of that I can only guess that there are different sync offsets for different segments of the programme. If that were the case, then when the file was edited, the different syncs would add up and give overall desync.

I'm working on making it easier to resync streams, the next version (0.80) will have half the work ready to do this, version 0.90 will hopefully have a new muxing engine which should enable resync of bad streams such as you describe.

In the meantime, does the desync happen when you pass it without making any edits in TSPE (with Fix Timecodes checked)?

Have you tried demuxing and remuxing the source (no edits) with something like TSMuxer?

Do you have desync in PowerDVD 7.3+ ?

I hope you are able to send me a sample that demonstrates the issue so I can work out what's wrong and provide a fix!

Regards,

Vent

[dylmaster]

Here is a very short sequence recorded without any ad breaks but showing all the symtoms previously mentioned:
http://rapidshare.com/files/247042903/06-14_21-22-15_ITV_HD_.ts.html

ITV HD is only broadcasting a few hours a week so chances to record new samples are few and far between!

[Ventolin]

Here is a very short sequence recorded without any ad breaks but showing all the symtoms previously mentioned:
http://rapidshare.com/files/247042903/06-14_21-22-15_ITV_HD_.ts.html

ITV HD is only broadcasting a few hours a week so chances to record new samples are few and far between!

Hi dylmaster,

Thanks for the sample.

I have made a test edit by loading the source and pressing the edit button and then tested the output in TSPE's preview window, MPC with Haali + CoreAVC and also with Haali + FFDShow but found no sync issues with the source or test edit.

Is there something I need to do to be able to reproduce desync other than the above?

Note that in MPC you need to disable the internal splitter by unchecking the MPEG/TS/PVA option, (if you have MPC-HD uncheck the h.264 options as well). This will enable Haali as the splitter.

Also, this sample has long GOPs - 62 fields (interlaced frame) so be sure to set your start points at I frames and note where your P frames are for end points. TSPE should find these automatically, but if you set them manually you have more control over where the edit point will be.

I don't know what networked media player you have and would be hard to help with that anyway as I don't have one at all to test with. If you do have issues with the *source* with the networked media player, I would advise getting in touch with the manufacturer to find out if they can help.

I suspect it may a direct show filter setup that may be causing the desync. Do you get the Haali icon when you play in MPC?

Regards,

Vent

[dylmaster]

OK - doing some more experiments there is obviously quite a complex scenario with my various players (and choice of decoders) on my HTPC. So let us initially concentrate on my Network Media Player (a Buffalo LinkTheater HD).
It turns out that the audio sync problem with ITV HD is fixed if I run the .ts file through tsMuxeR and set it to TSMux. I'm not sure exactly what it does but it fixes the problem! Funnily enough I was already using this technique to cure another problem with BBC HD where the Buffalo wouldn't recognise the AC3 track.
Since I want to edit most recordings, it would be great if TSPE could do whatever tsMuxeR is doing! I could post you some more examples if required.

[Ventolin]

OK - doing some more experiments there is obviously quite a complex scenario with my various players (and choice of decoders) on my HTPC. So let us initially concentrate on my Network Media Player (a Buffalo LinkTheater HD).
It turns out that the audio sync problem with ITV HD is fixed if I run the .ts file through tsMuxeR and set it to TSMux. I'm not sure exactly what it does but it fixes the problem! Funnily enough I was already using this technique to cure another problem with BBC HD where the Buffalo wouldn't recognise the AC3 track.
Since I want to edit most recordings, it would be great if TSPE could do whatever tsMuxeR is doing! I could post you some more examples if required.
HI dylmaster,

Glad to hear you have found a solution!

Hopefully the Muxing Engine will be the major feature of TSPE 0.90 - the next release after the upcoming release (0.80) which is Timelines.

After those will be Frame Accurate which will take us upto 1.0!

Regards,

Vent

[dylmaster]

To put this back into context, If I take a film I've recorded and run it through tsMuxeR, then all is well and I get good sync. If I then edit out the ad breaks with TSPE, it loses sync after the first edit. Running the edited file through tsMuxeR doesn't make any difference. I was careful to only edit on I frames.
I shall look forward to the later versions and keep you in touch with my success (or failure!).
cheers

[rebkell]

To put this back into context, If I take a film I've recorded and run it through tsMuxeR, then all is well and I get good sync. If I then edit out the ad breaks with TSPE, it loses sync after the first edit. Running the edited file through tsMuxeR doesn't make any difference. I was careful to only edit on I frames.
I shall look forward to the later versions and keep you in touch with my success (or failure!).
cheers

eac3to will probably cure the edited file gaps for you. All edits seem to leave small gaps in the audio at the edit points. My players don't mind, since TSPE keeps the timestamps aligned.

But if you try to remux the audio and video after editing it will be out of sync for sure.

If you plan on remuxing after editing or if your player doesn't handle the audio gaps or if you want to convert the edited file to an mkv or something, I'd advise eac3to, it will create an audio and video stream that will sync up and can be muxed back together in your choice of muxers.

[dylmaster]

I'm doing nothing other than cut editing on I frames. It jumps out of sync by about a second after the first edit!.
I'll try eac3to when I get a chance (just off to work).

[dylmaster]

eac3to seems to be an audio decoder. My problem is that the recorded TS files don't seem to have the audio and video properly 'connected' for my Network Media Player (or for several computer based players). There is clearly more than one way to do this since DVBViewer and TSPE both play the recorded files fine. Running the file through tsMuxeR somehow adds the required 'connection' (and corrects the header info so that my N.M.P. recognises the AC3 track). This is rather mysterious since tsMuxeR is nominally muxing a file that is already muxed! Editing this new file in TSPE loses the connection at the cut point.
Now I am sure that this has something to do with timecode and timestamps (whatever they are) but if somebody could explain what is going on to an old fart in language I can understand, then I'd be very grateful!

[Ventolin]

Hi dylmaster,

Both Video and Audio have a PTS - Presentation Time Stamp. This is the key factor in ensuring sync. Every Video frame should have a PTS but there can be many audio frames with 1 PTS.

A decoder's responsibility is to make sure a Video Frame is displayed at the same time as the corresponding audio frame is output.

Things would be easy if the decoder could decode 1 video frame and 1 audio frame and output them then move to the next frame. However, video is compressed using data from multiple frames. This means a group of frames (or pictures aka GOPs) are needed before the decoder can decode those frames and display them.

So a decoder has a buffer where it stores frames until enough frames are available to decode them and display them at the right time. Audio also has a buffer so the decoder's job now is to decode both the video and audio and store them in a buffer large enough to be able to display the video in sync with the audio.

Now consider if alot of frames are required before the decoder can display the first one - the audio needs a big enough buffer in the decoder so that the buffer doesn't fill up while waiting for the video to become ready. To help this situation somewhat, the audio is generally sent out of sync (offset) with the video - eg. a Video frame may arrive into the decoder with a PTS of 10.0 seconds followed by an audio frame with a PTS of 9.5 seconds.

The trouble comes when there are edits - the Presentation Timestamps (PTS) needs to be re-written so both audio and video remain in sync after the edit point. However, this can be tricky especially if the sync between different segments of the program changes. Furthermore, not all audio frames have PTS associated with them so the decoder may not be able to output audio until a PTS is received.

So here are a few explanations as to why you have desync:

1. The Decoder is not honouring the PTS of the Video and Audio for whatever reason.

2. The Audio sync is changing inbetween advert breaks and the decoder has too many or not enough audio frames so looses sync.

3. The PTS restamping is going wrong.

4. Something else is going on.

Without having a larger sample, I can't give you a precise answer as to what the exact issue is. I suspect it is option 1 though.

Remuxing or passing a transport stream through a muxer like TSMuxer basically repackages and restamps the video and audio. This can help the decoder honour the PTS of Video and Audio but it doesn't change the sync offset between the two. So if you cure desync by just remuxing, the Decoder you have is having a tough time honouring the timestamps. This really is a limitation of the Decoder in this case.

If remuxing does not cure the issue then something else is wrong and would need further analysis.

You can follow what is happening in your stream by looking at the PTS of both the Video and Audio and working out the difference between them. Use I frames as a reference point then move to the next audio frame by selecting "Frame Step - Audio" then using the F+ button and then use the PTS+ button to move to the next audio PTS.

I hope that clears a few things up a bit!

Regards,

Vent

[dylmaster]

Vent - that is very clear, thanks very much for that explanation.

It does prompt some obvious questions -
1) does TSPE have it's own H.264 decoder or does it use one of those already in the system?
2) does TSPE change the PTS after an edit, or is this what you mean by remuxing?

I have noticed whilst editing in TSPE that as I step through the video by repeatedly pressing the I+ button (next I frame) that the video doesn't always progress forwards on the timecode count. Indeed it is obvious by inspection that it is sometimes going back. It also sometimes gets the interlace reversed when playing at normal speed. Usually this is fixed by pausing and restarting. Any thoughts as to what these effects would be caused by?

thanks for your understanding and great support.

[dylmaster]

Actually - thinking on - it occurs to me that this particular programme is a film and therefore (in PAL land) is 'progressive'. The effect I sometimes see on the TSPE monitor window therefore cannot be a reversed interlace but is an actual incorrect playing order for the decoded frames.
FWIW

[Ventolin]

Vent - that is very clear, thanks very much for that explanation.
Hi dylmaster, glad it's helped!

It does prompt some obvious questions -
1) does TSPE have it's own H.264 decoder or does it use one of those already in the system?
2) does TSPE change the PTS after an edit, or is this what you mean by remuxing?
TSPE has partial decoders for all the codecs. For video preivew however, it uses the direct show filters already installed in your system. You can use a custom direct show filter chain however under the Options menu (Enable Direct Show Control and specify the filters under Direct Show Control).

TSPE restamps the PTS (and the DTS and PCR) by default when you make an edit. You can turn this off by unchecking the "Fix Timecodes" option next to the Edit button.

Some more theory:

There are 4 layers at work here:

1. Acutal Video Frames or Audio Frames
2. Elementary Stream Layer (Video & Audio etc)
3. PES Layer (Packetised Elementary Stream layer which holds the PTS timinig information). PSI (programme specific Information also comes here which contains the PAT and PMT (tables which say what streams make up a programme (PMT) and what programmes are contained in the stream (PAT)
4. Transport Stream Layer (which holds the PCR timing information)

Imagine each layer as a wrapper - starting with layer 1 being your actual content, layer 2 really sticks a bunch of Audio frames together into an Audio Frame (Video already is a frame but can be split into 2 fields). Layer 3 contains the magic PTS (and DTS) sync info and adds another wrapper and finally Layer 4 makes 188 byte or 192 byte "chunks" of all of the above layers which is the basic unit of a Transport stream and thus the basic unit of TSPE.

When you remux you unwrap all the layers back to the acutal Video and Audio frames (layer 2) and regenerate layer 3 and layer 4 by using the existing layer 3 and layer 4 as a base.

More to follow...

Vent

[Ventolin]

I have noticed whilst editing in TSPE that as I step through the video by repeatedly pressing the I+ button (next I frame) that the video doesn't always progress forwards on the timecode count. Indeed it is obvious by inspection that it is sometimes going back. It also sometimes gets the interlace reversed when playing at normal speed. Usually this is fixed by pausing and restarting. Any thoughts as to what these effects would be caused by?
This is more than likely a Direct Show filter issue but can be allieviated somewhat by experimenting with Direct Show Filter Control.

Note that as frames are stored out of order in the stream, when you Frame step (F+) you will notice the timecodes jump forwards and backwards depending on the type of frame but the Byte Position and Packet Count still go upwards. This is by design as you are navigating at the Transport Packet Layer (4) and not the 1st layer (1).

Now PAL and interlaced and movies:

Most movies are shot on Film and are called Progressive.
Most TV shows are shot on cameras and are called Interlaced (or Video).

In the days of old when CRT TVs were the norm, everything was broadcast as Interlaced. Since CRT TVs are interlaced by design everything matched together so you didn't notice.

Now with Flatscreens and High Definition etc. the Display end is mostly Progressive.

So for Film you have:

Film (progressive source) -> interlaced for broadcast -> Deinterlaced for Display
Video (interlaced source) -> interlaced for broadcast -> Deinterlaced for Display.

Now because in both situations you have an interlaced broadcast format, both Progressive frames and Video Interlaced frames are split into fields. This is where problems start.

Too often you have broadcasters inserting Logos or DOGs all over their content for channel branding (but will tell you its so you know what channel you are watching forgetting about EPGs etc). This can cause issues because the Logos can be pure interlaced. Add an interlaced logo to a progressive source means you now have Interlaced content which needs to be deinterlaced correctly to give you Progressive frames.

It's because of issues like this that sometimes the 2 fields that make up a progressive frame can get swapped or a field can get lost so a 1 frame is made up of 2 fields from different frames.

Now the Deinterlacer in your display needs to work out what fields go where when they are in the right order or when they are in the wrong order. Working out the correct order when things go wrong is down to the quality of the Deinterlacer you have.

A quality Deinterlacer will work out where the fields are and put them back in the right place. A less expensive deinterlacer might not check at all and just display each field as they arrive.

What's worse is that if the logo is interlaced you now have a pure interlaced source (the logo) which can not be made progressive in an exact way (like you can with progressive frames). So you end up with that horizontal "combing" effect you sometimes see.

Imaging what is required to convert a pure interlaced source into a progressive frame. In this case there are varied solutions ranging from making each field a frame (field interpolation) to scene analysis to work out which parts are moving (apply field interpolation) and which parts are static (weave both fields together only for the static areas).

You can quickly see that deinterlacing is a HUGE subject and I have just scratched the surface.

I hope that helps clear some more things up for you (and others) but I know all of this stuff is pretty technical!

thanks for your understanding and great support.

You are welcome!

Just in case you didn't know, there is some info on the Bitstreamtools Website (http://www.bitstreamtools.com) about all this, but I still have plenty more information to add there! I should be cutting and pasting from here to there and vice-versa!

Regards,

Vent

[dylmaster]

Just in case you didn't know, there is some info on the Bitstreamtools Website (http://www.bitstreamtools.com) about all this, but I still have plenty more information to add there! I should be cutting and pasting from here to there and vice-versa!
Vent

Great idea - this is all gold dust to my mind!

I have now purchased the CoreAVC codec and it has made a big difference to the sync in MPC. I also have most of the usual (free) suspects but my previous (trial) of CoreAVC had expired so I'm not sure what I was actually using. I'm going to systematically work through all the options (that's a lot!) and set everything correctly. That may well need some more education (there are lots of 'how to' web postings out there!) so I may well be some time!

[dylmaster]

Note that as frames are stored out of order in the stream, when you Frame step (F+) you will notice the timecodes jump forwards and backwards depending on the type of frame but the Byte Position and Packet Count still go upwards. This is by design as you are navigating at the Transport Packet Layer (4) and not the 1st layer (1).
Vent

Ah yes, I had forgotten this. So in order to make an edit in TSPE with least effect on the sync, how would I actually choose the points? Clearly I should always start a wanted clip on an I frame, but how do I finish it on the last frame in the GOP? I had been going to the nearest I frame (from my desired out point) and then stepping back one frame, but this ignores the jumbled frame ordering in the GOP (or does it, now I'm confused!).

[Ventolin]

Ah yes, I had forgotten this. So in order to make an edit in TSPE with least effect on the sync, how would I actually choose the points? Clearly I should always start a wanted clip on an I frame, but how do I finish it on the last frame in the GOP? I had been going to the nearest I frame (from my desired out point) and then stepping back one frame, but this ignores the jumbled frame ordering in the GOP (or does it, now I'm confused!).
Hi dylmaster,

CoreAVC works very well indeed with TSPE, better than FFDShow, but the latest FFDShow displays to P frame accuracy so is also useful.

Start points should be I frames, end points can be I or P frames. The "Clean Edits" option works out where the nearest I or P frames are for both start and end points and removes any unwanted B frames at the edit points. I would always advise setting the start/end points manually and then testing each edit point by selecting two rows in the EDL, right clicking and choosing Test Edit Point.

Next version with timelines will make this alot easier!

Most of the options in TSPE are really there to disable some part of the AI in case there is an issue with it. As time goes on though, TSPE is becomming more robust and so the options aren't really needed anymore.

Regards,

Vent

[joerg]

Tentative date for 0.80 end of 2nd Week of June, but don't hold me to that!

I'm sorry if I seem impatient, but I can't keep myself from asking: any news about that release, which was originally planned for May (IIRC) and then 2nd week of June?
I have been holding off editing most of my HD caps, hoping it will get easier with the new version of TSPE with fixed bugs and the timelines functionality.
Of course I know that it's hard to predict a release date, especially if an application is developed in one's limited spare time.

[Ventolin]

I'm sorry if I seem impatient, but I can't keep myself from asking: any news about that release, which was originally planned for May (IIRC) and then 2nd week of June?
I have been holding off editing most of my HD caps, hoping it will get easier with the new version of TSPE with fixed bugs and the timelines functionality.
Of course I know that it's hard to predict a release date, especially if an application is developed in one's limited spare time.

Hi guys,

Sorry I have not posted earler, I am very busy with quite a few things! (IRL, not related to TSPE).

Next version is still "almost ready" the bulk of the code is working but needs some refinement. I also have a big bug to fix which is proving to be quite tricky to resolve.

Comskip / Showanalyser import ability is already coded and working, the timelines also are working but are not so pretty and not too useful at the moment, but defintely is an improvement over the text based EDL.

TSPE will be heading towards full multi-track editing but this will be ready for 0.90, with 1.0 having frame accurate editing.

Hope that helps, please do keep checking here for the latest news and the website.

Regards,

Vent

[JonRead]

Hey Ventolin, sorry to ask as I know you are probably busy, but any news on how things are progressing now to the next beta ?

[Ventolin]

Hey Ventolin, sorry to ask as I know you are probably busy, but any news on how things are progressing now to the next beta ?

Hi,

Yeah, been very busy lately, here's what's left to do:

0.80:
Misc updates and bug fixes - done
Internal Demux engine - done
Comskip / showanalyser EDL import - done
Timelines (stage 1) - done
Bug1 in output - working on
Bug2 in output - todo
Tidyup & polish - todo

Once I get those 2 bugs sorted, just needs a little bit of polish and 0.80 be ready.

The 2 bugs are big ones though and are taking alot of time trying to fix.

After that will be:
0.90 - Muxing engine, timelines stage 2
1.00 - Frame accurate

2 versions will be available - Professional and Standard (limited analysis), the betas at the moment are all for Professional.

Sorry for the continued delay folks, but I hope you will enjoy the upcoming features!

Regards,

Vent

[NotAScooby]

Vent,
First off, congratulations on creating a tool (TSPE) that succeeds where many, many commercial packages fail. For the task of quickly editing an HD TS stream while keeping sound and video in sync, I've yet to find anything close, let alone better.

I've been trialling your current test version 0.755, have read this entire thread and many other references while I've been doing my trial, and I'm ready to purchase it from you, but there is one big concern stopping me:

Since your release of 0.755 in March you've been very busy with other things and your posts to this thread have been very few. This worries me, as there are clearly bugs in 0.755 and missing features that you've been promising for coming releases since June. So, I have to ask myself, is 0.755 stable and useable enough for me to buy it at the current price if, in the worst case, your development has stopped and there are not going to be any more releases?

Sadly, my answer to that is no.

The main bugs that are causing me problems are:

- If I open a recorded HD stream (say BBC HD), and do a simple edit (Set a start point, set an end point, click the edit button) the written file is created perfectly. If I then change the name of the 'save' file and move the slider, or delete the EDL entry to cut a different section, the program usually fails with a microsoft 'TS packet editor has encountered a problem and will close' error (usually something like Appver0.755.3526.31214 modname quartz.dll offset 000d0d7e). This is annoying rather than critical, and there's an easy work-around by simply re-opening the file, but I often need to split one recording into several parts so it's something I've hit a lot.

- when moving around in a file using the jump buttons, I often hit a memory exception 'this memory area cannot be written', and again the program is closed. This is more annoying as it's unpredictable and unreproducable and often happens after I've added several sets of edits to the EDL.

- If I try to use the > and < buttons or IF+/IF- to select an edit point, then most of the time the image displayed does not change, so I am 'in the dark' about where the cut point is being set.

- More often than not, the popup test window has no sound, or fails to open. I realise that this is probably a problem with the configuration of my sound / video drivers, and I haven’t spent much time trying to resolve it.

There are other problems, but these are the ones giving me most grief at the moment. Similarly, there are lots of things that I would like to ask you to add, but these are 'nice to have' rather than essential to the operation of the software. Now, I can understand from what I've read on this thread that all my problems are probably due to my configuration or the setup of my Filters, but if I cannot rely on the rapid response to teething problems that you were giving earlier in the year, then I'll be 'on my own' and without any source code or detailed documentation I'll be struggling with trial and error, and I'm not prepared to spend tens of euros to be left in that position.

I've installed the latest haali splitter (which made a huge improvement), and followed your other advice in this thread, but am left with the worrying suspicion that your development has stopped, and 0.755 is as far as this software is going to go. With no expectation of any additional functionality, or of any bugs being fixed, TSPE is not as appealing as it was.

To stop at 0.755 would be a loss for everyone who has supported you to date, and the many people like me who are ready to support you in the future, but I realise that life moves on and none of us can predict what other demands there will be on our time, or what other opportunities come knocking.

So, to quote your reply in message 104 when you were asked "If author goes out of business can I move licence", you said "It'll be "Acts of God" that will take the authors "out of business" and even then I'm sure TSPE will still live on!". I must ask - has there been an "act of god"? are you still "in business"? because if not then with a hardware locked key; if your development work is ended, then this software would become useless next time I upgrade my PC, and that's a real limit that I think will stop anyone from purchasing.

I hope you can come back with some reassurance, I really would like to have a copy of TSPE, fully supported with a planned upgrade schedule, and I would be happy to pay your asking price to be in that position. I can imagine how many hundreds of hours you have already invested in this development, and I would much rather pay you than try to start a similar development from scratch. I can only see the demand for your software growing with the proliferation of Media PCs, so I am sure your potential market is increasing rather than decreasing, but if development must stop, so be it.

I hope you're still checking this thread, and hope to hear from you.

Regards,

Dave

[Ventolin]

Hi Dave,

Thank you very much for your post and for your kind words!

You'll be pleased to know that development has been continuing and a new release is just days away!

If you check the download page of the website you can see that 0.80 is in final testing. Some full licence holders are in a closed beta test for 0.80 and the final bugs are being sorted out as we speak.

I have just 1 big but minor bug to fix before I can release the next version, but I hope to have that fixed in the next few days. Barring any other "show stoppers" (there's been one already) the release should go ahead.

Regarding 0.755, (and for 0.80) it sounds like you may have a directshow filter issue. To quickly see if this is the case, just turn off the Video checkbox and try and edit your file. Just go through the motions and see if TSPE crashes. If it doesn't then the directshow filter setup is the cause.

As you may have read, best results are obtained with Haali and CoreAVC or FFDShow. Make sure these filters are in use by checking your system tray icons and right clicking the video window. You can force CoreAVC or FFDShow to be the h264 decoder by enabling DirectShow control and selecting the filter in the filter selection dialog window.

I hope that fixes your issues and answers your questions for now, no doubt you will have more when the new version is released! I'll make an announcement here when it's ready, though you can also check the website.

If you have any specific questions feel free to PM me.

Regards,

Vent

[medp7060]

I tried 0.755, it is not attractive at all. I hope the coming 0.8 will be "frame acurate", otherwise, it cannot stand out among other similar unitlities.

BTW, a commercial one is Videoredo, which is in a alpha test now. It may be the one everyone is expecting.

[joerg]

I bought the software in May, when it looked like development was really progressing. I had some reservations about buying a beta at the price that was asked and about the hardware lock, but was convinced by the fast and helpful forum replys from Ventolin.
I hoped to be able to cut my HDTV caps properly during summer time with that new version which was initially announced to be available around the 2nd week of june. It never came.
Always if someone asks the new version "will be out soon" - the last answer stated the last bugs should be fixed "in a few days" - and again it's been three weeks. This is starting to feel like a "Duke Nukem Whenever" which is quite frustrating. I am starting to wonder if I made a mistake buying the beta so early.
In the meantime I have tried TSDoctor (which is free in it's beta stage) which allows to clean, convert to ts and cut my m2ts files in one go. It has had lots of updates during the last few months. Unfortunately it also doesn't really support accurate cutting.
@medp7060: Thanks for the hint about VideoRedo. I'll have a look at it once it's released with h264 support.

[Ventolin]

Hi Guys,

@medp7060
Sorry, 0.80 isn't frame accurate. The current plan is to have frame accurate by version 1.00. 0.90 will have a new muxing engine which is a prerequisite for frame accurate editing.

I am unaware of any other applications that are similar to TSPE offering editing and analysis of any transport stream regardless of content inside! If there is do let me know!

@joerg
I have what I hope to be the public release version of TSPE 0.80 ready to go. I'm in the process of updating the website as we speak. If you can't wait then send me an email and I'll send you the pre-release version.

My apologies for the continued delay but hope to release 0.80 today or tomorrow.

Regards,

Vent

[Ventolin]

Hi folks,

Well, after days and days of "almost ready" (more like months ahem!), the next version,

TSPE 0.810 Professional is finally out!

You can download it here:

http://www.bitstreamtools.com/download

This version is fresh out of the closed beta test so I've labeled it as "Unstable Development Alpha Release" mostly because there are probably bugs which haven't been caught yet and need to be fixed. Updates will be added as and when bugs are fixed and new features added.

If you have a full licence you will need to regenerate your new licence details with the new version. A temp licence is inlcuded with the TSPE setup program which is used if no existing licence is found in the installation folder. The temp licence gives 30 days unrestricted evaluation.

A list of new features are given if you follow the "What's New" link, but here's a brief summary:


Windows 7 support
GUI Update
Timelines - PIDLine, PESLine, Timeline
Comskip / ShowAnalyser import
Application Setup and Uninstall
Update checking
Enhancements and Bug Fixes!


So, what's next?
0.90 - New Muxing engine, independant PID editing
1.00 - Frame accurate

No timeframes for the above I'm afraid, but I'll try and keep a stable beta and unstable alpha available so you can try out the newest versions.

Thanks to all for your support and feedback, of course you are always welcome to give some more!

Regards,

Vent

[joerg]

Great to see the new version is finally out! I have just downloaded and installed it but will probably not be able to really test it before the weekend.
You write "If you have a full licence you will need to regenerate your new licence details with the new version.". After installing and starting TSPE 0.810 it tells me it is "Licensed to Joerg" - so to me everything seems fine with regard to the license. Do I still need to re-generate and mail in my activation key?

[canTsTop]

i did some tests with new version. here is sample http://www.mediafire.com/?mjm55izyaww and edl http://pastebin.com/f785da6b5

i open video in tspe, scan for errors, its clean. when cut some parts of video and save it. open edited video in tspe, scan for errors and there is:
Continuity Errors: 0, Transport Error Indicator Errors: 0
Timestamp gaps: 0, PCR gaps: 3
Total number of separate discontinuous regions: 3

edited video plays until first cut in vlc and mpc-hc with internal dxva decoder. mpc-hc with internal ffmpeg decoder plays all video, but there is some corruption in video near cuts. vlc version is 1.0.3, mpc-hc 1.3.1400.0

[medp7060]

select 2min, but only output 2.4 s.

[Ventolin]

i did some tests with new version. here is sample http://www.mediafire.com/?mjm55izyaww and edl http://pastebin.com/f785da6b5

i open video in tspe, scan for errors, its clean. when cut some parts of video and save it. open edited video in tspe, scan for errors and there is:
Continuity Errors: 0, Transport Error Indicator Errors: 0
Timestamp gaps: 0, PCR gaps: 3
Total number of separate discontinuous regions: 3

edited video plays until first cut in vlc and mpc-hc with internal dxva decoder. mpc-hc with internal ffmpeg decoder plays all video, but there is some corruption in video near cuts. vlc version is 1.0.3, mpc-hc 1.3.1400.0

Hi canTsTop,

Thanks for your sample and edl.

Looks like there might be a bug in the PCR fixup code. I tested it on 0.755 and it was ok so will hopefully be able to fix it.

@medp7060
Can you please give some more information? A sample and edl as canTsTop did would be very helpful.

Thanks guys.

Regards,

Vent

[prenz]

Hi Ventolino,

I have by accident found a sure way to crash TSPE (0.810):
- import any sufficiently large .ts
- scan it
- while scanning, move the time slider
- now TSPE will log a whole bunch of bogus inconsistencies, until it finally crashes after a minute or so.

[Ventolin]

Hi Ventolino,

I have by accident found a sure way to crash TSPE (0.810):
- import any sufficiently large .ts
- scan it
- while scanning, move the time slider
- now TSPE will log a whole bunch of bogus inconsistencies, until it finally crashes after a minute or so.

Hi prenz,

Thanks for that bug report (oops!) Will fix it for the next release.

Regards,

Vent

[joerg]

Hi Ventolin,

did you overlook my posting?:
You write "If you have a full licence you will need to regenerate your new licence details with the new version.". After installing and starting TSPE 0.810 it tells me it is "Licensed to Joerg" - so to me everything seems fine with regard to the license. Do I still need to re-generate and mail in my activation key?

I would still like to know if there's any reason for me to re-generate and mail in the activation key. Would be great to get a short answer.

Regards
Joerg

[Ventolin]

Hi joerg,

Sorry I didn't reply earlier, but I was expecting you to find that the old licence stopped working. I'm a bit surprised it is still working though!

No need to regenerate it if it's working fine, if it does fail then just resend your activation details as before.

Regards,

Vent

[joerg]

The audio streams (in the original m2ts stream as well as in the remuxed and fixed stream) are displayed as being 64 kbit and 44100Hz:
PID: 0x184D: MPEG1 Audio Layer 2, 64Kbit/s @ 44100Hz
PID: 0x184E: MPEG1 Audio Layer 2, 64Kbit/s @ 44100Hz
In fact the streams are 256 Kbit/s @ 48000Hz.
So it looks like they are being mis-detected (doesn't look like this has any negative effect as far as I can tell, but I thought I'd mention it).


Yes, this does look like a minor bug, it will only affect the display side of things and will not affect the output. Hopefully will have this fixed soon.


It seems like this bug is still present in version 0.810.


Also I noticed that the new PES Line mode does not affect the way the "Jump back a frame" and "Junp forward a frame" buttons work - i.e. they always seem to work in the physical order of the frames, even while I'm viewing the PES line. Is this expected?

And I have a suggestion: would it be possible to allow advancing through the video frame by frame in the TSPE Internal Viewer? Currenty I can just press play. Such a feature would make it easier to really properly test edit points.

[joerg]

Sorry I didn't reply earlier, but I was expecting you to find that the old licence stopped working. I'm a bit surprised it is still working though!

No need to regenerate it if it's working fine, if it does fail then just resend your activation details as before.

:thanks: That's what I wanted to know: will it fail visibly and let me know that it stopped working or can there be some side-effects that I won't notice. I guess your reply answers that question and I will get a clear message if it fails.

btw: I even exchanged my main harddisk in the meantime (my "old" WD Raptor broke (in the middle of the day while I was working with the PC) so I had to get a new disk and restore the system from a (much too old) backup) and the license kept working fine.

[joerg]

Hi Joerg,
1. With this sample you do need a preview offset of +3. The preview offset is dependant on the stream though which is why it is not remembered when you restart TSPE. I will have a look at ways to improve this situation.

Have you been able to make any changes for this with the new version? I tested this again and I am still not seeing the correct frame. I am using the latest CoreAVC (2.0.0) and Haali (1.9.355.21). For me this is a huge issue since I need to be able to see the correct frame in order to make proper cuts. Using the "test cut" feature is a workaround, but it is rather clunky. If the offset is dependent on the stream (as you said), I cannot even figure out the correct offset once and be done with it. :( Is there anything that can be done to resolve this or will we have to live with this?


2. When you set an end point, what you are actually setting the exact point at which to stop the edit. So if you are setting an I frame then everything upto that I frame is inlcuded but not the I frame itself. You would need to set the end point to the next frame in the PTS order.

This I agree is a bit confusing so I will have a look at changing this so the current frame you are looking at is also included.

It seems like this has not changed in the new version. This is a bit unfortunate since it makes making proper cuts very difficult - you never know where exactly to set the cut point if you want a specific frame included (due to the added difficulty of frame ordering within the stream).


[still 2.]However, if you do move past the I frame to the next B frame in PTS order, the resulting output is missing the last B frame before the target I frame (in PTS order) which is incorrect. This needs fixing.


It seems that this bug is still present, which is quite unfortunate since it makes clean cuts more or less impossible. After doing a cut I had a look at the resulting stream and found the following:

[...]
52:19.98 (I)
52:19.94 (I,P,B)
52:19.96 (I,P,B)
52:20.04 (I)
52:20.00 (I,P,B)
[end of stream]

Obviously the frame for 52:20:02 (probably an I,P,B frame) is missing.


3. M2TS vs TS - the sample has shown a few bugs which I need to fix before I can do any further analysis. Thanks for the sample again!

Also for this issue I would like to know if it has been fixed with the latest beta and clean edit now works with m2ts or if I need to convert the m2ts to ts first before doing my cut (I would obviously like to avoid that step). My first tests seem to suggest that this is fixed.


5. Jump time inaccuracy - this is a bug that I've known about but haven't got around to fixing yet.

This also seems to be unfixed.

I also found another bug, I believe. Move somewhere close to the end of the stream. Click the +10 seconds button (where no 10 seconds are left). The stream will jump to the last frame, which is fine. Click the "move back a frame" button one or more times. Now click the "move forward a frame" button. TSPE will tell you "No More Frames Found." and will refuse to move forward again.

The most important of these issues (at least for me) are the issue 1.) and the two issues listed at 2.) since they are essential for making proper cuts. I do not necessarily need true frame accurate editing (with re-encoding small parts (e.g. when I,P or I,P,B-frames are set as cut-in) of the video) - if I know I can set the cut-in at an I-frame and the cut-out at I or I,P frame that's good enough. But if I can't properly identify (see) the frames I actually cut at, this makes editing really difficult.

[Ventolin]

Hi Joerg,

Thank you for your excellent feedback! There's quite a bit, so I'll try and answer what I can now and look into the others later. Easy post first...

[Audio identification bug]
It seems like this bug is still present in version 0.810.
Yes, it's still there, I haven't fixed that bug yet, thanks for the reminder :)

Also I noticed that the new PES Line mode does not affect the way the "Jump back a frame" and "Junp forward a frame" buttons work - i.e. they always seem to work in the physical order of the frames, even while I'm viewing the PES line. Is this expected?
Try clicking the frames in the PES line! (Have a look at the What's New page on the website).

And I have a suggestion: would it be possible to allow advancing through the video frame by frame in the TSPE Internal Viewer? Currenty I can just press play. Such a feature would make it easier to really properly test edit points.
Try the mouse wheel after pausing! (This is listed in the Keyboard Shortcuts under the Help menu)

More in next post...

Regards,

Vent

[Ventolin]

Have you been able to make any changes for this with the new version? I tested this again and I am still not seeing the correct frame. I am using the latest CoreAVC (2.0.0) and Haali (1.9.355.21). For me this is a huge issue since I need to be able to see the correct frame in order to make proper cuts. Using the "test cut" feature is a workaround, but it is rather clunky. If the offset is dependent on the stream (as you said), I cannot even figure out the correct offset once and be done with it. :( Is there anything that can be done to resolve this or will we have to live with this?
For the samples you sent me, the offset is +3. I don't know why this is the case, but you are the only person to have reported issues with the accuracy - in all other streams it's frame accurate. This seems to be an issue with either Haali or CoreAVC but fixing this in TSPE will probably need internal filters to be written which might come in for version 2. It might be all your streams require the same offset - let me know if this is the case and I'll make the offset setting persistant.

2. When you set an end point, what you are actually setting the exact point at which to stop the edit. So if you are setting an I frame then everything upto that I frame is inlcuded but not the I frame itself. You would need to set the end point to the next frame in the PTS order.

This I agree is a bit confusing so I will have a look at changing this so the current frame you are looking at is also included.
It seems that this bug is still present, which is quite unfortunate since it makes clean cuts more or less impossible.
This isn't really a bug, as it performs as intended. Just remember that the end frame should be the first frame that you don't want. As I mentioned before it is a little confusing, but certainly possible. If the end point is going to be joined by another segment, then the end frame will be the last I or P frame if Clean Edits is on, so the actual end frame is determined by the stream frame order in this case.

After doing a cut I had a look at the resulting stream and found the following:

[...]
52:19.98 (I)
52:19.94 (I,P,B)
52:19.96 (I,P,B)
52:20.04 (I)
52:20.00 (I,P,B)
[end of stream]

Obviously the frame for 52:20:02 (probably an I,P,B frame) is missing.
In this sequence, the last displayable frame looks like 52:19.98(I) as the 2nd I frame (52:20.04) might not be complete. It's not as clean as it should be though, the last I and B frame shouldn't be there. I will probably have a go at cleaning this up for the next version, 0.90.


3. M2TS vs TS - the sample has shown a few bugs which I need to fix before I can do any further analysis. Thanks for the sample again!Also for this issue I would like to know if it has been fixed with the latest beta and clean edit now works with m2ts or if I need to convert the m2ts to ts first before doing my cut (I would obviously like to avoid that step). My first tests seem to suggest that this is fixed.
Yup, I hope I fixed that!

5. Jump time inaccuracy - this is a bug that I've known about but haven't got around to fixing yet.
This also seems to be unfixed.
Yeah, not quite fixed but alot better than it used to be. It gets it right mostly, but sometimes it's slightly out. Not too big an issue though as it's just for rough positioning anyway, but it can be fixed.

I also found another bug, I believe. Move somewhere close to the end of the stream. Click the +10 seconds button (where no 10 seconds are left). The stream will jump to the last frame, which is fine. Click the "move back a frame" button one or more times. Now click the "move forward a frame" button. TSPE will tell you "No More Frames Found." and will refuse to move forward again.
Nice bug! Looks like the +10 seconds goes over the end of the file, so seeking back frames fails - look at the byte position and compare it to the end byte position!

The most important of these issues (at least for me) are the issue 1.) and the two issues listed at 2.) since they are essential for making proper cuts. I do not necessarily need true frame accurate editing (with re-encoding small parts (e.g. when I,P or I,P,B-frames are set as cut-in) of the video) - if I know I can set the cut-in at an I-frame and the cut-out at I or I,P frame that's good enough. But if I can't properly identify (see) the frames I actually cut at, this makes editing really difficult.
Yeah, I agree it's sometimes hard to get the exact start / end frames matching with the actual frames in the edit, but for most streams this is the case. However, some streams such as yours are a bit more troublesome. That's why the Test edit point functions and Test Start / End point buttons are there along with fine tuning the frame displayed in the video window. They are compromises because getting it frame accurate for every transport stream with any content is pretty hard and using 3rd party directshow filters means some of that accuracy is out of my control.

However, I hope with the Timeline view and overlap detection finding the right edit points is alot easier with the new version.

To recap, to fine tune the edit point between two segments A and B:

First get 2 segments set:

Set the start & end point of segment A.
Click + to add to the timeline.
Set the start * end point of segment B.
Click + to add to the timeline.
Double click inbetween segment A and segment B to make a "Test Edit Point".

To fine tune:

Double click segment A or segment B to load that segment into the main viewer.
Change the start / end point.
Click + to add to the timeline. TSPE will auto detect if there's an overlap and change the relevant segment automatically.
Double click inbetween segment A and B to test again.


Remember you can leave the preview window open along side the main TSPE gui so it's doesn't take too long to fine tune the edit point.

Hope that helps, thanks again for your excellent feedback. I'll drop you an email about this also.

Regards,

Vent

[joerg]

This isn't really a bug, as it performs as intended. Just remember that the end frame should be the first frame that you don't want. As I mentioned before it is a little confusing, but certainly possible. If the end point is going to be joined by another segment, then the end frame will be the last I or P frame if Clean Edits is on, so the actual end frame is determined by the stream frame order in this case.

OK, to help me understand, let's assume I have the following:

01. 57:53.38 (I)
02. 57:53.34 (I,P,B)
03. 57:53.36 (I,P,B)
04. 57:53.44 (I)
05. 57:53.40 (I,P,B)
06. 57:53.42 (I,P,B)
07. 57:53.50 (I,P) <-- last visible frame that I want included
08. 57:53.46 (I,P,B)
09. 57:53.48 (I,P,B) <-- last frame that should be in the resulting stream according to the timecodes
10. 57:53.56 (I,P)
11. 57:53.52 (I,P,B)
12. 57:53.54 (I,P,B)
13. 57:53.62 (I,P)
14. 57:53.58 (I,P,B)
15. 57:53.60 (I,P,B)
16. 57:53.68 (I)
17. 57:53.64 (I,P,B)
18. 57:53.66 (I,P,B)

Where would I need to set the cut-out point if I want 07. to be the last visible frame in the resulting stream?

a) From a user standpoint (using physical frame order) considering what you said I'd go (from 07.) to the next frame (08.) using the next frame button and set that as cut-out (which is probably incorrect.)
b) From a user standpoint (using PTS frame order) considering what you said I'd go (from 07.) to the next frame in time order (11.) using the PES line feature (which is probably incorrect.)
c) From a logical standpoint I'd guess I have to select 10. as the cut-out-frame.

As you can see it is quite confusing if it works the way it is currently implemented.


In this sequence, the last displayable frame looks like 52:19.98(I) as the 2nd I frame (52:20.04) might not be complete. It's not as clean as it should be though, the last I and B frame shouldn't be there. I will probably have a go at cleaning this up for the next version, 0.90.

Actually from what I set as cut-out the 2nd I-frame SHOULD be there. If you look at the example I gave above, 07. is the frame I selected as cut-out. So considering what you said 06. should be the last (physical) frame included in the resulting stream. In my case 05. is the last frame included. Now you say 04. and 05. shouldn't be in the resulting stream while actually they should and 06. (which is missing) ahould be added!


Nice bug! Looks like the +10 seconds goes over the end of the file, so seeking back frames fails - look at the byte position and compare it to the end byte position!


Indeed it seems to go over the end of the file. Interestingly enough seeking back still works (still showing incorrect current byte position, but moving the picture and "Time" value). Only seeking forward frames again fails.


To recap, to fine tune the edit point between two segments A and B:


In my use case I (fortunately) have to cut only at the beginning and end so there is only one segment in the resulting list as there are no ads that I need to cut out. I just want these cuts at the beginning and end to be as exact as possible (call me pedantic if you want ;) )

[satnews]

Stable Release
Please note this version does not support Windows 7 for full licences and will not edit .m2ts files for temp licences.
Download both the beta version and trial Licence file.

1. TSPE 0.755 Main Application
2. TSPE 0.755 (17/03/09) 30 Day Trial Licence File

I'm using TSPE 0.755. Thank for great trial testing...
But i get big problem ;(
cant save my cuts, when i press [Edit] - nothing to do, just an error on log-window

Error:
Index was outside the bounds of the array.
PID: 0, iPii: 0, ProgIndex: 0
Error: Edit failed, only 376 bytes written.


I do not have that error on 0.810 with same ts-files at same time.

Like TSPE 0.755 cause it works more accurately for my side.

thank

p/s Windows XP Pro x64

[dylmaster]

Very quick test of 0.81 on ITV HD, which previously didn't work. It now successfully makes cuts that hold lip sync (which is a BIG improvement for me since previous problems were a deal buster), but there is an apparent faulty GOP generated around the cut resulting in a still frame or glitch for about half a second on replay. This is after setting both in and out points AFTER pressing the previous/next I frame button. At least these buttons now appear to work correctly (previously they didn't go to the same frames).
In general, the preview (using the latest version of CoreAVC) seems to be more touchy - yanking the slider to a different part of the file results in severe blockiness that clears after a few seconds of play or 'jogging'. I wish it was possible to lock the preview display to half size and thus avoid all the interlace artifacts. It should also be easy to force it to the correct aspect ratio. I hope these points can be considered for future versions but at least I can now consider using it again and finding some of the bugs!
Thanks for your effors Vent - however late!

[Cela]

Hi Vent,
since New Year 2010 (Y2k10) TSPE 0.755 gets lost after entering 'Preview' via 'Set Test' with error messages in the log pane:
Error: Quick-parsing packet header
Der Index war außerhalb des Arraybereichs.
"... Index outside array..."

This error occurs since Jan 1, 2010 with all my newly recorded TS files on my Windows XP machine. I just confirmed this also to happen with old TS files by testing an ArteHD TS file recorded in Nov 2009 and succussfully processed (trimmed) with TSPE 0.755 in Dec 2009.

On my secod computer (Vista) I get the following error message with a recording from Servus TV HD from Dec 27, 2009
Error:
Der Index war außerhalb des Arraybereichs.
PID: 0, iPii: 0, ProgIndex: 0
Error: Edit failed, only 376 bytes written.

In addition there may also be a data problem with recordings from newly launched german language DVB-S2 services at Astra 19.2° E:
ORF1 HD, ORF2 HD, Servus TV HD, Das Erste HD, ZDF HD, etc.

Edit:
****
I guess it may be a license problem although the TSPE main window starts with correct "...Licensed to:..."
info.

I just found that there are the following 5 positions (^) changed in the Activation Key when I rerunTSPE Activation:
-----------------^-------------------^^
-------------------------------------^^
(no hardware or operating system change in the last few months!)

What should be done?
Renewing the TSPE 0.755 license and/or upgrade to TSPE 0.8 or both?

Thanks

[joerg]

I am seeing exactly the same problem. 0.810 works, but since clean cut doesn't work properly with 0.810 anymore I don't want to use that version.
Is there any workaround other that re-setting the clock to 2009 (I confirmed that this resolved the problem)?

[Cela]

Now with TSPE Pro 0.810 Licensed to Beta Test:
+ All older TS work
- Problem with TS from new ORF2 HD:
Set Start and End Point works.
Add to EDL works.
Preview Start/End point does NOT work.
Edit the item in the EDL does NOT work.

Error messages:
Error: Calculating Timestamp Framerate
StartIndex darf nicht kleiner als Null sein. [~ startindex < 0 not allowed]
Parametername: startIndex
...
Testing Start Point
Error: Testing Start Point
Ein leerer Pfad ist unzulässig. [~ empty path not allowed]
...
Changing search PID to match Frame Search Type
Changing search PID to match Frame Search Type
Copying Data...
Processing EDL Item: 1 of 1
Warning: No PCRPID Found, Please update the PMT with the PCR PID in the Table Editor <<<how to do the update?>>>
Error: PID tree failed validation checks

First 26 seconds (40 MB) cut with DVBViewer's TSPLayer from a 2+hours DVBViewer capture from ORF2 HD from Jan 1, 2010 can be found in
http://rapidshare.com/files/329270384/ORF2HD_test.ts

Hope you can fix this.

Regards

P.S.
ProjectX reports
-> Filetype is TS (generic PES Container)
-> demux
-> Service ID 0x1330
-> PMT 0x6C refers to these usable streams:
Video:
PID: 0xB68(#3)(H.264)
Audio:
PID: 0xB69(ger)(#4)
PID: 0xB6B(AC-3)(ger)(#7)[PD]
PID: 0xB6A(ENG)(#6)
Teletext:
n/a
Subpict.:
n/a
and demuxes the audio streams 'adjusting audio at its own timeline'.

tsMuxeR GUI 1.10.6 only recognizes the h.264 stream but none of the audio streams (maybe a problem related to the same cause?). Thus does not report the audio delay values which would apply to the audio streams demuxed by ProjectX. Thus no workaround until TSPE gets fixed.

[rebkell]

I am seeing exactly the same problem. 0.810 works, but since clean cut doesn't work properly with 0.810 anymore I don't want to use that version.
Is there any workaround other that re-setting the clock to 2009 (I confirmed that this resolved the problem)?

I hope so, I'm currently using .755, haven't even downloaded the newer version. .755 has been doing everything I need and I have no need for anything else. I have my routine down pat and it's been working without a hitch for many many months.

The key problem is really the only problem I've had with TSPE since way way back.

[HD65]

I have another issue, not Y2K10 related. I just bought an AVCHD camcorder and a HD Mediaplayer. I needed an AVCHD editor, and TSPE seems one of the few editing the PGS stream as well. This will keep the correct timestamp in the video, which is important to me. Most other editors ignore the PGS stream and don't include it in the output at all.

But I think there is a bug in TSPE's PGS output stream when there are two (or more) edits. It looks as if the "gap" which is fixed for video (see output below) is not applied to the PGS stream. As a result, the last subtitle of the first cut stays on the screen for the duration of the gap between the two cuts. You can also see this problem when you demux the output file and process the SUP stream in SupRip.

Processing EDL Item: 1 of 2
New PAT Created.
New PMT Created.
Processing EDL Item: 2 of 2
New PAT Created.
New PMT Created.
Video PTS gap of 10,5 seconds fixed at 00:00:10 in output file, 0x92E7100.
Section duration is 00:00:20 (555 video frames)
Average bitrate is 16,96 Mbps
Total time taken 2,36 seconds, write speed 19,3 MB/s
Edit complete: successfully written file with 43929984 bytes.
Creating Custom Filter Graph...
Video Loaded.

If I create two separate files of the same edits and then join them with tsmuxer there is no problem.

[Ventolin]

Hi Guys,

Sorry for the Y2K10 issue and for the delay in replying, I hope you can understand sometimes I am not able to respond immediately.

I have put a fix up on the website, TSPE 0.760. I hope that version works as 0.755 did.

I took the opportunity to update it to support Windows 7 and it has an online version check.

I haven't had time to test this version thoroughly though, so please let me know if there are any issues.

HD65 - please bear with me, this issue will need to be addressed in 0.810+ and might take a while to fix. Can you please provide a sample and short explanation on how to reproduce the error?

Cela, I will try and have a look at why your streams are failing but might take a while also.


Regards,

Vent

[kirkusinnc]

Getting a zero byte file in Firefox and an error in IE when I attempt to download the new version...

Kirk

[Ventolin]

Getting a zero byte file in Firefox and an error in IE when I attempt to download the new version...

Kirk
Hi Kirk,

I just tested it with IE, Opera and Firefox and could download it fine. What error message did you get with IE?

Regards,

Vent

[HD65]

...

HD65 - please bear with me, this issue will need to be addressed in 0.810+ and might take a while to fix. Can you please provide a sample and short explanation on how to reproduce the error?

...


Thanks Vent. I'll work on a small sample. Here's how to reproduce the issue:

- Take a film clip with three streams: video, audio and subtitle (e.g. from a HD camcorder). Say the duration is 4 seconds, with one subtitle per second.
- Use TSPE and place two edits in the EDL. Say the first and last second.
- Hit the edit button (v 0.810) so a new clip is created. The duration of this new clip is 2 seconds.

If you demux this new clip and use SupRip to make an srt file of the PGS stream, you'll see that SupRip thinks that the first subtitle lasts three seconds. When you play this clip in a player supporting subtitles, you'll never see the second subtitle.

[Cela]

I just found a couple of posts (in German language, let Google translate to English) which claim that in the context of certain DVB-S2 CBR services (ORF1 HD, ORF2 HD, arte HD, etc.) VBR material is blown up to CBR by NALU padding and that repacking the TS to MKV may reduce filesize down to 50 percent of the original.
See:
http://forum.digitalfernsehen.de/forum/hd-tv-die-zeitschrift/238797-das-maerchen-von-der-hohen-hd-bitrate.html
http://forum.digitalfernsehen.de/forum/hd-tv-die-zeitschrift/190891-orf-hd-vorschau-143.html#post375732
http://forum.digitalfernsehen.de/forum/hd-tv-die-zeitschrift/190891-orf-hd-vorschau-143.html#post3757702

In these posts it is adviced to
>1. demux the recorded.TS by tsMuxeR (~ 6 min) and then to
>2. mux to MKV by mkvmerge GUI from MKVtoolnix (and enter the respective fps attribute for the MPEG-4 par 10 ES video stream) (~6 min)

After
>3. demuxing the MKV with MKVextractGUI from MKVtoolnix (~7 min)
>4. and finally muxing back to TS by tsMuxeR (~6 min).
my tests yield TS-size reductions to 50%-70% of the original recording size.

Mkvmerge from MKVtoolnix (with its AVC/h.264 ES-Demultiplexer and its MPEG-4 part 10 ES-Video-output modul) seems to do the magic!

-> Feature request for TSPE_0.810:
Couldn't TSPE's Edit take care of removing the superfluous NALU padding on the fly?
It would save the time, effort and temporary disk space of the 4-step workaround! That would be a real cool, useful and efficient feature.

Regards

P.S.
TSPE_0.760 cured the Y2k10 and works with the license files from TSPE_0.755. Thanks!

P.P.S-
MarcN explains in
http://forum.digitalfernsehen.de/forum/hd-tv-die-zeitschrift/190891-orf-hd-vorschau-144.html#post3762895
"Nalu type 12 are so-called "filler data", large packages composed almost entirely of zero bytes. They are for example required if one requests the encoder to produce a fixed, high bitrate, but the picture does not provide enough information to fill the stream with it."

[rebkell]

Getting a zero byte file in Firefox and an error in IE when I attempt to download the new version...

Kirk

I just tried it with Firefox and it downloaded just fine, I'd already downloaded before on another machine and had no problems.

[Ventolin]

@Cela,

Thanks for that info, I'll have a look into it. It will have to come after 0.90 though or even 1.0 but certainly a feature worth having!

Regards,

Vent

[Lennen]

Hi Everybody,

in the need for an appropriate H.264 capable cutting program, I came across this promising software and highly consider it for a purchase, but for now I have some trouble with my HDTV recordings from the German Astra channels (Das Erste HD, ZDF HD, Arte HD, Einsfestival HD).

Although the recordings seem to be fine and playback well with either DVBViewer or VLC player, I get endless PTS gap errors of 0,0 seconds when performing the file scan in TSPE. The gaps are written to the marker list and when I am scanning a large file, e.g. 2 hours Castaway movie, the application will crash after a while. I have tried TSPE 0.76 and TSPE 0.81 with different directshow codecs (ffdshow, PowerDVD, CoreAVC) on two different machines (XP 32bit and Vista 64bit), but without success.

I played with the scan options and it seems with the three options: >suppress related errors<, >suppress nearby errors< and >disable M2TS header checking< enabled, the application at least does not crash and finishes the scan after approximately half an hour, creating a 1.4 MB Log file with 25000 PTS gap errors!!!

The problem only occurs to 1280x720, 50fps material from the above mentioned channels. 1920x1080, 25fps, H.264 material and 720x576, 25fps, MPEG2 recordings, recorded with the same DVB hardware, passes the scan without any error. I am a newbe to HDTV and unaware if it is a problem with my PC configuration, with the recordings itself or with TSPE, so i have uploaded a very short sample. Maybe some helpful soul here can test the file with his working configuration.

You can download the sample here:
http://rapidshare.com/files/340484379/Einsfestival_HD_test.zip.html
The ZIP archive includes the .ts file, the .log file and the marker list.

The log looks like this:
PTS gap of 0,0 seconds in PID 0x0641 at offset 0x766B0
PTS gap of 0,0 seconds in PID 0x0641 at offset 0xCE448
PTS gap of 0,0 seconds in PID 0x0641 at offset 0x10E054
....
....
PTS gap of 0,3 seconds in PID 0x0641 at offset 0x8C5AC4
PTS gap of 0,0 seconds in PID 0x0641 at offset 0x925BD0

PID Tree:
PAT PID: 0x0000 Ver: 0 TS ID: 0x04B1
0x0640: PMT Ver: 1 Program : 28396
0x0641: H.264/AVC Video (as defined in ITU-T Rec. H.264 | ISO/IEC 14496-10 Video)
0x0642: MPEG1 Audio (ISO/IEC 11172-3 Audio)
0x0646: AC-3 Audio (DVB) Stream


PID List:
PID: 0641: H.264/AVC Video (as defined in ITU-T Rec. H.264 | ISO/IEC 14496-10 Video)
H.264 1280x720 (1280x720) @ 50,000 fps
PID: 0642: MPEG1 Audio (ISO/IEC 11172-3 Audio)
MPEG1 Audio Layer 2, 64Kbit/s @ 44100Hz
PID: 0640: PMT
PID: 0000: PAT PID: 0x0000 Ver: 0 TS ID: 0x04B1
PID: 0646: AC-3 Audio (DVB) Stream
AC3: 2/0: L,R 48KHz 256Kbit/s


Using the new timeline feature in TSPE 0.810, it seems that the gaps are detected on every single I and P Frame. I also noticed that the 256Kbit/s, 48kHz MPEG1 Audio is reported false as being 64Kbit/s, 44kHz and the Approximate System Bitrate is obviously not correct.


Thanks for your support!

[onitoo]

I came across an issue with TSPE 0.810. I used TSPE to cut out the commericals from a HDTV recording with 3 streams: video, audio and subtitles. I trimmed out only 1 commericals (65 seconds) in the middle of the film clip. When I played back the trimmed film clip, the first portion before the cut point is ok for video, audio and subtitles. But in the second portion after the cut point, the subtitles are delayed by the length of time of the commericals (65 seconds in this case).

In a second trial, I unchecked the "fix timecodes" box. This time the subtitles are in sync with the video and audio after the cut point. But of course the total time length of the film shown in the player is incorrect.

Is there a way to use the "fix timecodes" feature and without the subtitles out of sync problem? In fact I noticed that this issue also exists in earlier versions.

Regards,

[pausefisk]

It seems that there is an issue staring/running TSPE when bitdefender antivirus 2010 is installed and running.

I've been using TSPE for months in my vista 32bit, with bitdefender antivirus 2009, running fine, but now i switched to windows 7 64bit, running bitdefender anitivirus 2010.

If i set antivirus in game mode, TSPE starts fine, but crash when i load .ts file (HD), or if it loads fine, it crashed when playing.

I'm not sure if this is because of the antivirus or another thing... i got both ffdshow and haali installed. Any suggestion?

[Ventolin]

Hi Lennen,

Sorry for not replying earlier.

Thanks for your sample and your bug report. It's quite an interesting stream you have there! Will add this to the bug list and figure out how to deal with it.

The audio is indeed reported incorrectly (a known bug which needs to be fixed!)

I think the approx system bitrate is correct though unless the duration is calculated incorrectly by both TSPE and the codec.

av. system bitrate = total number of bits / seconds

total number of bits = size of file in MegaBytes * 8

Regards,

Vent

[Ventolin]

I came across an issue with TSPE 0.810. I used TSPE to cut out the commericals from a HDTV recording with 3 streams: video, audio and subtitles. I trimmed out only 1 commericals (65 seconds) in the middle of the film clip. When I played back the trimmed film clip, the first portion before the cut point is ok for video, audio and subtitles. But in the second portion after the cut point, the subtitles are delayed by the length of time of the commericals (65 seconds in this case).

In a second trial, I unchecked the "fix timecodes" box. This time the subtitles are in sync with the video and audio after the cut point. But of course the total time length of the film shown in the player is incorrect.

Is there a way to use the "fix timecodes" feature and without the subtitles out of sync problem? In fact I noticed that this issue also exists in earlier versions.

Regards,

Hi Onitoo,

Sorry for the delay in replying.

It looks like there's a bug there with the subtitle timestamp fixing. I'm having a look at the timecode fixing section already so will try and figure out why the subtitles aren't getting fixed also.

Regards,

Vent

[Ventolin]

It seems that there is an issue staring/running TSPE when bitdefender antivirus 2010 is installed and running.

I've been using TSPE for months in my vista 32bit, with bitdefender antivirus 2009, running fine, but now i switched to windows 7 64bit, running bitdefender anitivirus 2010.

If i set antivirus in game mode, TSPE starts fine, but crash when i load .ts file (HD), or if it loads fine, it crashed when playing.

I'm not sure if this is because of the antivirus or another thing... i got both ffdshow and haali installed. Any suggestion?

Hi pausefisk,

Sorry for the delay in replying.

I actually am running Windows 7 64bit on my development machine so will test with bitdefender 2010 and see if I can replicate the behaviour.

Does this happen with TSPE 0.760 and/or 0.810?

Regards,

Vent

[Lennen]

Hi Ventolin,

Thanks for your answer. I now can at least stop to search for the right configuration… Sad to hear that there is no simple solution so far.

Is there an understandable explanation, what exactly is different with this stream to cause such strange behavior? As I mentioned before the problem occurs exclusively to recordings in 1280x720, 50p format, which has become the new standard for all public German HD channels. I also noticed that different recordings from the same channel show completely identical average bitrates, so I think the streams will most likely be affected by the NALU padding issue, Cela mentioned above. When I apply the described procedure to my recordings, the file size indeed shrinks by about 3 percent, but in contrast to my slight hope, it has no effect on the zero seconds PTS gap issue.

To the average system bitrate question: File size was 9.698.920 bytes. Divided by 1024^2 and multiplied with 8 gives 73,997 Mbits. Divided by 9,3 seconds I come to 7,95665454 Mbits/s. TSPE on the contrary reports 795.665.453.839.046,00 Mbits/s. The digits are the same but the decimal point has shifted rightwards.

Regards,

Lennen

[pausefisk]

Hi pausefisk,

Sorry for the delay in replying.

I actually am running Windows 7 64bit on my development machine so will test with bitdefender 2010 and see if I can replicate the behaviour.

Does this happen with TSPE 0.760 and/or 0.810?

Regards,

Vent

Hi.

TSPE 0.755 = OK!
TSPE 0.760 = NOT LOADING
TSPE 0.810 = NOT LOADING

As for the crashing, it was my interal directoshow filters that caused it, i got that solved now. However i remember i had some issues with some .ts recordins in 0.755 wich worked fine in TSPE 0.810, so i hope you find a way to fix this ;)

PS!! I would be NICE to be able to change witch audio that should be played when cutting the video. Some times my recordings have several languages, and i have to listen to the "wrong" audio when cutting, wich can make it a bit hard to find the correct spots.

[TheFrenchDumb]

stable version crashed on my seven 64 bit

beta seems to work (just launching)

[jmac698]

Hello,
I have a good test case for you to work on. This stream isn't working in your latest beta
http://www.mediafire.com/?97306hu22241bhr
since your program can't find it, I'll give you a hint - the streams are 0x980 and 0x981, 1080p mpg2 and 5.1 ac3.
Though I don't know much about using it, however I was hoping your program would let me fix this stream.
thanks.

[jmac698]

The problem has been found - does your program support Transport Sync Error? And it was probably cut with mp2cut2.

xport Transport Stream Demuxer 0.96
program = 1, video channel = 1, audio channel = 1
Program Number = 1 (0x0001), Program Map PID = 55 (0x0037)
Transport Sync Error, skipped 73 bytes, at 44 <------------------------------ kills my PID detection!
program descriptor = 0x88, 0x04, 0x0f, 0xff, 0xc4, 0xfc
Video PID = 2432 <0x0980>, type = 0x02
Audio PID = 2433 <0x0981>, type = 0x81
ES descriptor for stream type 0x81 = 0x0a, 0x04, 0x65, 0x6e, 0x67, 0x00
Audio Bitrate = 384000, Audio Sampling Rate = 48000
Audio Mode = 2/0, bsid = 6, bsmod = 0
Sequence Header found
26 frames before first Sequence Header
Horizontal size = 1920
Vertical size = 1080
Aspect ratio = 16:9
Frame rate = 29.97
Sequence header bitrate = 25000000 bps
Progressive Sequence = 0
First Video PTS = 0xa89e0530
First Audio PTS = 0xa89dfd3b, -2037
ts rate = 450760, video rate = 3479986
packets for pid 0 <0x0000> = 197, first = 1, last = 48047
packets for pid 55 <0x0037> = 154, first = 1012, last = 48048
packets for pid 2432 <0x0980> = 42407, first = 50, last = 48178
packets for pid 2433 <0x0981> = 5424, first = 45, last = 48183
packets for pid 5269 <0x1495> = 1, first = 44, last = 44
coded frames = 454, video fields = 1135

[patja]

TSPE is proving invaluable for repairing H.264 ts files from my Hauppauge Colossus HDMI source.

But I really need to be able to script it from the command line. All I need is the ability to load a source file and hit the edit button's method.

Is there an undocumented feature that would enable me to accomplish the above repair-only scenario in the current release?

If not, what are the plans for a command line-enabled version?