FairUse 2.6 is out

[UMP]

I think you should ask about this to the original author of FU, since he's probably far more aware about these things than I am. After all, I'm only a poor average-skilled developer doing its best to further develop a tool he likes. :)

Anyway, I would be very surprised if FU would bring down 30 fps interlaced/progressive to 24 fps, since the field detection code seems to be a lot smarter than the not-so-smart deinterlacer. :rolleyes:

Regards,

ump

[len0x]

Originally posted by UMP
Anyway, I would be very surprised if FU would bring down 30 fps interlaced/progressive to 24 fps, since the field detection code seems to be a lot smarter than the not-so-smart deinterlacer. :rolleyes:


but you just said in the previous post that FU detects 30fps as telecined source which is always brought down to 24 fps! and only on 24/25 fps source deinterlacing is applied...

[UMP]

Quote from my first answer : fu2k would give a far better answer to this question, but it seems to me that FU works by looking at the framerate and the interlacing (and maybe the fields flags, but I'm not sure about this).

Please note "fu2k would give a far better answer" and "it seems to me". :p

This said, I had a look at the code, and I think I was wrong. here is what ***I think*** is happening in the field mode detection code :

Statistics are calculated with fields comparisons, and field flag counting (Repeat First Field). It makes me think to the "video type = FILM > 95%" determination in DVD2AVI. The result is compared against predefined thresholds to decide what's the source type :

For PAL source (25 fps) :

Native interlaced: The video appears to be interlaced, but the fields could not be re-aligned => BOB deinterlace.

or

Native progressive: The video appears to be progressive.

or

Field align: The video appears to be progressive with occasional orphaned fields.


For NTSC source (not 25 fps) :

Native: The video does not appear to be telecined.

or

Ivtc: The video appears to be telecined.


Hope this helps,

ump

[fu2k]

FairUse detects interlacing in a reconstructed frame by comparing the characteristics of the even and odd fields. If these characteristics are substantially different in the same small local area, then this area is considered interlaced (with varying degrees depending on the difference). These differences are accumulated across the entire frame (using a non-linear method) and if the total metric exceeds a threshold (experimentally determined) then the reconstructed frame is considered to be interlaced. See: "DecodeVideoInterlaced::_Interlaced(...)".

The information as to whether a frame is interlaced or not (and by how much) is used by field aligner to try to put fields back together to form progressive frames.

On PAL sources, the field aligner is limited to dropping the occasional orphan field to maintain alignment. This is often necessary on scene changes when the material has been edited on field based equipment. Of course, some full frames must be duplicated to maintain audio sync. See: "DecodeVideoFieldAlign::_Deinterlace(...)".

On NTSC sources where the user has selected IVTC, the field aligner attempts to dynamically match the IVTC pattern, realign fields and discard the duplicate fields added add by the telecine process. See: "DecodeVideoIvtc::_Deinterlace(...)".

The code that tries to predict whether IVTC is appropriate simply process a subset of the non-RFF parts of the input video using the IVTC algorithm and looks at the interlacing levels of the output. If there is "too much" interlacing then IVTC is not recommended to the user. See: "AutoModeNtsc::Process(...)".

Any interlacing that is present at the end of field realignment is removed by the resizing code, because a low pass filter is applied during this process.

[UMP]

Here is a link to the pre-release of FU 0.33 :

Link removed. Please see http://fairuse.sourceforge.net for latest release

Please not this release has not been fully tested. If some people gives it a try, it would be nice to send me some feedback to : fairuse AT free.fr

Changelog :

- huge improvement in encoding speed
- slightly changed the BeSweet command line to ensure optimal audio gain (solves the "low sound" some people reported)
- the "shutdown after encoding" feature should now work on Win98 systems (I can't test this myself because I haven't used Win98 for ages)
- added the ability to select several IFO chains to cache (no longer limited to one or all chains)
- added external language files so that FairUse can be easily translated (please email me if you want to contribute to translation)
- added project batch processing (use "Defer processing" on the encoding dialog, then "Defer project processing" on the project dialog)

[Boow]

Last time I used this program there was no option to use files already on your hardrive. will newer versions suport this or am i behind on this?

[UMP]

Originally posted by Boow
Last time I used this program there was no option to use files already on your hardrive. will newer versions suport this or am i behind on this?
FU does not take VOB files as input, and it won't. If you need to encode from previously ripped VOB files, you have *lots* of tools out there that will do the job, my personal favorite being GordianKnot.

ump

[spectra]

i'm ripping some DVD extras at the moment and i noticed, that fu ALWAYS uses the filesize as only parameter. even if i add my passes manually and change the bitrate to the bitrate i desire it ignores this manually entered bitrate and takes the bitrate that suites to the filesize that is entered. with normal movie files this is no problem cause you always want to fill 1,2,3 cd sizes, but with extras on dvds, i'd like to have a certain bitrate and do not know how big a 6-minutes extra will be in the end. at the moment i have to try a filesize check if my desired bitrate is reached and try again. is there a possiblity to update the filesize field automatically after i added a pass and manually changed the bitrate to the one i desire.

EDIT: Well ok i found out that with autoadd, you add the streams and then change the filesize and the bitrate in the streams change automatically. so that is good enough for me.

[lar1r]

Nice to see development continued again.

You already answered my first question about allowing FU to use VOB's as input (oh well).

How about adding AAC or OGG as potential audio options? (the output filename would have to be a MP4 or OGM respectively. The seeking in these files is so much better as is the sound quality). If not then the AC3 option will do since I can always use OGGmux or the aacenc/mpeg4creator tool seperately.

Regards

Larry

[UMP]

Ogg was once implemented, and should be back one day or the other, and AAC is on my TODO list, but some issues are still to be fixed before these can be implemented.

Regards
ump

[lar1r]

Ogg was once implemented, and should be back one day or the other, and AAC is on my TODO list, but some issues are still to be fixed before these can be implemented.

Sounds great! Keep up the good work!

I have a question regarding the resolutions FairUse comes up with based on a compression number. Should the setting more or less that's recommended be similar to how Gordian Knot analyzes the compression on a movie file, or how Nero Recode analyzes the mpeg in the vob file??

Thx

Larry

[chilledoutuk]

erm why are you so against being able to slect vobs that have been already ripped?

My encoding maching does not have a dvd rom in it and is not easily accessible surly it would make sense to be able to select a VIDEO_TS folder as the source otherwise many people like myself wont be able to use this tool that looks very nice indeed.

please consider this.

[spectra]

Originally posted by chilledoutuk
erm why are you so against being able to slect vobs that have been already ripped?

My encoding maching does not have a dvd rom in it and is not easily accessible surly it would make sense to be able to select a VIDEO_TS folder as the source otherwise many people like myself wont be able to use this tool that looks very nice indeed.

please consider this.

i had he same problem and i made a workaround by ripping a dvd with "iso mode" with dvd-decrypter. Then u can mount this file with daemon-tools on your fairuse computer. fu detects this mounted image as real dvd.

it's just a weak workaround, cause you always have to rip the complete dvd and also need this space additionally cause fu then produces it's own structure for ripping.

[bilu]

Does FairUse deal with hard-telecined (hardware telecine, without MPEG-2 RFF flags) and hybrid clips (mixed telecine and pure video)?


Bilu

[UMP]

I don't know much about this part but this is from the FAQ :

The IVTC functionality is integrated deeply into FairUse, and was written from scratch. It looks for interlaced frames and repeated fields, and should work properly even when the IVTC offset changes, or the video switches between hard and soft 3:2 pull down.

The same interlace metric is used on PAL titles to realign fields that have been either orphaned after editing or incorrectly interleaved due to mastering errors.

[mordant]

Just a follow up on interlacing. I recently did a HK flick with FairUse 34b where I wanted two CDs but wanted to choose the split point myself, so I used the range selection and did each disk as a sepperate session...what happened in the autodectect part for each half of the same film was different. The first half of the movie IVTC was suggested, the second half Native was suggested. I tried each part a couple of times and the same thing happened. Finally choose IVTC for both and they came out fine.

[bilu]

Originally posted by UMP
I don't know much about this part but this is from the FAQ :

The IVTC functionality is integrated deeply into FairUse, and was written from scratch. It looks for interlaced frames and repeated fields, and should work properly even when the IVTC offset changes, or the video switches between hard and soft 3:2 pull down.

The same interlace metric is used on PAL titles to realign fields that have been either orphaned after editing or incorrectly interleaved due to mastering errors. That's why nobody answered me yet in this thread: http://forum.doom9.org/showthread.php?s=&threadid=68653

The PAL method seems pretty simple and efficient but the NSTC methods are a bit unclear: hard and soft 3:2 pulldown are both pulldowns, i.e. 30 fps -> 24 fps conversions, the only difference being soft pulldowns being guided by the MPEG-2 RFF flags - you just need to remove the RFF fields.

But NSTC doesn't need pulldown everytime - you can have interlaced 30 fps sources. And sources that have 30 fps interlaced and telecined parts. Does FairUse know how to deal with them?

Bilu

[UMP]

The only person who can answer this did answer to this in the thread you mention.

ump

[Doom9]

But NSTC doesn't need pulldown everytime - you can have interlaced 30 fps sources. And sources that have 30 fps interlaced and telecined parts. Does FairUse know how to deal with them?You might want to give it a shot ;) I know a few easier sources where the film content is enough to force film but the result is not very good (the pulldown pattern changes and you can notice that in the output), but where FU has no problems. You'll want to either look in the source code or try to find the original author to get a more comprehensive answer.
I know the original author once said something to me regarding cut off of high frequency (flickering) but that's all I know and he never went into more detail than that.

[lakedude]

I can't say enough good to convey how much I love FairUse. I've used a bunch of different rippers/converters and IMO FU blows them all away. FU30 is bullet proof and very easy to use. It never fails to produce good results. I wonder if it would be much trouble to make a version the includes the audio and auto shut down features of 34b but that still uses the old divx 3.11 and the cross cutter?

The reason for asking is that I LOVE the idea/functionality of 34b but in practice it is more buggy than 30. Also 3.11 scales and plays better on the ole 450mhz/win 95 machine.

Is there a way to send the developers (fu2k,UMP)a donation? I use FU30 all the time and plan to use the newer version once the bugs ironed out.

Before I start in about the problems let me first say that the changes in 34b are great. It is wonderful to be able to set 34b up and just let it run till its done and have it auto shut down without having to do any extra manual steps.

Ok now the problems: I used 34b with divx5 to encode Once Upon a Time in Mexico with auto add 2 pass and default settings(700MB/128kbps mp3). The avi turned out small at only about 550MB and the quality was very variable. When the love birds are swinging by the chain outside the window the blocking is so bad it looks like the thing they do to whistle blower's faces so you can't see who it is.

I tried again with Xvid and it looks like every other frame is black so the xvid version flickers. I'm trying Xvid again after a reinstall as we speak. I hit a reset or load defauts button and one screen went from having frame 0 and frame 1 to just having frame 0. I suspected that the missing frame 1 was the problem so I'm trying again.

Using the same rip I tried FU30 and divx3.11 and got good results.

My trouble with 34b may just be settings cause I'm not used to using anything other then FU30.

Anyway TIA for any help.

[lakedude]

Update

That frame 0 / 1 thing musta been the problem with the xvid encode. The new one turned out great. If I can get ffdshow to work correctly with the version of media player on my laptop I'm switching from fu30/divx3.11 to FU 34b (or 36)/xvid.

[lakedude]

Ok final update:

My laptop has been goofey for a while now but it would play 3.11 divx fine so why fix what ain't completely broke? Well I just couldn't stand it anymore and reloaded all the software from scratch. Of course it works great now and will play xvid video with hardware acceleration enabled (something it couldn't do before). Now that the Laptop is fixed all my new movies will be fu34/36/xvid. Color me happy:-)

[UMP]

Hi lakedude, and thank you for your feedback :)

In some FU releases I tried to do too much XviD automation, but this thing has been removed from the latest releases. The newest FU 0.36c release fixes several issues, and should be the only release to use (until the next one, at least :p).
Every bug I could reproduce has been squashed from the latest FU release. Some people reported a "key not found" issue, but I've never been able to reproduce this on my system.

BTW, there is no way do donate money to the FairUse project. I won't accept donations because the original author never did so and the more important : I consider my work is nothing compared to his, since I would never have been able to code such software from scratch on my own.

About the undersize issue, you might want to search the forum for the terms "undersize" or "codec saturation".

Cheers,

ump

[lakedude]

Wow thanks for the quick response. You mentioned that development was suspended and I was expecting a much longer response time.

Where would be a good place to discuss bugs/problems?

I downloaded 36c to a win98se machine that had just been reloaded and to an XP machine and both had problems encoding to xvid. In both cases the video flickered like everyother frame was black and the frames that did contain video would get black squares in it. I had the same problem with 34b and xvid on one machine after hitting a load default button but the 34b version worked fine after a reinstall. The 34b version works fine on the laptop with 98se as well.

I did a drive image (ghost) right before installing Fair Use so I could go back to a machine with no versions of Fair Use to test em with clean installs instead of installs over previous versions. I thought that the problem with 36c was the xvid settings so I compaired them line for line with the settings on 34b and made them all the same and 36c still didn't work right as the video still flickered and got black blocks in it.

Could you make older versions available?

Oh BTW the download of 33 asks for a second disk that seems to be missing so the install fails.

[UMP]

About the latest releases

Oh BTW the download of 33 asks for a second disk that seems to be missing so the install fails.
The file is incomplete, I'll upload it again. Thank you for pointing this.

the video still flickered and got black blocks in it
What decoder do you use ? (this can be determined with GSpot)
Please note that FU 0.36+ uses XviD 1.0 RC1. I suspect the problem to be located on the decoding side.

BTW, there is a 0.37 release available.

Cheers,

ump

[lakedude]

Dear UMP thanks again. I Love what you are doing and wish I had the skill to help on the project.

What decoder do you use ?

I used Gspot and it shows:

src type as: 4cc xvid
video path is ffdshow

I used Gspot on both a clip that works fine and the blinky/blocky clip and all the info is the same except the "Stream Type and A/V interleave I/L:"

On the one that works it shows: "1 aud frame (24ms) Split: Yes"
On the one that blinks it shows: "22 vid frames (734ms) Split: Yes"

I could send you a short clip except my copy of Virtual Dub will not open the xvid video so I don't know how to edit the video for lenght.

Anyway it looks like an interleave issue??

[UMP]

I don't think this is an interleaving issue. To be able to open the video with VDub you have to properly install the codec.

Cheers,

ump

[lakedude]

I'm re drive imageing my c: partition right now and I'm gonna install the xvid-rc1 codec. I guess I assumed that fu36c came with it......

ok partition re-made, rc1 installed....VD works!

Ok I uninstalled ffdshow and now that the clip is using the Xvid decoder it looks fine. Picking the xvid option within ffdshow does not work (still flickers).

So far so good cept how come ffdshow don't work?

[UMP]

I'm not sure about this. On my system ffdshow works fine with XviD RC1 as long as there is no more than 1 consecutive BVOP.

You should use the search button, I guess...

Cheers,

ump

[lar1r]

Hey UMP, I just encoded a couple of movies using AC3 and on playback only the right channel was getting sound. It's happened everytime I've tried it.

I've checked my Sound control panel and watched what channels were outputting sound, and sure enough, the right channel was reporting everything fine, the left channel sometimes had very low sound coming from it, but the rest were silent.

I've tried this with AVI and OGM file names.

Does AC3 work for you (even testing with a two channel configuration ie: headphones, are you getting equal sound on the left and right channels when using ac3)?

Also I wanted to thank you big time again. The ability to change XVID settings (for setting zones, and configuring different encode settings), set a start/end frame, encode using OGG are all features that make this program great and standout from the rest.

Regards,
-Larry

[UMP]

Originally posted by lar1r
Hey UMP, I just encoded a couple of movies using AC3 and on playback only the right channel was getting sound. It's happened everytime I've tried it.
AC3 streams are simply extracted to AC3 files then muxed "as is". I fear that the problem is on the playback side somewhere on your system.

You might be able to check this by using DVDDecrypter to extract an AC3 stream to HD and play it back. It is likely that you'll experience the same behaviour.

Does AC3 work for you
AFAIK, AC3 works for all FU users, since as previously said AC3 streams are only demuxed and remuxed to AVI/OGM/MKV without further processing.

Regards,

ump

[Grassy]

Hi UMP,

First like many others here I have to say this is fantastic piece of software, really makes it easy to convert multi-episode discs.

I have only had two issues with this in that a couple of times when I have queued up 4 episodes of the one disc to all be done at the same file size, one of them will be about 20-30MB smalller than the others. The size of the original files are about the same so I don't know why that has happened, the rip has finished OK though, nothing was truncated and the file plays fine so not that big of an issue. I haven't repeated enough of these to give you more info on this but will post back with results after I do these ones again.

The other thing is that when I try to enter a new file comment (like INAM : Title, IART : Artist) I get a error message about "No room to insert", I don't get this error using other encoders so do you know what your program might be doing to cause that ?

Again thanks for creating a wonderful program, well done ! :)

Cheers.

[UMP]

This is likely a codec saturation issue. A search on this forum with the terms "undersize" or "codec saturation" will learn you more about this.

Concerning the other thing, FU currently uses the standard "AVI Mux Filter" to do the multiplexing of audio and video, in a graphedit-like way, so I can't do anything about this ATM. This could be subject to change in a future release.

Cheers,

ump

[lakedude]

How come 38 will not do old saved projects? All the other 3 series from 30 to 37 will take fups from one another.

[lakedude]

UMP

While i'm begging for favors......(are you sure you won't take any money?) An audio setup would be nice as an "expert mode" option. I'd like to make the sound a bit louder.

Many thanks for all you have done. Great program!

[UMP]

0.38 data and project file format changed to support forced subtitles.

The audio is already DRCed and normalised (postgain).

[flloyd]

Would it be possible to make FairUse minimize to the system tray? I would love to have this feature.

[UMP]

Maybe later, but this feature is not at the top of the todo list ;)

Regards,

ump

[lar1r]

Figured out the AC3 problem - it was the filter I originally installed (I think it was from ffdshow).

Once I installed ac3filter0.7 the ac3 sound from FairUse worked properly.

Hey UMP are Zones a feature you are thinking about (that could be set from the preview window)? Not really a big deal since it can manually be done in the codec setup.

[UMP]

If there are exactly TWO zones, the second one will be used for the credits encodings, else FU will ignore these, so you can set them up manually.

Regards,

ump

[chemmajik]

The ability of working off a hd vob directory of the dvd just add a screen to allow skipping dvd decrypt.... or reading dvd... option to decide whethere off hd or dvd...?!

[r6d2]

Originally posted by chemmajik
The ability of working off a hd vob directory of the dvd just add a screen to allow skipping dvd decrypt.... or reading dvd... option to decide whethere off hd or dvd...?! From the FAQ (http://fairuse.free.fr/faq.html):
Q: How can I use FairUse to transcode the VOB files that already are on my hard drive ?

A: You can't. FairUse was not designed to take input from VOB files. :)

[UMP]

I don't understand why this seems to be such a big issue ? One should finish any existing encodings with another tool, then switch to FU for a new backup.

Regards,

ump

[lar1r]

Just noticed that the frame number displayed in the preview window doesn't actually correspond to the right frame all the time.

I was setting ZONES near the start of the file (the Bond intros), and noticed that it was kicking in ~2000 frames too late and ending ~2000 frames later. I used the preview window to find out where to begin and end the zones and inputted these into the codec first and second pass setup options.

My workaround was to encode the first bit and then go back into virtualdub to determine the start and end frame numbers.

The interesting point is that when I set encoding credits near the end of the movie, it turns out fine.

-Larry

[UMP]

If we're talking about a R1 DVD, the frame count might change depending on the framerate if there is any IVTC (29.97 vs 23.976)

[lar1r]

I bet you that was it. They were older Bonds from the 60's which were probably at 29.97. (R1 DVD)

[lakedude]

Hmmm, seems to me like fu30 never had interlacing artifacts but a couple of my newer fu videos do. I think the "Smart Deinterlace" takes care of such things but if you hit "auto" it turns "Smart de" off by default. Is this correct so far?

Anyhow would it be possible to turn "SD" on by default? Or at least not have "auto" turn "sd" off?

[UMP]

SD will be turned on when the source is determined to be interlaced, but could not be reconstructed (ie. if an interlaced source was remastered as progressive without prior deinterlacing).

I did not change anything in the interlacing detection algorithm, so it works the same as it did in FU 0.30.

What you can do is re-enable it when you feel it could help. I think the deinterlacing thresholds would need more tweaking, but it's something I don't fully understand, and any help with the interlacing/IVTC mode determination process would be highly appreciated. :)

Regards,

ump

[fu2k]

The resizer that FairUse used would remove any remaining interlacing artifacts during resizing. So if the resizer has been changed then, this might be why more interlacing is visible.

[UMP]

Well, I see. I replaced the original resizing code with "simple" MMX optimized resizer because resizing was way too slow. Since I don't know much assembly myself, I couldn't convert the existing algorithm.

ump