HC encoder

[osho]

I must say HC works amazingly great on x64.

I my previous install of x32 win pro HC froze the whole pc basically every 10 mins.

now on x64 it runs far better. dont know if it is by reinstall of whoel win but it also runs faster. getting now around 70fps on the first part of the pass.

[_argos]

i can confirm this error on exit thingy. got an error when using AVISource but NO ERROR with DirectShowSource.

my Cpu is Athlon64 3200+ with WinXP SP2.

_argos

BTW: I uploaded an altered version of the HCProvider for Cuttermaran with the DirectShowSource-Workaround. Get it here (http://home.arcor.de/frostyboy/data/HCProvider_altered.7z). Source (AutoHotkey) is included.

[skipper152]

Hi,

My system (with the applications mentioned installed)
AMD Athlon 64 3500+
1 GB Memory
WinXP 32-bit SP1
ffdshow version oct 12 2004 08:43:06
AVISynth 2.55 build sep 1 2004 [16:49:49]
XviD XviD-1.1.0-Beta2-04042005 _Beta Release_

It seems the only time I have problems with HC_015 is when I have AOL running. So for me to get the avs to crash using AVISource on exit I have to load up AOL. So, I'm confused as to why the exiting problem doesn't happen on my Athlon 64...

Changing the script from AVISource to DirectShowSource works fine (mentioned by TFM_TheMask).
I have nothing checked in the ffdshow raw-only filter configuration (mentioned by LigH).

I've tried QuEnc and there have been no problems with the avs scripts at all.

I don't know too much about video processing, but if there is something I can look at or try to help out, let me know. (I'm working night shift this week, so I'll try to get back as quickly as possible!)

Skip

[_argos]

another thing i've noticed is: the crash only seems to happen if the hc_batch.exe is called from within another program as in dvd-rebuilder or in my case from Cuttermaran-"Provider" (which is an AutoHotkey script turned to .exe). when i launch the program from the command line the error does not happen. strange, eh?!

_argos

[DrShrinker]

another thing i've noticed is: the crash only seems to happen if the hc_batch.exe is called from within another program as in dvd-rebuilder or in my case from Cuttermaran-"Provider" (which is an AutoHotkey script turned to .exe). when i launch the program from the command line the error does not happen. strange, eh?!

_argos

I ran it from the command line and got the error. :confused:

[_argos]

I ran it from the command line and got the error. :confused:
Ok, screw my theory. seems it'll happen random. :devil:

[jul059]

Hey.

Once again Hank does a brilliant job. Guys please remember, Hank does this in his free time so lets appreciate his work.

Suck up !!!!!!!!!!!! :-)

Hank: When encoding PAL movies with a low bitrate ( Ave:1600 Max:4000 ) I get a brilliant picture and a 90 min movie to +- 1.2 Gig. However I noticed night areas with a minimum difference in color cause blocky playback ( PC with WinDVD through a TV out ). Could it be the matrix ( HVSBest ) ??

I actually got a simillar scenario as yours and as urvieh's, but with medium bitrates. There is abnormal blocking on certain scenes (molsty darker but not necessarly black background) when I encoded the movie Hitch. I used the default matrix from HC for everything except menus and extras (which were 33% decreased in quality and used BACH1). the extras are actually VERY blocky, I would say much more than 1000 kbps MPEG1.

I can give more information if needed.

[Kika]

What encode mode do you use? 2pass or single pass?
At single pass, sometimes the constand quantizer itself is the problem. And there's not much you can do.
At 2pass, you can change the VBR behavior to get better results.

[jul059]

I always use 2 pass with the profile best. My matrix for the movie was HC default (I don't know if it's any good. Is it?). for the extras, BACH1. DC and GOP use the DVD-RB defaults. What exactly did you mean by changing the VBR behavior?

[Kika]

What exactly did you mean by changing the VBR behavior?

VBR bias
Set it around 20-40 and you will get better results in low motion/low contrast scenes.

[Prodater64]

I got a Visual Fortran run-time error twice, with same source:

http://img150.imageshack.us/img150/473/screenshot7gv.png

What is this?

[PJones2101]

I actually got a simillar scenario as yours and as urvieh's, but with medium bitrates. There is abnormal blocking on certain scenes (molsty darker but not necessarly black background) when I encoded the movie Hitch. I used the default matrix from HC for everything except menus and extras (which were 33% decreased in quality and used BACH1). the extras are actually VERY blocky, I would say much more than 1000 kbps MPEG1.

I can give more information if needed.


Hi jul059 - Help would be appreciated PLEASE.
Here are the settings I normally use.
Bitrate Ave : 1600
Bitrate Max : 4000
DC Precision : 10
CPU : SSE3
Scan Method: ZigZag
Matrix: HVSBest
Profile: BEST

As stated by you and me :-) The light areas come up perfectly, however the dark areas, if the colors are closely the same, gets blocky.

You are welcome to contact me via EMail directly - if we are allowed to post an EMail address here.

[Fishman0919]

Hi jul059 - Help would be appreciated PLEASE.
Here are the settings I normally use.
Bitrate Ave : 1600
Bitrate Max : 4000
DC Precision : 10
CPU : SSE3
Scan Method: ZigZag
Matrix: HVSBest
Profile: BEST

As stated by you and me :-) The light areas come up perfectly, however the dark areas, if the colors are closely the same, gets blocky.

You are welcome to contact me via EMail directly - if we are allowed to post an EMail address here.

The matrix you are using (HVSBest) is compressing to much with the dark scenes and causing the blocks most likely, try BACH1 (it's a little better with the dark scenes and great for encoding around 1000 to 1800k)it will help with the dark scene blocks... also try DC Precision 8, with as low of a bitrate you are using it will also help... setting VBR Bias to high with a low bitrate can hurt more then help, with the avg and max bitrate you are using the default setting is probably the best way to go.

[gigah72]

is it possible to implement or use such an auto matrix option in hc encoder like inautoqmat?

[SAPSTAR]

is it possible to implement or use such an auto matrix option in hc encoder like inautoqmat?
You can use QMatOp...it's not as good/fast as the version implemented in AutoQmatEnc ...

[hank315]

I got a Visual Fortran run-time error twice, with same source:Seems HC tries to write or read from unit 23 (database file) which doesn't exist.
Could you give some more info about your encoding settings.

@jul059
2 questions:
- is playback the same on your PC and HW player for those scenes
- what is the average bitrate for these scenes

[jul059]

@hank315

yes, the playback is pretty much the same on both my hardware (Panasonic RV31) and my computer (both nero showtime and winDVD).

for the average bitrate, the only way I thought about is reading the values given from SmartRipper (because I have deleted the .m2v files...) and I got extremely strange results:

average bitrate is 456 kb/s for the main movie!!!!

the extras are all around 200 kp/s.

I don't understand, the DVD is full and the resulting files were almost oversized (but fitted just right on a DVD). maybe it's just SmartRipper. If you could give me another way, maybe I could check the "real" bitrate...

anyway, here's a screenshot of one of the extras: http://img110.imageshack.us/img110/2713/badblocks4fw.th.png (http://img110.imageshack.us/my.php?image=badblocks4fw.png)

[Mug Funky]

try bitrateview :)

[Kika]

@jul059

That's the result of high quantisation at low Bitrate (or too low bitrate only).
This often happens if the Matrix is one of those optimized for high compression (Bach1, as an example).
Playing around with the Matrix isn't as effectiv as using good settings for Bitrate. Like i wrote: Change the Settings for VBR bias to get a higher Bitrate in such parts of the Video.

[PJones2101]

Thanks to all for the advice - I know my rates are very low but HC does such a great job.

Anyway, Im going to spend a few days trying all the advice.

[jul059]

@jul059

That's the result of high quantisation at low Bitrate (or too low bitrate only).
This often happens if the Matrix is one of those optimized for high compression (Bach1, as an example).
Playing around with the Matrix isn't as effectiv as using good settings for Bitrate. Like i wrote: Change the Settings for VBR bias to get a higher Bitrate in such parts of the Video.

indeed, I think you were right. The average quantisation on the extras is extremely high: around 17, while the bitrate is around 1400. (thanks Mug Funky for bitrateviewer!)

but that doesn't explain why I get abnormally blocky (but not unwatchable) scenes in the main movie, where the average bitrate is around 3400.

Also, is there a way in DVD-RB to automatically change the VBR biais on the extras?

[Kika]

but that doesn't explain why I get abnormally blocky (but not unwatchable) scenes in the main movie, where the average bitrate is around 3400.

Always look at the Bitrate AND the quantisation. HC acts very similar to TMPGEncs new 2pass Mode. If you are interested in some testings, try TMPGEnc with the old(!) 2pass Mode. After that compare the encodings with Bitrateviewer and look at the image quality. I know the result right now: in some scenes, HC will do the better job, but in some scenes, TMPGEnc will be the winner. And with the data from Bitrateviewer, you can see, where and what the problem is.

[Rumbah]

In the German Cuttermaran Forum a user also got the error message:

Die Anweisung in "0x09168bfc" verweist auf den Speicher in "0x09043628". Der Vorgang "read" konnte nicht auf dem Speicher durchgeführt werden.

He has a P4 3Ghz HT on WinXP SP1 with Avisynth 2.55 and XVid 1.03. As I thought it might be something with the YUV Codec (Cuttermaran delivers YUV avi), I suggested that he might try the Helix YUV codecs. So he deinstalled XVid and installed the Helix codecs and everything worked fine. Then he deinstalled the Helix codec again and installed Xvid and the error was there again. He didn't try to install Xvid and the Helix codecs, so I cannot say about that. So the error could be between Xvid and HC. So you guys with the error could perhaps try the Helix YUV codec and report if it works for then.

Rumbah

[Amnon82]

Is there a switch for the batch-version of HC to enable"check max br"? I hope with this HC doesn't go over the max bitrate anymore. ... will be helpful for AutoCQ.

[manolito]

Is there a switch for the batch-version of HC to enable"check max br"? I hope with this HC doesn't go over the max bitrate anymore. ... will be helpful for AutoCQ.
It's not a switch, it is a different parameter in the HC.ini file (introduced in version 0.15). Instead of "*CQ" you have to specify "*CQ_MAXBITRATE".

Cheers
manolito

[LigH]

According to katjarella, HC encoder works well with AviSynth 2.56 b4:

Example picture (http://katjarella.gleitz.info/sub/hc15beta.jpg)

[PJones2101]

Does anybody know of a good Matrix to use for animation. I have South Park episodes in DVD format, however I want more than 4 on a DVD and plan to put 8-10 on a disk.

Thanks in advance.

[LigH]

You may try the Kika matrices. They are included in my collection you can find in the thread about my "Custom Quantizer Matrix Editor" (CQME).

[gigah72]

Does anybody know of a good Matrix to use for animation. I have South Park episodes in DVD format, however I want more than 4 on a DVD and plan to put 8-10 on a disk.

Thanks in advance.

how do you want to put 8-10 on one disc, like create own menu and then author yourself, or what way?

[PJones2101]

Yup.

I have the 8 files in AVI format and will convert to m2v with HC.
I use DVDLabPRO so I'll create my main menu, chapters, episode menu etc with that an imported my convert m2v and ac3 into that.

I just need a good matrix for compressing the animation. I know this is picky ( fussy me ) - but the lower I can get the filesize the better. I test all files compressed on a ATI TV-Out card and on a DVD-/+RW on my DVD player first.

I'm just looking for that extra space.

[PJones2101]

You may try the Kika matrices. They are included in my collection you can find in the thread about my "Custom Quantizer Matrix Editor" (CQME).

OK I must be a complete idiot - well most of the time - can you possible post a link please ( If it's allowed here ).

Thanks in advance.

[LigH]

Okay - for someone who may be too lazy to type "CQME" into this board's "Search" function:

http://forum.doom9.org/showthread.php?t=73666&highlight=CQME

http://www.ligh.de/software/qmatrix.zip

[PJones2101]

I Typed "Custom Quantizer Matrix Editor" instead - I'm just useless.
Thanks, besides, I'm sure you helped someone else as well.

By the way - Awsome tool - Great job and very appreciated.

[LigH]

My fault - it's called "Custom Quantization Matrix Editor" instead. :o Little typo, big fault.

[PJones2101]

Not a problem because it led me to a lot of Matrix's to play with.

If anyone else is looking for Matrix's type "Custom Quantizer Matrix Editor" into the Search box and go through the 5 forums.

[feedback]

@PJones2101
Here is a link to a matrix zip folder. Some included are the same as in LigH's zip but some are different.

Matrix Zip Folder 1. (http://forum.doom9.org/attachment.php?attachmentid=4081)

Regards,:)

[Xesdeeni]

I authored a DVD with two different authoring programs based on an MPEG created with HC. On two DVD players, I get audio (AC3 from the original ATSC stream) with no video. If I fast forward 8x or more, I get individual frames. On two others, and on four PCs I tried, everything is fine. Any idea what could be the issue?

Xesdeeni

[Mug Funky]

interesting. what sort of player was it (if you don't mind my asking)?

[Xesdeeni]

One of the ones without the picture was a Terapin I got at Fry's. The other was my sister's, but I can't recall what type it was (sorry, but not the same--I'll e-mail her and find out). A GE (3-Disc) and a Samsung (VCR+DVD) worked fine.

Xesdeeni

[Xesdeeni]

My sister's is a Phillips Progressive Scan.

Xesdeeni

[video_magic]

WMV3 problem via Avisynth (version 2.56 from July 10th 2005) encoding in HC gui v0.15
What happens is that I create an avs file with this line:

DirectShowSource("c:\atmp\jrmoved\converted\JaMes_Randi__Homeopathy - The Test.wmv", fps=25)
Lanczosresize(352, 288)
ConvertToYV12()

it makes NO difference if I also include the , convertfps=true after the fps=25 or not.

The AVS will play absolutely fine using Media Player Classic and also in Virtualdub v1.6.9
However, when I try encoding the AVS through HC gui v0.15 there is a problem with the first 2 to 9 seconds being duplicated and then throughout the clip it gets even more out of synch between video and audio.

My solution is to put the avs into VirtualDub 1.6.9 and turn it into a HuffYuv avi. I check the Huffyuv and the AVS has encoded to it fine, and then make another AVS of this huffyuv clip to encode it through HC gui :)

This then produces my mpeg-2, but anyway, it looks like an HC with AviSynth issue I think, I have produced several clips using avs from different formats with exactly the same HC settings (it's to put on a DVD compilation) and this is the only one to have this problem, and the only WMV, it's not DRM'd.

Here is the mediaInfo report on the original video:

File size : 75.9 MiB
Format : Windows Media
Overal BitRate : 242 Kbps
PlayTime : 43mn 44s
Title : Horizon - Homeopathy - The Test
VBR Peak : 618725
Buffer Average : 144149

Video #0
Codec : WMV3
Width : 384
Height : 288
Aspect ratio : 4/3

Audio #0
Codec : WMA2
Bit rate : 48 Kbps
Channels : 2
Sampling rate : 44 KHz

[Delphin]

I have an idea for a unique feature that I think would set HC apart from ALL other MPEG2 encoders.

A fast capture feature using I Frames only, with a 'smart re-encode' feature in HC that would use this pool of I frames to create a standard high compression IPB structured MPEG2 video stream WITHOUT RE-ENCODING THE IFRAMES.

Ok, the "I-Frame only" thing has been done before (I have an old ATI 'all in wonder' card which does that).

Where they dropped the ball, was that there was no 'smart render' feature provided with the tools to do a minimum-loss re-encode of the resulting stream without re-encoding the I frames.

I found the "I frame only" MPEG2 captures to be very comparable to MJPEG (no surprise there), so this would allow analog captures without using the huge amounts of drive space required by the HUFFYUV codec.

I would approach this with a two part stratagy . . .

1. Write an " I-Frame AVI codec " which could be used for capture and editing.

2. Add 'smart render' support to HC for 'minimum-loss' re-encoding of these I-Frame only files.

The AVI file could also be edited more efficiently since it would contain all I frames (equivelent to AVI codec 'key frames'), and unlike MPEG or MPEG2, 'smart rendering' such a file (rendering only changed frames) would be trivial, since we would not have to mess with any complex GOP structures.

The idea here is not to speed up HC (though I would expect a tiny improvement since HC could just pick rather than encode the required I frames), but rather to eliminate the lossy MJPEG format as an intermediate format and facilitate high quality 'direct to MPEG' captures from analog sources.

This approach to analog capture would give the 'best of both worlds' in that we would get the speed of MJPEG (with an I-frame only capture codec) and then could do a truly optimal minimum-loss translation of the resulting file into a full IBP structured MPEG2 file by taking advantage of the better bit allocation and motion search capabilities of a non-realtime off line encoder like HC.

Sadly I know that 'analog capture' and 'editing' features may not be on the high priority list for a lot of forum members who are ONLY concerned with how they can, make more efficient copies of existing DVD material (though we all assume that this is for the nobel purpose of making 'fair-use' backups of their own legally owned DVD's :)

On the other hand I suspect that some members who are captureing from older analog sources or who are involved in the creation and editing of original content would find a "I-Frame AVI codec" and associated 'smart re-render' feature in HC to be INCREDIBLY usefull.

If Hank is not interested in this even for his 'blue sky' future features list, then maybe I should toss it over to NIC for inclusion in QuEnc (since he has worked with the Xvid binaries and should have some experence with AVI codec code).

In any case I would hope that regardless of who creates the I-Frame AVI codec, that HC would support a 'smart re-render' mode with that codec as the source (in addition to the current AVS source).

A good quality "I-Frame AVI codec" would actually fill TWO important needs, because most of the decent MJPEG codecs cost money, so we would not only get 'optimal re-encode' to mpeg, but some might users might want to use it as a jumping off point for XVID or DIVX encodes (as an alternative to MJPEG).

If the DCT and Quantization Coefficients are flexibly defined in the proposed “I-Frame AVI” container format, then the same files could be used for optimal 'minimum-loss' transcoding to MPEG4 as well as MPEG2 by simply selecting suitable I-Frame codec settings before capture.

Just a thought . . .

[Revgen]

I have an idea for a unique feature that I think would set HC apart from ALL other MPEG2 encoders...


Do you think that all of these "smart" features will turn out a DVD compliant MPEG2 stream?

If so, then I would be interested in seeing it done.

[Delphin]

Do you think that all of these "smart" features will turn out a DVD compliant MPEG2 stream?

If so, then I would be interested in seeing it done.

The idea is NOT to have an AVI codec which DIRECTLY encodes a MPEG compliant stream, just use this as a good INTERMEDIATE format (like the current AVS signpost file input) when doing ANALOG CAPTURE.

For any of a number of reasons, trying to encode a compliant MPEG directly into an AVI container would be a problem, but that’s not what I am talking about.

The idea it to have a AVI CODEC which captures nothing but I-Frames (in a DVD compliant format) right from the start, then later when the HC encoder does motion search and bit budgeting, it would decide which frames need to be I-Frames and which frames need to be dropped and replaced by B or P frames. Then the I-Frames at least can simply be taken from the already encoded stream.

The B and P frames in all MPEG streams (including DVD compliant ones) reference back to the I-frames ultimately, so avoiding a re-encode of these frames should improve quality for the whole MPEG stream by using I-frames which were themselves directly encoded from the ORIGINAL VIDEO SOURCE (during the AVI capture operation).

Of course the frame order gets juggled in assembling a GOP but that would all be taken care of by HC during it’s post processing compression phase (just as it is now).

So all of the really critical formatting would be taken care of in ‘non-real time’ as part of the MPEG2 encoding step (again, just as HC does it now), so the ‘DVD compliance’ of the stream would ONLY be dependant on the care with which HC assembles the final M2V elementary stream (not the nasty old Microsoft AVI format of the captured I-Frames:)

So the short answer to your question is that it should have NO IMPACT on the "DVD compliance" of the stream. The flexibility is still there to do whatever is needed to insure that full GOP structure is correct and that the right bits go in the right places in the final output M2V file.

[Revgen]

...So the short answer to your question is that it should have NO IMPACT on the "DVD compliance" of the stream. The flexibility is still there to do whatever is needed to insure that full GOP structure is correct and that the right bits go in the right places in the final output M2V file.

Sounds good to me. :)

As long as the whole complete process results in a playable DVD with great quality, then I'm all for it.

[LigH]

I remember another MPEG2 I-frame AVI codec - made by Etymonix.

And you might probably even get ffdshow's VfW interface to write MPEG frames into AVI output...

[Delphin]

I remember another MPEG2 I-frame AVI codec - made by Etymonix.

And you might probably even get ffdshow's VfW interface to write MPEG frames into AVI output...

Yes thanks for reminding me, I had stumbled upon that link before but lost it.

http://www.justdosoft.com/Etymonix-MPEG2-Video-Codec-15066.html

The reason I wasn't more enthusiastic about this particular "I-Frame" codec is that the performance they are quoting is no better than I get with the FREE HUFFYUV codec if I let it do color space conversion to YUV.

This is because they apparently are not Truncating the Quantitization of the I-Frames at all, so we only see a SMALL amount of compression due to the DCT, Runlength, and Huffmen coding (pretty much the same as Huffyuv which is free and also does Huffmen coding). This is like saving a JPEG at 100 percent quality (which gives GREAT quality, but eliminates most of the advantages of the compression).

Also it costs 50 dollars which is the same problem which afflicts most of the MJPEG codecs out there (too damn expensive for what it does).

Maybe I'm a little cheap, but paying 50 bucks for a MJPEG Codec (or this Codec) seems a little strange when HC, QeEnc, Xvid and some versions of DivX are Free.

Kind of like living in a world where a 12 course Banquite with Lobster, Filet Mignon (and Baked Alaska for desert) is FREE but a BIG MAC cost 50 bucks. :rolleyes:

The I-Frame codec I am talking about would apply the I-Frame COMPRESSION appropriate to the data rate and quantitization that is to be applied to the final IBP frame encoded file by HC in it's last pass, giving HC a source of already properly compressed I-Frames on the one hand, and saving a huge amount of disk space on the other hand.

This would be MUCH more disk space efficient while remaining essentially 'lossless' in a 'no additional degradation to the I-Frame' sense verses a true lossless codec like HUFFYUV or the Etymonix codec you pointed out.

Let's lay it out . . .

Current lossless capture:

1. YUV color space Capture 100% quality (minus minor color space conversion hit)

2. Huffyuv codec giving huge intermeadiate file

3. HC encodes compressed I-Frames and B,P frames (which point to these frames) based on the LOSSLESS Huffyuv AVI file.

Result: Very good quality with HUGE intermediate file.


Current LOSSY CAPTURE:

1. YUV color space Capture 100% quality (minus minor color space conversion hit)

2. MJPEG codec giving barely manageable file (still very large because quality must be set very high to avoid messing with later MPEG2 encode).

3. HC encodes compressed I-Frames from already compressed MJPEG frames, B and P frames also point to these doubly compressed frames.


Proposed 'Quasi-Lossless' I-Frame CODEC Capture:

1. YUV color space Capture 100% quality (minus minor color space conversion hit)

2. AVI "I-Frame CODEC" pre-encodes properly compressed I-Frames with compatible DCT quantitazition based upon the HC profile to be applied in the post processing MPEG2 encoding passs. This will give a smaller intermediate file even than MJPEG because the 'quality' level of the encoding does not need to be set to a very high 'overkill' setting like MJPEG to avoid degrading the final output, but is rather set to EXACTLY the value used in the final HC encode.

3. HC uses the I-Frame which were pre-encoded straight from the uncompressed YUV capture device to encode the MPEG2 file.

RESULT: Smaller files than practical MJPEG encoding, with about THE SAME QUALITY AS HUFFYUV (because all I-Frames used in the final MPEG2 encode were pre-encoded losslessly directly from the ORIGINAL capture source).

There are some implications to doing the motion serch on the compressed I-Frames vs. the uncompressed YUV converted sources but overall I think we are talking about SMALLER CAPTURE FILES THAN MJPEG, WITH QUALITY ROUGHLY EQUIVILENT TO HUFFYUV LOSSLESS.

[winny]

Has anyone seen an issue where HC encoder shows the "input statistics" line counting down from 10 to 0 and says "ERROR, can not process AVS file"?

Because the file is not processed it causes dvd rebuilder to report an expected m2v file is missing.

please forgive the double post as this was originally raised in the rebuilder forum, however the problem may be more relevant here.

rebuilder thread:-
http://forum.doom9.org/showthread.php?p=684630#post684630

i have only seen this issue on one disk out of many, the rest have worked flawlessly with HC.

Many thanks,
Winny

[vladislav67]

Hahk , in 3 pass will better encoding..
Will is added in new version 3 pass?
Otherwise your coder is very, very good .

Best regards ,
Vlad

[dragongodz]

Hahk , in 3 pass will better encoding..
please go to page 5 of this thread and read from there.