Multithreaded XviD - official thread

[Gromozeka]

Happy new 2009 year
People, how can add in XviD new 1vbr method from this:
http://forum.doom9.org/showthread.php?p=1230473#post1230473
Its give best quality and speed

[cweb]

Happy new 2009 year
People, how can add in XviD new 1vbr method from this:
http://forum.doom9.org/showthread.php?p=1230473#post1230473
Its give best quality and speed
trellis? XviD already has trellis... happy new year

[Gromozeka]

trellis? XviD already has trellis... happy new year
NOT! 1 passVBR (fast 2pass VBR) :)

[cweb]

trellis? XviD already has trellis... happy new year
NOT! 1 passVBR (fast 2pass VBR) :)
eh ok I see what you mean 1 pass VBR trellis then...

[alexins]

XviD-1.3.0 (CVS)-090528.06.15-VAQ-MTK x86/x64 (VAQ. MTK, DivX profiles) (not stable) (http://www.xvidvideo.ru/content/view/770/1/)

Changes:
improved precision and rounding for RGB->YV12 conversion;
attempt at fixing a RGB24 access violation;
Added Darkshikari's variance masking as an option to lumimasking;
GUI for variance masking.

Building in the Intel (R) C + + Compiler for Windows 10.1.030

[LigH]

Hail alexins! :D

Now it's our task to test it thoroughly before it can be called stable?

[hajj_3]

what exactly is "Added Darkshikari's variance masking as an option to lumimasking" ?

are there any multithreaded speed improvements with 1.3.0 over 1.2.1?

[juGGaKNot]

XviD-1.3.0 (CVS)-090528.06.15-VAQ-MTK x86/x64 (VAQ. MTK, DivX profiles) (http://www.xvidvideo.ru/content/view/770/1/)

Changes:
improved precision and rounding for RGB->YV12 conversion;
attempt at fixing a RGB24 access violation;
Added Darkshikari's variance masking as an option to lumimasking;
GUI for variance masking.

Building in the Intel (R) C + + Compiler for Windows 10.1.030

what exactly is "Added Darkshikari's variance masking as an option to lumimasking" ?

are there any multithreaded speed improvements with 1.3.0 over 1.2.1?

Is it somewhat stable ? i want to replace 1.2.1

[alexins]

XviD-1.3.0 (CVS)-090529.07.01-VAQ-MTK x86/x64 (VAQ. MTK, DivX profiles) (not stable) (http://www.xvidvideo.ru/content/view/773/1/)

Bugfix:
Added missing resync marker range check in decoder.c;
return E_FAIL instead of S_FALSE upon XVID_ERR_MEMORY error in dshow frontend

Building in the Intel (R) C + + Compiler for Windows 10.1.030

[alexins]

XviD-1.2.2 x86 / x64 Stable release (http://www.xvidvideo.ru/content/view/774/1/)

This release is Xvid 1.2.2 bugfix release. It is API compatible with the previous 1.2.1 stable release. This release contains important, security-related fixes. Update is highly recommended.

Changes since 1.2.1:

xvidcore library


Workaround for nasm bug with Mach-O/OSX target
Fix for missing resync marker range check (reported by IBM X-Force. Thanks go to John McDonald and Christopher Valasek)
Improved precision for RGB<->YUV conversions
Fix for potential RGB24 access violation
Updated compiler options for Apple PPC target
Fixed MSVC6 projects to work for path names with spaces

VFW frontend

Updated mingw makefile

DShow frontend

Bugfix for wrong handling of xvidcore XVID_ERR_MEMORY return code (reported by IBM X-Force. Thanks to John McDonald and Mark Dowd)
____________________________________
Building in the Intel (R) C + + Compiler for Windows 10.1.030

[hajj_3]

great, i'll wait for koepi to add it to his site.

*begs darksharki to improve the speed on multi-core cpu's* :P

[henryho_hk]

Oops... VAQ is not official yet....

[Lugia25000]

@alexins

I have a question about the Encraw in your builds. Its a fps rate detection from the input file possible?
The input file is 23.976 and the output file show 25.00 in Mediainfo.
And a change from the High and the Width in the output file?
Width and high are interchanged in the display from Mediainfo and Windows, but high and width are correct.

Thanks for alle improvements to Xvid.

[kypec]

@alexins

I have a question about the Encraw in your builds. Its a fps rate detection from the input file possible?
The input file is 23.976 and the output file show 25.00 in Mediainfo.
And a change from the High and the Width in the output file?
Width and high are interchanged in the display from Mediainfo and Windows, but high and width are correct.

Thanks for alle improvements to Xvid.

+1 to all these requests. :goodpost:
Alexins, you can look at the source code of xvid_encraw compiled by squid_80 (http://forum.doom9.org/member.php?u=67865), he has fixed the bug with swapped WxH values. Actually, he completely fixed the AVI output saving routines and also the calculation of encoding speed -> it shows true speed including processing of input frames/encoding/outputting instead of raw Xvid encode time spent on single frames which is very misleading for real life purposes.

@Lugia2500: you have to specify output framerate manually with -framerate switch. Otherwise it will be set to default 25fps (PAL).Rate control options:
-framerate float : target framerate (25.0)

[Barough]

I found some time to compile the new Xvid release 1.2.2. A VAQ-patched version is still missing though. Since the new version fixes two potential security vulnerabilities it is advised to install it ASAP!


http://www.koepi.info/Xvid-1.2.2-07062009.exe

[jsquare]

http://www.koepi.info/Xvid-1.2.2-07062009.exe
Why bother with Koepi's version?
Alexins builds works very well, also have the VAQ option and a bunch of profiles.

[olnima]

Why bother with Koepi's version?
Alexins builds works very well, also have the VAQ option and a bunch of profiles.

...because in my tests koepis versions were a bit faster...


Olnima (...waiting for koepis 1.22 VAQ-Version...)

[juGGaKNot]

@Lugia2500: you have to specify output framerate manually with -framerate switch. Otherwise it will be set to default 25fps (PAL).Rate control options:
-framerate float : target framerate (25.0)


Yes, i forgot to add to to the second pass but i've added it to the mux and info says original fps 25 and fps 30

Was scared for a minute there :)

[Asrial]

Can someone explain, in easier terms, what the vulnerabilities in XviD allow?

Someone could make a malicious AVI that runs code on someone's computer?

[henryho_hk]

Yes, i forgot to add to to the second pass but i've added it to the mux and info says original fps 25 and fps 30

Was scared for a minute there :)

Or use squid80's much improved version of xvid_encraw

[juGGaKNot]

Or use squid80's much improved version of xvid_encraw

I have the latest one from megui, is it not squid80's build ?

[Sharktooth]

yep. megui comes with squid's enhanced xvid_encraw.

[squid_80]

Alexins, you can look at the source code of xvid_encraw compiled by squid_80 (http://forum.doom9.org/member.php?u=67865), he has fixed the bug with swapped WxH values.
That was actually my fault in the first place, it was adopted into xvid's CVS before I found and fixed it.

[christinedawson]

WOW!
Does this technology applicable to AVC? Is it possible to make same for x264?

[buzzqw]

x264 is already SMP able, a lot able

BHH

[alexins]

XviD.1.3.0-(CVS).100814.02.44-VAQ-(MTK, DivX profiles) x86/x64 (http://www.xvidvideo.ru/xvid-video-codec/xvid-1-3-0-cvs-100814-02-44-vaq-mtk-x86-x64.html)

changes:

skip mv_bits assert in _DEBUG mode;
app-level multi-threading for xvid_encraw;
typo with sequence splitting;
fixed multithreaded AVI input (hopefully);
fixed rounding issue for app-level multi-threading;
API change: signal fourcc to xvidcore;
decoder: better distinguish between xvid and non-xvid streams.


Microsoft Visual C++ 2008 SP1, Microsoft Windows SDK for Windows 7.1, DirectX SDK v.9.29.1962, Intel® C++ Compiler Professional Edition for Windows 11.1.065

[laserfan]

XviD.1.3.0-(CVS).100814.02.44-VAQ-(MTK, DivX profiles) x86/x64 (http://www.xvidvideo.ru/xvid-video-codec/xvid-1-3-0-cvs-100814-02-44-vaq-mtk-x86-x64.html)What is the difference between the 1.3.0 in this link, and the 1.3.0 in your signature? :confused:

[Sharktooth]

none...

[Penecho]

Hey guys, i tried several Xvid builds now, but none seems to use the full CPU power...

Does any1 have a recommendation for me, which xvid build to use?

I have a core i7@3,8Ghz and running Win7 x64...


Cu



Penecho

[Sharktooth]

xvid multithreading is not optimal. your CPU will never get fully used.
you should start using x264 instead. it has a way better threading model and obviously a much better compression/quality.

[LigH]

Nevertheless - nice to see that there is still *any* development around Xvid... ;)

[Gser]

Nevertheless - nice to see that there is still *any* development around Xvid... ;)

It should be stopped. And people should stop using xvid.

[olnima]

It should be stopped. And people should stop using xvid.

???

Should any project be stopped which user 'Gser' does not need?
What about viewing videos via DVD-player? :devil: :devil: :devil:

I use xvid and I will definately continue so, as long as x264-support is not available or much more expensive in DVD-players, so please stop talking such a nonsense.

[cweb]

It should be stopped. And people should stop using xvid.

For your information, nobody can stop XviD (it's XviD by the way not xvid).
It's an open source project and anyone is free to continue it and to keep copying it.

[Didée]

(it's XviD by the way not xvid).
The spelling XviD (capital "D") is outdated. The notation has officially been changed to Xvid. Check www.xvid.org. ;)

[cweb]

The spelling XviD (capital "D") is outdated. The notation has officially been changed to Xvid. Check www.xvid.org. ;)

I see. So the thread title should be appropriately changed too.

[Gser]

???

Should any project be stopped which user 'Gser' does not need?
What about viewing videos via DVD-player? :devil: :devil: :devil:

I use xvid and I will definately continue so, as long as x264-support is not available or much more expensive in DVD-players, so please stop talking such a nonsense.

Stop using a dvd-player. The xbox360, ps3 and blu-ray players are not expensive.

[cweb]

Stop using a dvd-player. The xbox360, ps3 and blu-ray players are not expensive.
Blu-rays play xvid too..

[Sharktooth]

... blah... stop using blu-ray and use streaming.
every technology fits in its own place.
you cant just say "it should be stopped"... it will naturally die when some other tech will replace it AND when it will completely adopted.
in the meanwhile its completely legit to continue developing it, especially for bugfixing or multithreading implementation.

[Guest]

Stay on topic please. This thread is about multithreaded xvid, not whether DVD players should be abandoned. Thank you.

[LigH]

IMHO there are good uses for Xvid still - like: Backwards compatibility (not everyone is rich enough to be always able to buy most current hardware); or: ratio of encoding speed, quality and size (e.g. capturing video with a compression smaller than MJPEG and faster than AVC).

[PiPPoNe92]

I can't download latest version of xvid from xvidvideo.ru.
My antivirus detect it like a virus. So, I upped xvid_encraw to virustotal...

http://www.virustotal.com/file-scan/report.html?id=3e4c79d123bedb0a8c1867d1a8fa9c3211360d35264b3e4ad006c31e2db9218a-1283337253

[alexins]

I can't download latest version of xvid from xvidvideo.ru.
My antivirus detect it like a virus. So, I upped xvid_encraw to virustotal...

http://www.virustotal.com/file-scan/report.html?id=3e4c79d123bedb0a8c1867d1a8fa9c3211360d35264b3e4ad006c31e2db9218a-1283337253

This is a false positive.

[PiPPoNe92]

ok, thanks. Another question: when I encode with last xvid_encraw with this command line:

xvid_encraw -progress -i "TESTNNDM2 - HQ --FILM.avs" -bitrate 363 -stats -max_bframes 1 -vhqmode 4 -full1pass -pass1 "pass.stats" -bvhq -bquant_ratio 162 -bquant_offset 0 -threads 6 -type 2 -avi "First_pass.avi"


xvid_encraw -progress -i "TESTNNDM2 - HQ --FILM.avs" -bitrate 363 -stats -chigh 30 -clow 15 -max_bframes 1 -vhqmode 4 -pass2 "pass.stats" -bvhq -bquant_ratio 162 -bquant_offset 0 -threads 6 -type 2 -avi "Movie_2pass.avi"

the final .avi that I have got, has like fourCC mp4v, not XVID, why? With previous version I obtain a .avi XVID. Now not. How can I do?

[LigH]

If xvid_encraw has no undocumented switch for this purpose (at least I did not see one in the "-help" output), you can at least use the "FourCC Changer" to fix it...

But I wonder how this happened to you. The "xvid_encraw.exe" installed from "XviD.1.3.0-(CVS).100814.02.44-VAQ-MTK-icl11_x86.exe" from xvidvideo.ru (204800 bytes, not EXE-compressed) does not even contain a string "mp4v".

[PiPPoNe92]

yes, I tried to use fourcc changer, but doesn't work, because It don't recognize the fourcc mp4v.

[LigH]

Incredible ... the FourCC Changer I know ("Nic's Mini AviC FourCC changer" - included in Koepi's Xvid as well as the one from XvidVideo.RU) simply overwrites the FourCC. No matter if knows the previous FourCC.

It just makes the mistake to offer upper case FourCCs for both, although one of them should be lower case. But you can type them manually.

So - which program tells you that this AVI has a FourCC of "mp4v"?

[PiPPoNe92]

vlc...

[LigH]

Well - an AVI should have two FourCCs for the video. Looking into the AVI header of the file I just created (using a hex editor), I see "xvid" right after the "vids" chunk, and "XVID" a bit after "strf". As it should be. My VLC 1.1.4a displays "Codec: MPEG-4 Video (XVID)" in the Codec Details tab (Ctrl+J).

[PiPPoNe92]

I see with VLC 1.1.3 "Codec: MPEG-4 Video (mp4v)" in the Codec Details tab.

with smplayer I see:

Demuxer: mpeg4es
Codec: ffodivx