XviD 1.1.2 bugfix release

[unskinnyboy]

From the devel mailing list..

-----Original Message-----
From: xvid-devel-bounces@xvid.org [mailto:xvid-devel-bounces@xvid.org] On Behalf Of Michael Militzer
Sent: Tuesday, October 31, 2006 3:52 PM
To: xvid-devel@xvid.org
Subject: [XviD-devel] Xvid 1.1.1 is available!

Hello!

This is Xvid 1.1.1 release.

This release is Xvid 1.1.1 bugfix release. It is API compatible with the previous 1.1.0 stable release.

Changes since 1.1.0:

* xvidcore library
- Fixed bug when frame-drop (N-VOP) feature is used in combination with packed B-frames
- Fix for premature EOF in xvid_decraw example
- Fixed potential crash on Intel EMT64 architecture
- Several fixes for IA64 platform (patch by Thomas Koeckerbauer)
- Fix for visual_object_verid vs. video_object_layer_verid problem
- Ensure intervening bytes are preserved in BitstreamInit()

* VFW frontend
- Prevent segfault when encoding application calls compress_end with NULL
codec context (PerfectDark at yandex dot ru)
- Profile definitions updates

* DShow frontend
- Add missing debug.c

The files are available in the download section of Xvid.org:
http://www.xvid.org/downloads.html

-- The "Xvid Team"

[imcold]

new version + new website ^_^ many thanks xvid people :)

[Inventive Software]

That website was a long long time coming... I almost lost faith in the XviD team after they said "The new website should be up next week", in May! :D But I'm gonna have a compile and a play all the same.

[celtic_druid]

http://home.wildit.net.au/celticdruid/mirror/XviD/XviD-1.1.2.exe
http://home.wildit.net.au/celticdruid/mirror/XviD/xvid-1.1.2.7z

[prOnorama]

Cheers, I'll wait until 1.1.1 is on Koepi's site (the official link from XviD.org still shows 1.1.0)

[IgorC]

Good news.

Koepi's compilation should not differ from Celtic's. Even Koepi's compilations aren't free of bug totally. Like in case with uninstal of all codecs from system.
Celtic's compilations are REALLY well knowed. In big mayority they are stable.

[celtic_druid]

Well usually I stick with providing svn/cvs builds so there are usually plenty of bugs, just hopefully not caused by anything that I did.

Koepi has definatly become synonymous with XviD binaries though. Speaking of the installer bug. The 1.1 build on afterdawn from recollection is effected so there could be some major instances as people uninstall their 1.1 builds to install 1.1.1. Very easy mistake to make in inno setup though (setting it to delete an entire key instead of just the one value).

[clsid]

Have these same fixes been applied to the 1.2 branch as well? If so, any chance of a new 1.2 build?

[henryho_hk]

~_~ .... I want multi-thread ~~~~

[celtic_druid]

You want a 1.2 cvs build then. New one up by the way, but this isn't a thread for discussing 1.2 builds.

[clsid]

Xvid 1.1.2 released
This release is Xvid 1.1.2 bugfix release. It fixes and replaces previous Xvid 1.1.1 release.

Changes since 1.1.1:

Fixed typo in xvid.c
Updated bitstream.h to fix the previous faulty non-aligned bitstream buffers fix

[Koepi]

...and I finally have binaries available, too :)

Cheers
Koepi

[Inventive Software]

2 compiles in 6 hours. GCC's been busy on my system. :D

[unskinnyboy]

Xvid 1.1.2 released
This release is Xvid 1.1.2 bugfix release. It fixes and replaces previous Xvid 1.1.1 release.

Changes since 1.1.1:

Fixed typo in xvid.c
Updated bitstream.h to fix the previous faulty non-aligned bitstream buffers fix
There is one more change to xvidcore which you missed:

Added a test_bits bench to xvid_bench.c to test for non-aligned bitstream buffers.skal added that^^

P.S: Can a MoD change the title of the thread to 'XviD 1.1.2 bugfix release'?

[bob0r]

celtic_druid, Koepi:
Mind sharing your install script(s) with me?

[KornX]

hi folks, especially koepi,
what about smp support?
went the code from that release in?

(Latest unstable binary
XviD-1.2.-127-25022006.exe (630kb)
Changelog to XviD-1.1:
- {core}: New experimental SMP support.
- {core}: Trellis improvements (according to sysKin).
- On uniprocessor machines set number of threads to 0!)

KornX

[clsid]

SMP support is not included in 1.1.x, but only in 1.2.x (= current CVS).

[EEPROM]

Excused the ignorance, but if during encoding the cpu(AMD 3800X2) works to 100%, does not mean that codec included smp support?
Because i've tested the new build and cpu has worked at 100%
http://forum.doom9.org/showpost.php?p=895038&postcount=4232

[henryho_hk]

Not if you are using new version of Virtualdub, which seems to be using separate threads for reading (i.e., incl. avisynth processing) and encoding.

[EEPROM]

If I have not understood badly,my CPU work at 100% with AutoGK 2.36b(VirtualDubMod) but with new version of VirtualDub or another program, my cpu Work at 50%?

[numaios]

By the way, does anybody know why the name XviD changed to Xvid with this new build?

[Liisachan]

After updating XviD, the histogram doesn't show anymore in the Status window. This never happened before. What could be the reason?
http://ffdshow.faireal.net/tmp/xvid.png
celtic_druid's 1.1.2 on Win2K...

[MatMaul]

I have a "similar" problem with xvid_encraw, the fps, bitrate and evolution of the encode are not reported in xvid_encraw (celtic_druid binaries).
I have made my own build with gcc and no problem.

You can download my build here (http://www.etud.insa-toulouse.fr/~mvelten/xvid-1.1.2.zip), it doesn't include the directshow filter.

[Liisachan]

Koepi's 1.1.2 works ok. Strange...

[kurt]

use a gcc version of xvidcore.dll and everything works fine. I also reported this issue: http://celticdruid.no-ip.com/phpBB2/viewtopic.php?t=206
(look at latest post)

[Liisachan]

Thanls MatMaul. and thanks kurt, your last post there is really informative! I'll try the gcc version then later (I'm currently encoding something, and so cannot over-install files right now)

EDIT:
This should be something one would consider obvious, but the result by 1.1.0 and the result 1.1.2 are (almost?) identical. I just wanted to check that with my own eyes, using a longer sample, and they are.

EDIT2:
Yes! xvid-1.1.2gcc works ok :D

[Taurus]

...and I finally have binaries available, too :)

Cheers
Koepi
Thanks Koepi (and of course celtic_druid) for sharing your compiles with us.

The slowest Download ever: 280 Byte/sec on DSL line from Koepi's Website:D

Cheers
Taurus

[Wesmosis]

Thanx

[squid_80]

The experimental SSE2 GMC code can cause problems if gcc isn't used to compile, but it was left out of 1.1.2.

[celtic_druid]

The problem is caused by a combo of ICL9.1 and MSVC6. Happend in the past with a 1.2cvs build, so I switched to MSVC7.1, but forgot for the last builds (I don't generally use the status window). I'll run some tests later (gotta go out to lunch) and replace the builds. GMC will still be a problem though for the 1.2 build.

[Romario]

What's going on with Xvid 1.2 SMP builds, can somebody tells me something new?

Etc, can Xvid 1.2 speed-up encoding process, and boost quality on lower bitrates?

[SeeMoreDigital]

What's going on with Xvid 1.2 SMP builds, can somebody tells me something new? Why not start a new 1.2.x thread and ask their?

[SeeMoreDigital]

http://home.wildit.net.au/celticdruid/mirror/XviD/xvid-1.1.1.exe
http://home.wildit.net.au/celticdruid/mirror/XviD/xvid-1.1.1.7zHi CD,

Unfortunately MPEG4 Modifier reports "Invalid VOL" and GSpot (latest version) reports: -

http://img140.imageshack.us/img140/590/gspotzi6.png


Cheers

[celtic_druid]

Don't use the 1.1.1 builds. Use 1.1.2 instead which works fine. I'll replace the links in this thread now.

[Liisachan]

Celtic Druid's 1.1.2 (revised)
http://ffdshow.faireal.net/mirror/XviD/XviD-1.1.2.exe

And if you need to use GMC...
http://ffdshow.faireal.net/mirror/XviD/gcc/xvid-1.1.2gcc.7z

[SeeMoreDigital]

Don't use the 1.1.1 builds. Use 1.1.2 instead which works fine. I'll replace the links in this thread now.Thanks...

A few months ago you created a special "Frame Drop" Xvid compile (http://82.10.220.174/Uploaded_Files/Doom9_Forum_files/XviDCore.dll_Frame_Drop_Build_by_Celtic_Druid.7z). Any chance of doing the same again?


Cheers

[celtic_druid]

GMC shouldn't be an issue unless you use a 1.2cvs build, then you need the gcc compile for working GMC with SSE2.

Refresh my memory. What was special about it? If it was the one that didn't do N-VOP's when the ratio was set to 0. According to the changelog, XviD 1.1.2 doesn't do N-VOP's if packed bitstream is enabled. Also my MTK builds don't do N-VOP's for HT profiles, even if bframes are disabled.

[SeeMoreDigital]

Refresh my memory. What was special about it? If it was the one that didn't do N-VOP's when the ratio was set to 0. According to the changelog, XviD 1.1.2 doesn't do N-VOP's if packed bitstream is enabled. Also my MTK builds don't do N-VOP's for HT profiles, even if bframes are disabled.Yes that's the one...

However, with the new builds (yours or Koepi's) I get this: -

http://img168.imageshack.us/img168/5458/framedrop0mf0.png

With your special build, I got nothing but I-VOPs and P-VOPs ;)


Cheers

[celtic_druid]

You need to use the 1.2cvs+MTK build and enable Home Theatre or one of the other DXN equiv type profiles. Should be no N-VOP's then. Which is as it should be since DivX doesn't create them.

With 1.1.2 with the above (bframes disabled) I think there is no way to disable N-VOP's and if I went changing things, then it wouldn't be 1.1.2.

[SeeMoreDigital]

With 1.1.2 with the above (bframes disabled) I think there is no way to disable N-VOP's and if I went changing things, then it wouldn't be 1.1.2.Bummer.....

That's a shame. How did you manage to do it before?

[celtic_druid]

If I recall, I modified the core so that a ratio of 0 really did disable N-VOP's. Anyone who wants to encode without N-VOP's, I would suggest a 1.2cvs+MTK build with a HT profile.

If it is really important for you to encode with 1.1.x and not have N-VOP's I could do another build. Just won't be marked 1.1.2.

[SeeMoreDigital]

If it is really important for you to encode with 1.1.x and not have N-VOP's I could do another build. Just won't be marked 1.1.2.For my own very selfish requirements I would like such a build.... It's really useful for generating my test cards ;)

[numaios]

I don't know if this has something to do with what you are talking about. Encoding a movie with XviD 1.1.0 (Koepi's build):
Packed bitstream: No
QPel: No
GMC: No
Interlaced: No
Aspect ratio: Square pixels
Quant type: H.263

I-VOPs: 2422 (1,25%)
P-VOPs: 97238 (50,33%)
B-VOPs: 93534 (48,41%)
S-VOPs: 0 (0,00%)
N-VOPs: 0 (0,00%)

Max consecutive B-VOPs: 1

With XviD 1.1.2 (Koepi's build):
Packed bitstream: No
QPel: No
GMC: No
Interlaced: No
Aspect ratio: Square pixels
Quant type: H.263

I-VOPs: 2519 (1,30%)
P-VOPs: 97239 (50,33%)
B-VOPs: 93435 (48,36%)
S-VOPs: 0 (0,00%)
N-VOPs: 1 (0,00%)

Max consecutive B-VOPs: 1

With XviD 1.2 (included in AutoGK 2.36)

Packed bitstream: No
QPel: No
GMC: No
Interlaced: No
Aspect ratio: Square pixels
Quant type: MPEG Custom

Custom intra matrix:
8 9 10 11 12 14 16 18
9 11 12 13 15 16 18 20
10 12 14 16 17 19 21 22
11 13 16 18 20 22 24 26
12 15 17 20 23 25 28 30
14 16 19 22 25 29 34 38
16 18 21 24 28 34 46 52
18 20 22 26 30 38 52 72

Custom inter matrix:
16 16 17 17 18 19 21 23
16 17 17 18 19 20 22 24
17 17 19 20 21 23 24 27
17 18 20 22 24 25 28 32
18 19 21 24 26 29 36 40
19 20 23 25 29 40 48 52
21 22 24 28 36 48 60 64
23 24 27 32 40 52 64 106

I-VOPs: 2530 (1,31%)
P-VOPs: 97291 (50,36%)
B-VOPs: 93372 (48,33%)
S-VOPs: 0 (0,00%)
N-VOPs: 1 (0,00%)

Max consecutive B-VOPs: 1

Why do I get one N-VOP with XviD 1.1.2 and 1.2? What can I do to avoid that?

Thanks.

[SeeMoreDigital]

I don't think it has anything to do with what I'm talking about. Also, in my case I'm generating encodes "without" B-VOPs....


Cheers

[numaios]

Ok, I see...

Have you got any idea of why I get 1 N-VOP with some versions and none N-VOP with other versions?

[SeeMoreDigital]

Ok, I see...

Have you got any idea of why I get 1 N-VOP with some versions and none N-VOP with other versions?Sadly... I don't. Sorry

[celtic_druid]

With the 1.2 build from AutoGK, as I said earlier if you use a Home Theatre profile, then you shouldn't get any N-VOP's. I did that to make sure that the ESS option in AutoGK was actually compatible with ESS based players since I believe if a movie was very compressible AutoGK would disable bframes which will most likely cause N-VOP's.

As for your N-VOP. v1.1.0 I don't think generated N-VOP's with bframes enabled at all. Do you have packed bitstream enabled?

I really can't see a problem with N-VOP's personally. Any compliant decoder can handle them. Anything that can't is broken.

[Liisachan]

I used celtic_druid's 1.1.0 to encode many things; it generated a few N-vops (cartoon mode is enabled), but I have no reason to dislike that at all. The file plays everywhere afaik and timing is frame accurate.
What's wrong about N anyway? I'm not one of them but there are even a lot of ppl who do quasi-vfr 120 fps with a ton of null frames.

[celtic_druid]

http://celticdruid.no-ip.com/test/xvidcore-1.1.x.n-vop.7z

Anyone else noticed the effect of bitrate and quant matrix on the number of N-VOP's? 100 identical frames = 1 I-VOP, 99 N-VOP's with h.263/Q2. h.263/some bitrate = 95 N-VOP's, CQM/Q2 = 66 N-VOP's & CQM/some bitrate = 5 N-VOP's. Above build = 99 P-VOP's, 0 N-VOP's.

[squid_80]

The decision to code as N_VOP is based on whether macroblocks are skipped or not. More bitrate/better matrix = less blocks skipped = less N-VOPs.

[SeeMoreDigital]

Many, many thanks CD :D

Just tried your build and it appears to be working fine. Should anybody else be interested (apart from me). Here are some sample Test_Cards_(with_N_&_P_VOPs) (http://82.10.220.174/Uploaded_Files/Doom9_Forum_files/Test Cards with N-VOPs and P-VOPs).7z).


Cheers

[numaios]

:thanks: As for your N-VOP. v1.1.0 I don't think generated N-VOP's with bframes enabled at all. Do you have packed bitstream enabled?I didn't get any N-VOP at all with v1.1.0 (only with v1.1.2), and packed bistream was disabled:
Packed bitstream: No
QPel: No
GMC: No
Interlaced: No
Aspect ratio: Square pixels
Quant type: H.263

I-VOPs: 2422 (1,25%)
P-VOPs: 97238 (50,33%)
B-VOPs: 93534 (48,41%)
S-VOPs: 0 (0,00%)
N-VOPs: 0 (0,00%)

Max consecutive B-VOPs: 1

I really can't see a problem with N-VOP's personally. Any compliant decoder can handle them. Anything that can't is broken.An N-VOP means there is no frame? Why do they appear if packed bistream is enabled/disappear when disabled?

Anyone else noticed the effect of bitrate and quant matrix on the number of N-VOP's? 100 identical frames = 1 I-VOP, 99 N-VOP's with h.263/Q2.I see... Precisely, that N-VOP was inserted in a long and completely black scene...

With the 1.2 build from AutoGK, as I said earlier if you use a Home Theatre profile, then you shouldn't get any N-VOP's. I did that to make sure that the ESS option in AutoGK was actually compatible with ESS based players since I believe if a movie was very compressible AutoGK would disable bframes which will most likely cause N-VOP's.
OK, now I understad. And using a HT profile with 1.1.2 would avoid N-VOPs too?

[celtic_druid]

So you get N-VOP's when packed bitstream is enabled? Because the HT profiles forced packed bitstream so if that was the case then definatly using it in 1.1.2 would not avoid N-VOP's.

Anyone who 100% wants to avoid N-VOP's should either use my 1.2cvsMTK build with a HT profile or the above xvidcore-1.1.x.n-vop build with the drop ratio set to 0.

HT and other DXN equiv profiles encode without N-VOP's in my MTK builds because I modified it to override the drop settings in such a case. You could set the ratio to 100 and still get no N-VOP's. Regular XviD builds don't have this. With the 1.1 build, I simply modified the core so that a ratio of 0 actually disables N-VOP's.

[numaios]

I didn't get any N-VOP at all with v1.1.0 (only with v1.1.2), and packed bistream was disabled
So you get N-VOP's when packed bitstream is enabled? Because the HT profiles forced packed bitstream so if that was the case then definatly using it in 1.1.2 would not avoid N-VOP's.
No, I get 1 N-VOP with NO packed bistream and with B-frames enabled, Advanced Simple@L5 profile, in version 1.1.2 and in version 1.2 bundled with AGK. As I posted before:

With Koepi's XviD 1.1.0, AS@L5 profile, drop ratio set to 0:
Packed bitstream: No
QPel: No
GMC: No
Interlaced: No
Aspect ratio: Square pixels
Quant type: H.263

I-VOPs: 2422 (1,25%)
P-VOPs: 97238 (50,33%)
B-VOPs: 93534 (48,41%)
S-VOPs: 0 (0,00%)
N-VOPs: 0 (0,00%)

Max consecutive B-VOPs: 1

With Koepi's XviD 1.1.2, AS@L5 profile, drop ratio set to 0:
Packed bitstream: No
QPel: No
GMC: No
Interlaced: No
Aspect ratio: Square pixels
Quant type: H.263

I-VOPs: 2519 (1,30%)
P-VOPs: 97239 (50,33%)
B-VOPs: 93435 (48,36%)
S-VOPs: 0 (0,00%)
N-VOPs: 1 (0,00%)

Max consecutive B-VOPs: 1

With XviD 1.2 bundled with AGK, unrestricted profile, drop ratio set to 0:
Packed bitstream: No
QPel: No
GMC: No
Interlaced: No
Aspect ratio: Square pixels
Quant type: MPEG Custom

I-VOPs: 2530 (1,31%)
P-VOPs: 97291 (50,36%)
B-VOPs: 93372 (48,33%)
S-VOPs: 0 (0,00%)
N-VOPs: 1 (0,00%)

Max consecutive B-VOPs: 1

BTW, what does "CVS" stand for? I always read XviD 1.2 cvs but I haven't found its meaning in any page.

[Terranigma]

BTW, what does "CVS" stand for? I always read XviD 1.2 cvs but I haven't found its meaning in any page.
Concurrent Versions System. :)

[numaios]

Thank you! :)
And it means that the codec gets updated although the version number doesn't change?

[foxyshadis]

In this sense, it basically means "development version", which is sort of pre-alpha usually, as opposed to a beta or final build. 1.2 final is still a ways down the line, even though it's currently pretty stable.

[celtic_druid]

Try 1.1.2 with packed bitstream enabled and 1.2 with a HT profile.

[plugh]

For various reasons I decided I need to do my own build (first time).
I already had Visual Studio 6 + SP5 + processor pack installed on a W2K system, so I downloaded the 1.1.2 zip file.

Followed the 'INSTALL' instructions...

Should I be getting warnings during build?

Compiling...
Command line warning D4002 : ignoring unknown option '/Qipo'
bitstream.c
J:\xvid\xvidcore-1.1.2\src\bitstream\bitstream.c(1387) : warning C4018: '<' : signed/unsigned mismatch
J:\xvid\xvidcore-1.1.2\src\bitstream\bitstream.c(1405) : warning C4244: '=' : conversion from '__int64 ' to 'int ', possible loss of data
J:\xvid\xvidcore-1.1.2\src\bitstream\bitstream.c(1406) : warning C4244: '=' : conversion from '__int64 ' to 'int ', possible loss of data
J:\xvid\xvidcore-1.1.2\src\bitstream\bitstream.c(1407) : warning C4244: '=' : conversion from '__int64 ' to 'int ', possible loss of data
plugin_2pass2.c
J:\xvid\xvidcore-1.1.2\src\plugins\plugin_2pass2.c(468) : warning C4244: 'function' : conversion from 'const double ' to 'const float ', possible loss of data
J:\xvid\xvidcore-1.1.2\src\plugins\plugin_2pass2.c(474) : warning C4244: 'function' : conversion from 'const double ' to 'const float ', possible loss of data
J:\xvid\xvidcore-1.1.2\src\plugins\plugin_2pass2.c(1358) : warning C4244: '=' : conversion from 'unsigned __int64 ' to 'double ', possible loss of data
J:\xvid\xvidcore-1.1.2\src\plugins\plugin_2pass2.c(1541) : warning C4244: 'initializing' : conversion from 'double ' to 'const float ', possible loss of data
plugin_single.c
J:\xvid\xvidcore-1.1.2\src\plugins\plugin_single.c(191) : warning C4244: '=' : conversion from 'double ' to '__int64 ', possible loss of data
encoder.c
J:\xvid\xvidcore-1.1.2\src\encoder.c(848) : warning C4018: '==' : signed/unsigned mismatch
J:\xvid\xvidcore-1.1.2\src\encoder.c(848) : warning C4018: '==' : signed/unsigned mismatch
J:\xvid\xvidcore-1.1.2\src\encoder.c(1557) : warning C4018: '!=' : signed/unsigned mismatch
(edit - found the vfw file :o )

[movmasty]

then, 1.1.2 or 1.2 ?

and, i forgot, have to uninstall the old xvid?

[numaios]

In this sense, it basically means "development version", which is sort of pre-alpha usually, as opposed to a beta or final build. 1.2 final is still a ways down the line, even though it's currently pretty stable.Try 1.1.2 with packed bitstream enabled and 1.2 with a HT profile.:thanks: foxyshadis and celtic_druid

[Nil Einne]

Any chance of getting a 1.1.2 x64 build Celtic or squid_80 or someone?

Also, I know this is OT but how about a x64 from the current CVS (1.2)?

Cheers