reimaginemedia / webm Goto Github PK

I see subtitles as an fundamental part of a media specification, specially
for accessibility purposes, and I don't see them specified, not even
mentioned in the WebM Project site faqs and docs.

While subtitles don't have as many different specifications and
implementations as video and audio formats, and even the Matroska
documentation writers took their time to describe the few of them, I'd like
to see some sort of guarantees if webm-specific media players will show my
embed [optionally multi-language] subtitles or not.

So this is request to the project team to discuss and clarify this
standardization issue of the Subtitle Track, if there is any.

We've have been trying to compile ffmpeg with VP8 here, but run into a problem. 
The compilation 
would fail, because PIC support wasn't had not been added to libvpx-v8 
previously.

We recompiled libvpx-v8 again, previously specifying --enable-pic in the 
configure script, but this 
didn't help.

We then realised that the -fPIC CFLAG was not being added by configure. We 
added it manually to 
the Makefile and finally, after recompiling the library, ffmpeg could 
successfully link to it, and 
compiled correctly.

This is on Ubuntu x86_64 on an EC2 instance with GCC 4.3.3

Directshow filters sources are not available.
It will be usefull to move source code into the WebM Project public repository.

What steps will reproduce the problem?
1. Execute regsvr32 webmsource.dll webmsplit.dll vp8decoder.dll 
vp8encoder.dll webmmux.dll

What is the expected output? What do you see instead?

I expect to see the following for each dll:
DLLRegisterServer in <dllfile> succeeded.

Instead this message appears four times:
Assertion failed!
Program: C:\Windows\SysWOW64\regsvr32.exe
File: .\dllentry.cpp
Line: 82
Expression: SUCCEEDED(hr)

Followed by:
DLLRegisterServer in webmsource.dll succeeded.

Followed by:
[Window Title]
RegSvr32

[Content]
The module "webmsplit.dll" was loaded but the call to DllRegisterServer 
failed with error code 0x80070005.

For more information about this problem, search online using the error code 
as a search term.

[OK]

Followed by:
2 more assertions

[Window Title]
RegSvr32

[Content]
The module "vp8decoder.dll" was loaded but the call to DllRegisterServer 
failed with error code 0x80070005.

For more information about this problem, search online using the error code 
as a search term.

[OK]

2 more assertions

[Window Title]
RegSvr32

[Content]
The module "vp8encoder.dll" was loaded but the call to DllRegisterServer 
failed with error code 0x80070005.

For more information about this problem, search online using the error code 
as a search term.

[OK]

[Window Title]
RegSvr32

[Content]
The module "webmmux.dll" was loaded but the call to DllRegisterServer 
failed with error code 0x80070005.

For more information about this problem, search online using the error code 
as a search term.

[OK]

What version are you using? On what Windows version?
webmdshow-0.9.5.0-20100518
Windows 7 Home Premium 64-bit

The attached patches makes building the vpx project (decoder only) warning-
free on Visual Studio 2008 (with /W3, without /Wp64).

Some upstream Directshow filters may supports only FORMAT_VideoInfo2, so the 
encoder filter must support it, at least on input, for compatibility.

Minimal changes (for input only), would be in vp8encoderinpin.cpp, where 
support for FORMAT_VideoInfo2 must be added.

Information on the VIDEOINFOHEADER2 Structure (which is not cast-compatible 
with VIDEOINFOHEADER):
http://msdn.microsoft.com/en-us/library/dd407326(v=VS.85).aspx

... because in case of something going wrong, this will leave a nasty
8GB-zombie. And the "usual user" will just wonder why the harddisk gets
full so fast

Attached .gitignore works for me.

WebM should have Codecs for subtitles, song lyrics, transcripts, etc. I
propose OggWrit [1].

[1] http://wiki.xiph.org/OggWrit

What is the expected behavior? What do you see instead?

Expect clean compile with no warnings.

What version are you using? On what operating system?

Latest MacOS Snow Leopard with all updates installed.

Can you reproduce using the ivfdec or ivfenc tools? What command line are
you using?


Please provide any additional information below.

    [CC] vp8/encoder/ethreading.c.o
vp8/encoder/ethreading.c: In function 'vp8cx_create_encoder_threads':
vp8/encoder/ethreading.c:447: warning: implicit declaration of function
'mach_task_self'

    [CC] vp8/decoder/threading.c.o
vp8/decoder/threading.c: In function 'vp8_decoder_create_threads':
vp8/decoder/threading.c:405: warning: implicit declaration of function
'mach_task_self'

    [CC] vp8/encoder/onyx_if.c.o
vp8/encoder/onyx_if.c: In function 'encode_frame_to_data_rate':
vp8/encoder/onyx_if.c:4537: warning: format '%6ld' expects type 'long int',
but argument 11 has type 'int'
vp8/encoder/onyx_if.c:4537: warning: format '%6ld' expects type 'long int',
but argument 12 has type 'int'
vp8/encoder/onyx_if.c:4537: warning: format '%6ld' expects type 'long int',
but argument 13 has type 'int'
vp8/encoder/onyx_if.c:4537: warning: format '%6ld' expects type 'long int',
but argument 14 has type 'int'
vp8/encoder/onyx_if.c:4537: warning: format '%5ld' expects type 'long int',
but argument 15 has type 'int'
vp8/encoder/onyx_if.c:4537: warning: format '%5ld' expects type 'long int',
but argument 16 has type 'int'
vp8/encoder/onyx_if.c:4537: warning: format '%5ld' expects type 'long int',
but argument 17 has type 'unsigned int'
vp8/encoder/onyx_if.c:4537: warning: format '%8ld' expects type 'long int',
but argument 18 has type 'int'
vp8/encoder/onyx_if.c:4537: warning: format '%8ld' expects type 'long int',
but argument 23 has type 'unsigned int'
vp8/encoder/onyx_if.c:4539: warning: format '%6ld' expects type 'long int',
but argument 11 has type 'int'
vp8/encoder/onyx_if.c:4539: warning: format '%6ld' expects type 'long int',
but argument 12 has type 'int'
vp8/encoder/onyx_if.c:4539: warning: format '%6ld' expects type 'long int',
but argument 13 has type 'int'
vp8/encoder/onyx_if.c:4539: warning: format '%6ld' expects type 'long int',
but argument 14 has type 'int'
vp8/encoder/onyx_if.c:4539: warning: format '%5ld' expects type 'long int',
but argument 15 has type 'int'
vp8/encoder/onyx_if.c:4539: warning: format '%5ld' expects type 'long int',
but argument 16 has type 'int'
vp8/encoder/onyx_if.c:4539: warning: format '%5ld' expects type 'long int',
but argument 17 has type 'unsigned int'
vp8/encoder/onyx_if.c:4539: warning: format '%8ld' expects type 'long int',
but argument 18 has type 'int'
vp8/encoder/onyx_if.c:4539: warning: format '%8ld' expects type 'long int',
but argument 22 has type 'unsigned int'

I'm following the instruction on your web page on how to configure and then 
make to create the VS project and solution files.  It should create 
solution files.  Instead, I get the following error.  It doesn't matter 
whether I configure for VS7 or VS8, I get the same make error.

make[1]: *** No rule to make target `vp8/vp8.h', needed by `vpx.vcproj'.  
Stop.
make: *** [.DEFAULT] Error 2

What version are you using? On what operating system?

I have both VS 2003 and VS 2005 installed on Win XP.  I'm using the latest 
cygwin with make v3.81

In vp8/vp8_dx_iface.c, in function vp8_xma_set_mmap, if ctx->priv is
initially null and mmap->id is not VP8_SEG_ALG_PRIV, then ctx->priv will
still be null when this code:

  if (ctx->priv->alg_priv)

on line 529 is executed.

Assembly code can not be linked in SAFESEH mode on win32 due to lack of
support in yasm. Also, configuring libvpx with --enable-pic causes bugs
with yasm-1.0.0 that aren't present in yasm-0.8.0

I'm trying to add WebM/Matroska support to Android-StageFright, but as 
libebml makes use of libraries like "string.h"  and "algorithm.h" which 
are not available on android's bionicC I'm facing porting issues.

Matthew Heaney told me on the forum that google has a WebM / Matroska 
parser library for Android already that doesn't use any STL. 

Can you please make it available?

What is the expected behavior?
same quality
What do you see instead?
different quality

What version are you using?
may 27, 2010.
On what operating system?
win7

Can you reproduce using the ivfdec or ivfenc tools?
yes
What command line are
you using?

d:\mediatests\tulip>timex ivfenc_msys32 --passes=2 --timebase=1/30
t-ref=1 --minsection-pct=0 --maxsection-pct=1200 --lag-in-frames=1
0 --max-q=63  -t 4 -w 1280 -h 720 --target-bitrate=2000 tulip.1280
Pass 1/2 frame  500/501    64128B    1026b/f   61562b/s   31345 ms
Pass 2/2 frame  500/537  4059691B     119F   4372F   5847F   5557F
Pass 2/2 frame  500/552  4168472B   66695b/f 4001733b/s  183284 ms
timex 713411.35ms

d:\mediatests\tulip>timex ivfdec_msys32 tulip.vp8.ivf --summary -o
vpx Technologies VP8 Decoder v0.9.0-8-g6cd4a10
552 decoded frames/500 showed frames in 4607477 us (108.52 fps)
timex 6263.42ms

d:\mediatests\tulip>psnr tulip.1280x720_30Hz_P420.yuv tulip_vp8.12
Frame    PSNR-Y          PSNR-U          PSNR-V          PSNR-All
Global:  38.033677       41.838680       41.917746       38.979976
Avg:     38.122031       41.931030       42.009934       39.064884
Min:     36.556165       40.174052       40.094529       37.497207

d:\mediatests\tulip>timex ivfenc_cyg32 --passes=2 --timebase=1/30
-ref=1 --minsection-pct=0 --maxsection-pct=1200 --lag-in-frames=16
 --max-q=63  -t 4 -w 1280 -h 720 --target-bitrate=2000 tulip.1280x
Pass 1/2 frame  500/501    64128B    1026b/f   61562b/s   25578 ms
Pass 2/2 frame  500/537  4060330B     116F   4520F   5396F   5698F
Pass 2/2 frame  500/552  4169751B   66716b/f 4002960b/s  205933 ms
timex 238802.94ms

d:\mediatests\tulip>timex ivfdec_cyg32 tulip.vp8.ivf --summary -o
vpx Technologies VP8 Decoder v0.9.0-8-g6cd4a10
552 decoded frames/500 showed frames in 6746000 us (74.12 fps)
timex 7947.49ms

d:\mediatests\tulip>psnr tulip.1280x720_30Hz_P420.yuv tulip_vp8.12
Frame    PSNR-Y          PSNR-U          PSNR-V          PSNR-All
Global:  38.033865       41.844806       41.914986       38.980427
Avg:     38.121606       41.934641       42.007598       39.064614
Min:     36.561687       40.266481       40.091443       37.507611

d:\mediatests\tulip>timex ivfenc_vc32 --passes=2 --timebase=1/30 -
ref=1 --minsection-pct=0 --maxsection-pct=1200 --lag-in-frames=16
--max-q=63  -t 4 -w 1280 -h 720 --target-bitrate=2000 tulip.1280x7
Pass 1/2 frame  500/501    64128B    1026b/f   61562b/s   61048 ms
Pass 2/2 frame  500/537  4060284B     121F   4330F   5844F   5529F
Pass 2/2 frame  500/552  4169423B   66710b/f 4002646b/s  326559 ms
timex 420847.90ms

d:\mediatests\tulip>timex ivfdec_vc32 tulip.vp8.ivf --summary -o t
vpx Technologies VP8 Decoder v0.9.0-8-g6cd4a10
552 decoded frames/500 showed frames in 7910257 us (63.21 fps)
timex 9530.13ms

d:\mediatests\tulip>psnr tulip.1280x720_30Hz_P420.yuv tulip_vp8.12
Frame    PSNR-Y          PSNR-U          PSNR-V          PSNR-All
Global:  38.031713       41.835912       41.908388       38.977316
Avg:     38.119750       41.927665       42.001722       39.061948
Min:     36.569751       40.261691       40.062244       37.504356

d:\mediatests\tulip>timex ivfenc_vc64 --passes=2 --timebase=1/30 -
ref=1 --minsection-pct=0 --maxsection-pct=1200 --lag-in-frames=16
--max-q=63  -t 4 -w 1280 -h 720 --target-bitrate=2000 tulip.1280x7
Pass 1/2 frame  500/501    64128B    1026b/f   61562b/s   59565 ms
Pass 2/2 frame  500/537  4060391B     122F   4538F   5817F   5460F
Pass 2/2 frame  500/552  4169677B   66714b/f 4002889b/s  316246 ms
timex 402394.19ms

d:\mediatests\tulip>timex ivfdec_vc64 tulip.vp8.ivf --summary -o t
vpx Technologies VP8 Decoder v0.9.0-8-g6cd4a10
552 decoded frames/500 showed frames in 8344515 us (59.92 fps)
timex 10423.46ms

d:\mediatests\tulip>psnr tulip.1280x720_30Hz_P420.yuv tulip_vp8.12
Frame    PSNR-Y          PSNR-U          PSNR-V          PSNR-All
Global:  38.032442       41.833797       41.911142       38.977971
Avg:     38.122506       41.926802       42.005882       39.064489
Min:     36.559204       40.260343       40.096123       37.500843

Please provide any additional information below.
Performance varies alot, but thats somewhat expected.
cygwin built ivfenc is both fastest, and highest quality.
But its msys and visual c I want to use.

WebM multiplex with video, audio, and others should have application/webm,
not video/webm. Video/webm should be reserved only to video.

Hola, quisiera saber si existe o no informacion de Webm en español.

gracias.

What steps will reproduce the problem?
1. Get a video h.264/aac 480x270 (see GSpot screenshot)
2. Use MS GraphEdit to build the following filter graph
Haali-Splitter => ffdshow => WebM VP8&Vorbis => WebM muxer => file writer
Haali & Ogg updated today
see screenshot of filtergraph
3. run filter an wait, wait, wait - about 5 minutes later (transcoding with
any codec else finishes after about 1 minute!) the muxer crashes:
see screenshots

Running on Vista 32bit SP2, Core2 2.4GHz, 2 GB RAM

Hi,
commit 6cd4a10e167203d1deb79abf60ee72599e97891b added to most defines the 
prefix VPX_. 
The PLANE_* defines still don't have the VPX_ prefix yet.

From Sebastian Dröge <[email protected]>:

Another thing I noticed is, that the assembly doesn't specify
that it doesn't need a executable stack. This will prevent using libvpx
on some systems where an executable stack is not allowed for security
reasons.

See
http://etbe.coker.com.au/2007/10/07/executable-stack-and-shared-objects/
for more information. The solution is quite easy because your assembly
really doesn't need an executable stack. Just add
.section .note.GNU-stack,"",@progbits
to the start of the assembly files.

Hi,
when compiling the compiler with --enable-psnr it will output PSNR packets, no 
matter if 
VPX_CODEC_USE_PSNR is provided during encoder initialization or not.

HAVE_CONFIG_H and vpx_config.h is a known pain point for many people. 

Recently from Collabra:

another thing I've noticed is, that the vpx_codec.h header includes
vpx_config.h if HAVE_CONFIG_H is defined. HAVE_CONFIG_H is often defined
by autotools build systems too and if vpx_codec.h is included from there
the vpx_config.h is not found (because it's not installed by make
install).

The vpx_config.h header should probably be included by some private
header instead.

What is the expected behavior?
should configure and build like other compilers

What do you see instead?
The configure does not work from a windows command prompt.  Could you make 
configure work from a regular shell?  (low priority)
I use studio 2008, so first it tries to convert.  Could you add vs9?
You could automate this with either a native 2008 target, or a devenv 
/upgrade command.  You shouldnt need to go into the GUI... it can do the 
convert and build from command line.
Then it fails to upgrade unless you follow the yasm install procedure.
I have yasm in my path.  Can you make it use it?  Or detect that it needs 
to be copied and do that?  Or at least report the error and solution.

What version are you using?
studio 2008
On what operating system?
windows 7

Can you reproduce using the ivfdec or ivfenc tools? What command line are
you using?

cd /c/on2
mkdir win32
cd /c/on2/win32
../libvpx/configure --target=x86-win32-vs8 --disable-install-docs --
disable-codecs --enable-vp8 --disable-ccache --disable-debug --disable-psnr 
--disable-postproc
make clean
make
make install DIST_DIR=/c/libvpx/win32

Please provide any additional information below.
both 32 and 64 bit do build once configured correctly.

Add options for manual bitrate video

In vp8/encoder/onyx_if.c starting at line 4593, there are a few
lines similar to this:

        cpi->ref_frame_flags &= !VP8_GOLD_FLAG;

It looks like those '!'s should be '~'s. 

Assertion in webmux while writing video only graph


I am using 64 bit Windows 7

I am using graph edit to construct the graph

The main reason for the assertion might be that the audio is not attached 
to mux.

If Video only file is allowed in the mux then this is a bug other wise not.

Complete graph with assertion message is shown in the attached picture

../libvpx/ivfenc.c:28:22: error: sys/mman.h: No such file or directory

I can find the file here.  Do we need to patch msys?

C:\msys>dir /s mman.h
 Directory of C:\msys\1.0\fixincludes\tests\base\sys
08/31/2004  02:27 AM               530 mman.h

Is it possible for webm to provide motion detect capability?

Once directshow filter started we cannot change properties of the encoder
on the fly (Ex: Targetbitrate, Deadline, etc...).

Call to a method (setter) of IVP8Encoder is not apply to the encoder once
the filter is started.

There should be some official vp8 encoded files available via download
section, for public test.

Running makewebm.exe to make a webm from sample.avi creates a webm.
However when playing back it with playwebm.exe, it says it is invalid format.
The same thing happens in graphedit too.

When playing with sample.avi file in filter/Video, I got assertion messasge.

The graph is simple.

source avi-> avi splitter -> vp8 encoder -> webm muxer -> file writer

                          -> Vorbis encoder ->

Please refer to the attached png.  

When using the direct show encoder, currently there is no (easy?) way to
set encoding parameters. Typical third-party components for video encoding,
that handle all the low level direct show stuff for you can only call a
codecs property page to set it up.

https://groups.google.com/a/webmproject.org/group/webm-discuss/browse_thread/thr
ead/56f3d5aca779c2c3#

What is the expected behavior?
DIST_DIR should contain libs, bins etc as on msys builds.

What do you see instead?
 Directory of c:\libvpx\win32
05/27/2010  09:51 PM    <DIR>          .
05/27/2010  09:51 PM    <DIR>          ..
05/27/2010  09:51 PM    <DIR>          build
05/27/2010  09:50 PM    <DIR>          include
05/27/2010  09:51 PM             2,269 md5sums.txt
05/27/2010  09:50 PM    <DIR>          src

What version are you using?
May 27 version.  How do you get this?

On what operating system?
Win 7 64 bit.

Can you reproduce using the ivfdec or ivfenc tools? What command line are
you using?

cd /c/on2
mkdir win32
cd /c/on2/win32
../libvpx/configure --target=x86-win32-vs8 --disable-install-docs --
disable-codecs --enable-vp8 --disable-ccache --disable-debug --disable-psnr 
--disable-postproc
make clean
make
make install DIST_DIR=/c/libvpx/win32

The default is to convert the project, but if you say build, it builds 
Debug into win32\Debug
If I manually switch to Release and build that, the install works
 Directory of C:\libvpx\win32
05/28/2010  01:47 PM    <DIR>          .
05/28/2010  01:47 PM    <DIR>          ..
05/28/2010  01:47 PM    <DIR>          bin
05/27/2010  09:51 PM    <DIR>          build
05/27/2010  09:50 PM    <DIR>          include
05/28/2010  01:47 PM    <DIR>          lib
05/27/2010  09:50 PM    <DIR>          src

The files are put into a Win32 subfolder of bin and lib.
Seems like it should be done the same as linux 32 and 64 bit builds, which 
is the files go in lib and bin itself.

lib\Win32\vpxmd.lib

Please provide any additional information below.
Visual Studio 2008

Packaging libvpx for Ubuntu for use by GStreamer will require shared 
libraries. configure currently has no way of producing shared libraries.

Ideally pkg-config should be used to simplify the configure scripts of 
projects using libvpx.so.

steps to reproduce



ffmpeg -y -f mjpeg -i orig\B000GW69Y8x2E01-
A13RZ30EC2G5CFx2E_SS75_x2E.jpgx2E01x2E_SL110_SCLZZZZZZZ_x2Ejpg -vcodec 
rawvideo -pix_fmt yuv420p -vframes 1 -s 75x75 yuv\B000GW69Y8x2E01-
A13RZ30EC2G5CFx2E_SS75_x2E.jpgx2E01x2E_SL110_SCLZZZZZZZ_x2Ejpg.75x75_P420.y
uv 
if not exist yuv\B000GW69Y8x2E01-
A13RZ30EC2G5CFx2E_SS75_x2E.jpgx2E01x2E_SL110_SCLZZZZZZZ_x2Ejpg.75x75_P420.y
uv echo READERR orig\B000GW69Y8x2E01-
A13RZ30EC2G5CFx2E_SS75_x2E.jpgx2E01x2E_SL110_SCLZZZZZZZ_x2Ejpg 
ffmpeg -y -s 75x75 -i yuv\B000GW69Y8x2E01-
A13RZ30EC2G5CFx2E_SS75_x2E.jpgx2E01x2E_SL110_SCLZZZZZZZ_x2Ejpg.75x75_P420.y
uv -an png\B000GW69Y8x2E01-
A13RZ30EC2G5CFx2E_SS75_x2E.jpgx2E01x2E_SL110_SCLZZZZZZZ_x2Ejpg.75x75_P420.p
ng 
ivfenc --passes=1 --timebase=1/30 --static-thresh=0 --drop-frame=0 --good -
-minsection-pct=0 --maxsection-pct=1200 --min-q=1 --max-q=1 -t 1 -w 75 -h 
75 yuv\B000GW69Y8x2E01-
A13RZ30EC2G5CFx2E_SS75_x2E.jpgx2E01x2E_SL110_SCLZZZZZZZ_x2Ejpg.75x75_P420.y
uv vp8\B000GW69Y8x2E01-
A13RZ30EC2G5CFx2E_SS75_x2E.jpgx2E01x2E_SL110_SCLZZZZZZZ_x2Ejpg.vp8 
ivfdec vp8\B000GW69Y8x2E01-
A13RZ30EC2G5CFx2E_SS75_x2E.jpgx2E01x2E_SL110_SCLZZZZZZZ_x2Ejpg.vp8 --
summary -o unvp8\B000GW69Y8x2E01-
A13RZ30EC2G5CFx2E_SS75_x2E.jpgx2E01x2E_SL110_SCLZZZZZZZ_x2Ejpg_vp8.75x75_P4
20.yuv 
ffmpeg -y -s 75x75 -i yuv\B000GW69Y8x2E01-
A13RZ30EC2G5CFx2E_SS75_x2E.jpgx2E01x2E_SL110_SCLZZZZZZZ_x2Ejpg.75x75_P420.y
uv -pix_fmt yuv420p -qscale 8 -an jpg\B000GW69Y8x2E01-
A13RZ30EC2G5CFx2E_SS75_x2E.jpgx2E01x2E_SL110_SCLZZZZZZZ_x2Ejpg.jpg 
ffmpeg -y -s 75x75 -i unvp8\B000GW69Y8x2E01-
A13RZ30EC2G5CFx2E_SS75_x2E.jpgx2E01x2E_SL110_SCLZZZZZZZ_x2Ejpg_vp8.75x75_P4
20.yuv -an unpng\B000GW69Y8x2E01-
A13RZ30EC2G5CFx2E_SS75_x2E.jpgx2E01x2E_SL110_SCLZZZZZZZ_x2Ejpg.75x75_P420.p
ng 

original (jpeg) and output (png) attached

Hi. 
Please, add support for other media types then MEDIASUBTYPE_YV12. 
Standart system "Color space converter" does not supports 
MEDIASUBTYPE_YV12. So, it's not possible to connect capture source 
filter (web-camera, tv-tuner) to the WebM encoder without third-side 
filters.

What steps will reproduce the problem?
1. Create a video only webm file.
2. Play this file using any webm player.

What is the expected output? What do you see instead?
Video was recorded at 29.976 fps but is not being played at the same rate.

What version are you using? On what Windows version?
I am using Windows 7 64 bit and Webmdshow 0.9.7.0

Please provide any additional information below.

Recording Duration of the attached video is about 10 seconds, as reported 
in graph edit but playback of file finishes in about 1-2 seconds both in 
Media Player Classic and in playwebm.exe

I have also attached picture of average frame time at which file was 
recorded.

From [email protected]:

The decode only version of the lib should be as small as possible for 
several reasons. one is physical size - smaller helps download faster and 
improves startup time. another is security - we dont want code in there 
that is a security risk.  less code is better.

Looking at the lib with size, there are some object files that dont seem 
necessary.  Below is the full list
we dont need the encoder
  text    data     bss     dec     hex filename
  1360       0       0    1360     550 vpx_encoder.c.o (ex 
c:/libvpx/lib/libvpx.a)
scalers?  is this for spatial resampling?
   256       0       0     256     100 scalesystemdependant.c.o (ex 
c:/libvpx/lib/libvpx.a)
  3296       0       0    3296     ce0 gen_scalers.c.o (ex 
c:/libvpx/lib/libvpx.a)
we dont need postproc
  6736       0       0    6736    1a50 postproc.c.o (ex 
c:/libvpx/lib/libvpx.a)
  1040       0       0    1040     410 postproc_mmx.asm.o (ex 
c:/libvpx/lib/libvpx.a)
  1472       0       0    1472     5c0 postproc_sse2.asm.o (ex 
c:/libvpx/lib/libvpx.a)
shouldnt this be inline?
    16       0       0      16      10 emms.asm.o (ex 
c:/libvpx/lib/libvpx.a)
text?
   528       0       0     528     210 textblit.c.o (ex 
c:/libvpx/lib/libvpx.a)

Here is the full list
  text    data     bss     dec     hex filename
  1392       0       0    1392     570 vpx_codec.c.o (ex 
c:/libvpx/lib/libvpx.a)
   784       0       0     784     310 vpx_decoder.c.o (ex 
c:/libvpx/lib/libvpx.a)
   688       0       0     688     2b0 vpx_decoder_compat.c.o (ex 
c:/libvpx/lib/libvpx.a)
  1440       0       0    1440     5a0 vpx_image.c.o (ex 
c:/libvpx/lib/libvpx.a)
  1360       0       0    1360     550 vpx_encoder.c.o (ex 
c:/libvpx/lib/libvpx.a)
   432       0       0     432     1b0 vpx_mem.c.o (ex 
c:/libvpx/lib/libvpx.a)
  5072       0      96    5168    1430 vpxscale.c.o (ex 
c:/libvpx/lib/libvpx.a)
   480       0       0     480     1e0 yv12config.c.o (ex 
c:/libvpx/lib/libvpx.a)
  1296       0       0    1296     510 yv12extend.c.o (ex 
c:/libvpx/lib/libvpx.a)
   256       0       0     256     100 scalesystemdependant.c.o (ex 
c:/libvpx/lib/libvpx.a)
  3296       0       0    3296     ce0 gen_scalers.c.o (ex 
c:/libvpx/lib/libvpx.a)
   256       0       0     256     100 segmentation_common.c.o (ex 
c:/libvpx/lib/libvpx.a)
  1712       0       0    1712     6b0 alloccommon.c.o (ex 
c:/libvpx/lib/libvpx.a)
   592       0       0     592     250 blockd.c.o (ex 
c:/libvpx/lib/libvpx.a)
  1440       0       0    1440     5a0 debugmodes.c.o (ex 
c:/libvpx/lib/libvpx.a)
  6704     256     208    7168    1c00 entropy.c.o (ex 
c:/libvpx/lib/libvpx.a)
  1904       0       0    1904     770 entropymode.c.o (ex 
c:/libvpx/lib/libvpx.a)
   128       0       0     128      80 entropymv.c.o (ex 
c:/libvpx/lib/libvpx.a)
   736       0       0     736     2e0 extend.c.o (ex 
c:/libvpx/lib/libvpx.a)
 11264       0       0   11264    2c00 filter_c.c.o (ex 
c:/libvpx/lib/libvpx.a)
  1360       0       0    1360     550 findnearmv.c.o (ex 
c:/libvpx/lib/libvpx.a)
   352       0       0     352     160 systemdependent.c.o (ex 
c:/libvpx/lib/libvpx.a)
  2048       0       0    2048     800 idctllm.c.o (ex 
c:/libvpx/lib/libvpx.a)
  2192       0       0    2192     890 invtrans.c.o (ex 
c:/libvpx/lib/libvpx.a)
  8160       0       0    8160    1fe0 loopfilter.c.o (ex 
c:/libvpx/lib/libvpx.a)
  6656       0       0    6656    1a00 loopfilter_filters.c.o (ex 
c:/libvpx/lib/libvpx.a)
  2400       0       0    2400     960 mbpitch.c.o (ex 
c:/libvpx/lib/libvpx.a)
    96       0       0      96      60 modecont.c.o (ex 
c:/libvpx/lib/libvpx.a)
  4000       0       0    4000     fa0 modecontext.c.o (ex 
c:/libvpx/lib/libvpx.a)
   128       0       0     128      80 predictdc.c.o (ex 
c:/libvpx/lib/libvpx.a)
  1360       0       0    1360     550 quant_common.c.o (ex 
c:/libvpx/lib/libvpx.a)
  2256       0       0    2256     8d0 recon.c.o (ex 
c:/libvpx/lib/libvpx.a)
  9696       0       0    9696    25e0 reconinter.c.o (ex 
c:/libvpx/lib/libvpx.a)
  9088       0       0    9088    2380 reconintra.c.o (ex 
c:/libvpx/lib/libvpx.a)
  2064       0       0    2064     810 reconintra4x4.c.o (ex 
c:/libvpx/lib/libvpx.a)
   352       0       0     352     160 setupintrarecon.c.o (ex 
c:/libvpx/lib/libvpx.a)
    64       0       0      64      40 swapyv12buffer.c.o (ex 
c:/libvpx/lib/libvpx.a)
   528       0       0     528     210 textblit.c.o (ex 
c:/libvpx/lib/libvpx.a)
  1648       0       0    1648     670 treecoder.c.o (ex 
c:/libvpx/lib/libvpx.a)
   800       0       0     800     320 x86_systemdependent.c.o (ex 
c:/libvpx/lib/libvpx.a)
  2496       0       0    2496     9c0 vp8_asm_stubs.c.o (ex 
c:/libvpx/lib/libvpx.a)
  2944       0       0    2944     b80 loopfilter_x86.c.o (ex 
c:/libvpx/lib/libvpx.a)
  6736       0       0    6736    1a50 postproc.c.o (ex 
c:/libvpx/lib/libvpx.a)
  2928       0       0    2928     b70 vp8_dx_iface.c.o (ex 
c:/libvpx/lib/libvpx.a)
  1504       0       0    1504     5e0 dboolhuff.c.o (ex 
c:/libvpx/lib/libvpx.a)
 16208       0      96   16304    3fb0 decodemv.c.o (ex 
c:/libvpx/lib/libvpx.a)
 20896       0       0   20896    51a0 decodframe.c.o (ex 
c:/libvpx/lib/libvpx.a)
  5024       0       0    5024    13a0 demode.c.o (ex 
c:/libvpx/lib/libvpx.a)
   736       0       0     736     2e0 dequantize.c.o (ex 
c:/libvpx/lib/libvpx.a)
  7856      32       0    7888    1ed0 detokenize.c.o (ex 
c:/libvpx/lib/libvpx.a)
    96       0       0      96      60 dsystemdependent.c.o (ex 
c:/libvpx/lib/libvpx.a)
  2048       0      16    2064     810 onyxd_if.c.o (ex 
c:/libvpx/lib/libvpx.a)
  5568       0       0    5568    15c0 threading.c.o (ex 
c:/libvpx/lib/libvpx.a)
   368       0       0     368     170 x86_dsystemdependent.c.o (ex 
c:/libvpx/lib/libvpx.a)
   128       0       0     128      80 vpx_config.c.o (ex 
c:/libvpx/lib/libvpx.a)
   480       0       0     480     1e0 idctllm_mmx.asm.o (ex 
c:/libvpx/lib/libvpx.a)
   288       0       0     288     120 iwalsh_mmx.asm.o (ex 
c:/libvpx/lib/libvpx.a)
   592       0       0     592     250 recon_mmx.asm.o (ex 
c:/libvpx/lib/libvpx.a)
  2192       0       0    2192     890 subpixel_mmx.asm.o (ex 
c:/libvpx/lib/libvpx.a)
  3968       0       0    3968     f80 loopfilter_mmx.asm.o (ex 
c:/libvpx/lib/libvpx.a)
   512       0       0     512     200 recon_sse2.asm.o (ex 
c:/libvpx/lib/libvpx.a)
  1888       0       0    1888     760 subpixel_sse2.asm.o (ex 
c:/libvpx/lib/libvpx.a)
  5424       0       0    5424    1530 loopfilter_sse2.asm.o (ex 
c:/libvpx/lib/libvpx.a)
   256       0       0     256     100 iwalsh_sse2.asm.o (ex 
c:/libvpx/lib/libvpx.a)
  1040       0       0    1040     410 postproc_mmx.asm.o (ex 
c:/libvpx/lib/libvpx.a)
  1472       0       0    1472     5c0 postproc_sse2.asm.o (ex 
c:/libvpx/lib/libvpx.a)
   848       0       0     848     350 dequantize_mmx.asm.o (ex 
c:/libvpx/lib/libvpx.a)
    16       0       0      16      10 emms.asm.o (ex 
c:/libvpx/lib/libvpx.a)
     0       0       0       0       0 x86_abi_support.asm.o (ex 
c:/libvpx/lib/libvpx.a)

I'm okay with things being in the .a file, so long as they dont get linked 
into the final ffmpeg dll

Is it possible to implement:

Reversed complexity encoding / Z-frames: A type of frame that in encoding 
complexity is lower than a regular P-frame but is more compact. The catch 
is decoding computation needs to be higher. Specific technical approaches 
might include:

1. a combination of multi-frame processing with Wyner-Ziv coding

2. mixed quality encoding, where the Z-frames are non-reference frames 
interspersed with regular P-frames and coded at lower quality than the 
target quality. The decoder recovers a higher quality version of the Z-
frames by multi-frame processing using information in the neighboring 
higher quality P-frames.

3. supplement the information transmitted for the Z-frame with additional 
helper information to regularize the reconstruction process on the decoder 
side. This could be a Wyner-Ziv layer, or something different.

mplayer@31189_libvpx@6cd4a10-VP8__WebM.diff patch rejected in  
libavformat/avformat.h and libavformat/matroskadec.c.

and after corrected rejected parts of diff,compile ok.
but compiled mplayer can`t play webm file.
sound ok but video not working.error message is 
Cannot find codec matching selected -vo and video format 0x30385056


[/media/sdc1/mplayer-vp8-encdec-support/mplayer-31189]$ ffprobe-
webm /media/sdc1/shu1.webm
FFprobe version SVN-r23210, Copyright (c) 2007-2010 the FFmpeg 
developers
  built on May 23 2010 08:33:55 with gcc 4.3.3
  configuration: --enable-gpl --enable-version3 --enable-nonfree --enable-
postproc --enable-pthreads --enable-libfaac --enable-libfaad --enable-
libmp3lame --enable-libopencore-amrnb --enable-libopencore-amrwb --
enable-libtheora --enable-libx264 --enable-libxvid --enable-x11grab --
enable-libvpx --enable-avfilter --enable-avfilter-lavf --enable-bzlib --
enable-zlib --enable-libfaadbin --enable-libgsm --enable-libnut --enable-
libschroedinger --enable-libvorbis --enable-libspeex --enable-libopenjpeg -
-enable-libdirac --disable-debug
  libavutil     50.15. 3 / 50.15. 3
  libavcodec    52.69. 3 / 52.69. 3
  libavformat   52.64. 0 / 52.64. 0
  libavdevice   52. 2. 0 / 52. 2. 0
  libavfilter    1.20. 0 /  1.20. 0
  libswscale     0.10. 0 /  0.10. 0
  libpostproc   51. 2. 0 / 51. 2. 0
[libvpx @ 0x8b46130]v0.9.0-8-g6cd4a10
[matroska @ 0x8b427d0]Estimating duration from bitrate, this may be 
inaccurate
Input #0, matroska, from '/media/sdc1/shu1.webm':
  Metadata:
    doctype         : webm
  Duration: 00:00:53.91, start: 0.000000, bitrate: N/A
    Stream #0.0: Video: libvpx, yuv420p, 480x360, PAR 1:1 DAR 4:3, 30 
tbr, 1k tbn, 1k tbc
    Stream #0.1: Audio: vorbis, 44100 Hz, stereo, s16
[libvpx @ 0x8b46130]v0.9.0-8-g6cd4a10
^_^[/media/sdc1/mplayer-vp8-encdec-support/mplayer-31189]$ 
./mplayer /media/sdc1/shu1.webm
MPlayer SVN-r31189-4.3.3 (C) 2000-2010 MPlayer Team
150 audio & 338 video codecs
mplayer: could not connect to socket
mplayer: No such file or directory
Failed to open LIRC support. You will not be able to use your remote 
control.

Playing /media/sdc1/shu1.webm.
[mkv] Track ID 1: video (V_VP8), -vid 0
[mkv] Track ID 2: audio (A_VORBIS), -aid 0, -alang und
[mkv] Will play video track 1.
Matroska file format detected.
VIDEO:  [VP80]  480x360  24bpp  25.000 fps    0.0 kbps ( 0.0 kbyte/s)
Opening video filter: [screenshot]
Opening video filter: [hqdn3d]
Opening video filter: [pp=de]
==================================================
========================
Cannot find codec matching selected -vo and video format 0x30385056.
==================================================
========================
==================================================
========================
Opening audio decoder: [ffmpeg] FFmpeg/libavcodec audio decoders
AUDIO: 44100 Hz, 2 ch, s16le, 0.0 kbit/0.00% (ratio: 0->176400)
Selected audio codec: [ffvorbis] afm: ffmpeg (FFmpeg Vorbis)
==================================================

The VP8 encoder filter doesn't notify the downstream filter about the 
allocator chosen by the outpin.  Under these circumstances the AVI Mux filter 
refuses to pause/run, so the graph cannot even start.

The solution is to call IMemInputPin::NotifyAllocator on the downstream pin.

Compiling from git master at the moment, on a Fedora 13 x86_64 box. 
Compilation was fine.

The "make install" target though, which also looked ok, additionally
reported a trivial error:

  /bin/sh: line 3: [: =: unary operator expected
  ../src/configure --target=generic-gnu

Thought you guys might like to know.

Further error text below:

**********************************************************************

$ make install
    [INSTALL] vpx-vp8-nopost-nodocs-generic-gnu-v0.9.0/include/vp8.h
    [INSTALL] vpx-vp8-nopost-nodocs-generic-gnu-v0.9.0/include/vp8e.h
<snipped for brevity>
    [INSTALL] vpx-vp8-nopost-nodocs-generic-gnu-v0.9.0/bin/vp8cx_set_ref
make[1]: Nothing to be done for `install'.
/bin/sh: line 3: [: =: unary operator expected
../src/configure --target=generic-gnu
Configuring selected codecs
  enabling vp8_encoder
  enabling vp8_decoder
Configuring for target 'generic-gnu'
  enabling generic
  disabling ccache
Creating makefiles for generic-gnu examples
    [MD5SUM] vpx-vp8-nopost-nodocs-generic-gnu-v0.9.0
$

**********************************************************************

What is the expected behavior? What do you see instead?

When building libvpx with --enable-pic, it should be possible to build a 
linked library using gcc -fPIC -shared ... -lvpx

The minimal test "library" I'm compiling is attached in init_destroy.c. 
Put this and reproduce.sh in the same directory as a libvpx checkout and 
run it on a 64-bit Linux system.

This is the gcc command:

gcc -shared -fPIC -o init_destroy.so init_destroy.c -Itmp/include/ -Ltmp/
lib/ -lvpx -lm

It fails with this error:

/usr/bin/ld: tmp/lib//libvpx.a(subpixel_mmx.asm.o): relocation 
R_X86_64_PC32 against symbol `vp8_bilinear_filters_mmx' can not be used 
when making a shared object; recompile with -fPIC
/usr/bin/ld: final link failed: Bad value
collect2: ld returned 1 exit status

This does not happen on 32-bit Linux.

What version are you using? On what operating system?

libvpx git commit 1689564bb5c0f03bb2f35244bf40bcf58c9fec35 (Merge "expose 
vp8_deblock")

gcc (Ubuntu 4.4.3-4ubuntu5) 4.4.3

Please provide any additional information below.

Adding --target=generic-gnu to the configure line "fixes" the problem, 
suggesting that this another problem with the relocation code in the 
assembly.

--disable-mmx --disable-sse2 is also enough to "fix" the problem, but 
perhaps this is just a side-effect of the broken thing (whatever it is) 
not having an optimized version for sse, sse3 and the rest.

WebM should have codecs for metadata and tags. I propose XML format
Continuous Media Markup Language [1] or RDF format (in development)
Ontology for Media Resource [2].

[1] http://www.annodex.net/TR/draft-pfeiffer-cmml-01.html
[2] http://dev.w3.org/2008/video/mediaann/mediaont-1.0/mediaont-1.0.html

Reporting as a bug mostly to test out our bug tracking

---
The Videolan project have released a (very stable) RC of their upcoming 
release, with WebM support (decoding only for now, but encoding support is 
on its way as well).

http://people.videolan.org/~jb/webm/ 

Given that it's an exceedingly popular media player on Windows/Mac, do you 
think the link could be added to http://www.webmproject.org/users/ in the 
media players section?

As a side-note, their stabilization branch for 1.1.0 has been ready for 
several weeks now, but they've been holding off on cutting the RC in the 
hopes that the rumors about VP8 would turn out to be true. I hereby relay 
to your their kudos :).

At the moment, tagging is declared as "Undecided" in the container spec[1]. In 
my opinion, tagging should be supported. Unlike subtitles, tagging is useful 
starting from the (often offline) source (e.g. an encoder attaching its name 
and settings or a cam recording to WebM and storing creator name) and should 
therefor be supported directly in the container format spec - other methods are 
unlikely to be sufficiently coupled to the file and/or to be standardized.

[1] http://www.webmproject.org/code/specs/container/#tagging

sync up chromium and libvpx
build libvpx and attempt to build ffmpeg
CC      libavcodec/libvpxenc.o
libavcodec/libvpxenc.c: In function 'vp8_init':
libavcodec/libvpxenc.c:266: error: 'IMG_FMT_I420' undeclared (first use in 
this function)

As it exist an RFC for using H.264 with RTP/RTSP,
http://www.faqs.org/rfcs/rfc3984.html, specifications for using VP8 with
such protocols should be written and implemented into HTML5 browsers.

RTP/RTSP is the only transport protocol for video which ensure low-latency,
it is tolerant to packet loss, it enable multi-cast transmission with UDP.

These protocols are use by most of the IP based surveillance cameras
(latest Camera all support RTSP + RTP + H.264).

HTML5 browsers should support this transport protocol and should allow RTSP
URL inside <video> tags 
( <video src="rtsp://my.webtv.com/myChannel/myProgram" /> )

To be able to directly install the library on *nix systems, we should use 
the standard file system layout in make install. We should also maintain the 
current behavior, perhaps as 'make dist'