July 2005 - ffmpeg-devel - lists.ffmpeg.org

mpegts pid values
by roussel＠lri.fr 10 Aug '05

10 Aug '05

Hi, I'd like to be able to specify the pid used by each MpegTSWriteStream of an MPEG transport stream. I thought I could use the sub_id field of the AVCodecContext for that. Is that ok? Here's the change I propose:

3 8

wrong! stream ts file, help me
by canbaby＠21cn.com 10 Aug '05

10 Aug '05

Hello: help me! I have a ts file that is: #file some.ts some.ts: MPEG ADTS, layer I, v1, Monaural I can use ffplay play it, I add the file to ffserver's config this: <Stream some.ts> File "/home/debwu/some.ts" </Stream> so I use: ffplay http://myip:8090/some.ts to play it, but there are a msg "Error while opening file", then exit. What's the wrong? Could some one show me a light? Thanks -canbaby-

3 4

OSX linker option error
by sms＠2BSD.COM 10 Aug '05

10 Aug '05

Hi An extraneous option (-d) crept in and causes building on OSX to fail make -C vhook all cc -fPIC -O3 -g -Wall -Wno-switch -no-cpp-precomp -pipe -fomit-frame-pointer -force_cpusubtype_ALL -Wno-sign-compare -faltivec -I.. -I'/usr/local/src/ffmpeg' -I'/usr/local/src/ffmpeg'/libavformat -I'/usr/local/src/ffmpeg'/libavcodec -DHAVE_AV_CONFIG_H `freetype-config --cflags` -c -o null.o null.c null.c: In function 'Process': null.c:59: warning: pointer targets in passing argument 2 of 'avpicture_fill' differ in signedness cc -Wl,-d,-search_paths_first -g -o null.so -dynamiclib -flat_namespace -undefined suppress null.o /usr/bin/libtool: unknown option character `d' in: -d Usage: /usr/bin/libtool -static [-] file [...] [-filelist listfile[,dirname]] [-arch_only arch] [-sacLT] Usage: /usr/bin/libtool -dynamic [-] file [...] [-filelist listfile[,dirname]] [-arch_only arch] [-o output] [-install_name name] [-compatibility_version #] [-current_version #] [-seg1addr 0x#] [-segs_read_only_addr 0x#] [-segs_read_write_addr 0x#] [-seg_addr_table <filename>] [-seg_addr_table_filename <file_system_path>] [-all_load] [-noall_load] make[1]: *** [null.so] Error 1 rm null.o make: *** [videohook] Error 2 -search_paths_first is a valid option but -d is not. The attached patch simply removes the un-needed -d Cheers, Steven Schultz

5 10

Joining Multiple Input Streams
by syllogism711＠gmail.com 05 Aug '05

05 Aug '05

First of all, is there any plan to add such a feature into ffmpeg? Second, if no one planned on doing this, could someone give me some information on what all would have to be done in order to add this? Being able to join Input streams into one output stream would definitely be a nice feature to have IMO. . . --Brad

3 2

PATCH: Build Suffix
by charles.yates＠pandora.be 04 Aug '05

04 Aug '05

Hi all, I wondered if you might consider this patch - it's a build on a previous patch to allow applications to install multiple versions of the shared libraries (in particular, it's geared toward framework usage). The usage is as follows: ./configure --enable-shared --build-suffix=.mlt make sudo make install You will then get installed versions of ffmpeg.mlt, ffplay.mlt and both libavformat and libavcodec are similarly decorated. In theory at least, it should be possible to have multiple builds without clashing (I've only tested two instances). I'm not sure about the ldconfig usage though - in the current CVS, invoking ldconfig via the sudo make install fails because it's not on the path - this patch assumes it's in /sbin which may not be correct for all distros? Anyone care to comment? Cheers, Charlie

13 35

Win32 compile
by jeffrey.s.behrbaum＠boeing.com 02 Aug '05

02 Aug '05

Hi all. Quick question. I was reading the web page regarding how to build ffmpeg for a win32 platform (http://www.salyens.com/mingw/) On that page, it list a whole multitude of config options. My question is, for the lastest source, do you still need all those config flags? Thanks to any and all for any of your help. The web page suggest the following: ./configure --extra-cflags="-mno-cygwin -mms-bitfields" --extra-ldflags="-Wl,--add-stdcall-alias" --enable-mingw32 --enable-shared --disable-a52 --disable-pp --disable-ffserver --disable-ffplay Jeff

3 3

[PATCH] libavutil - splitting parts from libavcodec
by eclipse7＠gmx.net 01 Aug '05

01 Aug '05

Hi, i moved some of the generic utility code of libavcodec to a lib of its own: libavutil. The idea is to have the generally useful stuff in an independent directory/library for making it conviniently usable from other projects. It IMHO also improves FFmpeg project's internal structure/layout. I did no functional changes to the code. I did only code moving, buildsystem changes and exporting of functions through avutil.h . To test it: 1. create a libavutil directory in FFmpeg project root 2. copy the attached C and C Header files into it Makefile avutil.h intfloat_readwrite.h intfloat_readwrite.c math.c 3. move the following files from libavcodec into it: bswap.h common.h integer.c integer.h rational.c rational.h 3. apply the attached patch It would be good to have some more testing as i only tested it on a sinlge machine so far. The pkg-config stuff is still to do. I am not familiar with pkg-config so it will take me a bit more time to do it. I am also not sure if I got the version/buildnumber stuff right. Alex (beastd)

2 9

grabbing ffmpeg output for my app
by patcito＠gmail.com 01 Aug '05

01 Aug '05

hey all, im writing a gui to ffmpeg for my app, i would like to grab the output for my progress bar. Whats the best way to read the output of mplayer using c++, is there a special API or somethiung i couldnt find it in the doc. thx in advance Patcito

2 1

H.264 decoding bug in either lavc/JM (x264 too?)
by tuukkat＠ee.oulu.fi 01 Aug '05

01 Aug '05

I encoded a video clip using x264 and decoded with both Mplayer (lavc) and JM 9.6. The two decoded videos differ slightly on some frames. I think this is bug, as H.264 should be bit exact, shouldn't it? Here's frame-by-frame comparison: Comparing size 640x480 ssd_y ssd_u ssd_v ssd psnr_y psnr_u psnr_v psnr Frame 0 0 0 0 0 inf inf inf inf Frame 1 0 0 0 0 inf inf inf inf Frame 2 0 0 0 0 inf inf inf inf Frame 3 0 0 0 0 inf inf inf inf Frame 4 9342 1 8 9351 63.3006 96.9844 87.9535 65.0573 Frame 5 0 0 0 0 inf inf inf inf Frame 6 0 0 0 0 inf inf inf inf Frame 7 0 0 0 0 inf inf inf inf Frame 8 0 0 0 0 inf inf inf inf Frame 9 0 0 0 0 inf inf inf inf Frame 10 66739 150 46 66935 54.7612 75.2235 80.3568 56.5094 Frame 11 0 0 0 0 inf inf inf inf Frame 12 0 0 0 0 inf inf inf inf Frame 13 0 0 0 0 inf inf inf inf Frame 14 0 0 0 0 inf inf inf inf Frame 15 0 0 0 0 inf inf inf inf Frame 16 668 0 4 672 74.7573 inf 90.9638 76.4922 Frame 17 0 0 0 0 inf inf inf inf Frame 18 0 0 0 0 inf inf inf inf Frame 19 69884 55 25 69964 54.5612 79.5808 83.0050 56.3172 Frame 20 0 0 0 0 inf inf inf inf Frame 21 0 0 0 0 inf inf inf inf Frame 22 11 0 0 11 92.5911 inf inf 94.3520 I used latest CVS Mplayer and latest SVN x264 (but same results were same with slightly older versions). x264 command line: --bframes 2 --8x8dct --b-pyramid --qp 26 --direct spatial The problem occurs only if using all three options --8x8dct, --b-pyramid, and --direct spatial. Most other options seem not to matter. The funny thing is that the image quality is actually a bit better with Mplayer decoded video than with JM decoded video, which would indicate that the bug would be in JM. However, as JM should be quite mature, I would rather guess there's a bug in both x264 and in Mplayer. (Tell me if it's otherwise, so I can go to bug JM developers :) Test files: http://www.ee.oulu.fi/~tuukkat/tmp/bug/ (Frame #11 below is frame 10 above, as my yuvcmp counts from 0, Mplayer apparently from 1) bug.tgz - includes all diff-fr11.png - Frame #11 difference picture, contrast enhanced diff.yuv - Difference YUV file, contrast enhanced differs.txt - Frame-by-frame PSNR between two decoded videos jm-fr11.png - Frame #11 decoded with JM 9.6 mpl-fr11.png - Frame #11 decoded with Mplayer test-640x480.yuv - Original test video test.264 - Bitstream produced with x264 test_dec-mpl.yuv - Decoded video with Mplayer test_dec.yuv - Decoded video with JM 9.6 testme.sh - Command lines used

3 6

GCC 4 Compiler warnings
by norman-k-williams＠uiowa.edu 01 Aug '05

01 Aug '05

Hi! I'm new to messing around with ffmpeg, but I have built it and have 2 questions 1. I've checked out the CVS version, and grabbed the x264 library. I built the x264 library, and installed it in a local 'target tree' -- i.e. x264 is configured with 'configure --prefix=/scratch/opt/video.' In most autoconf projects, if you define LDFLAGS before the configure you can add -L commands, in this case I'd need -L/scratch/opt/video/lib added to the ffpmeg link line before -lx264, but the configure/Makefile for ffmpeg ignore LDFLAGS. So the question is WHERE is x264 supposed to live surch that ffmpeg can link to it? 2. There are several hundred warnings reported by gcc4. Most of them can be silenced with trivial changes that won't affect the code generated. Does anyone care about these? Should I try and fix them and submit a patch?

5 6