Recherche avancée

Sur d’autres sites (5325)

• ffmpeg image to video sequence stops after 10

  18 mai 2012, par Jim

  Probably a fast anwser, but my first time trying to do this. I have 77 jpeg pictures, I have renamed them to be 000.jpg - 076.jpg. The next step is using ffmpeg and here is my command statement :

  ffmpeg -f image2 -r .1 -i %d.jpg -r 25 test.avi

  It creates an avi with 10 images for 10seconds each, why only 10 ? I've tried other iterations of %d.jpg in the ffmpeg command with no success.

  thanks for the help all !

  -Jim

  Here is the listing of the directory with the images :

  ls -l
  
  total 77472
  
  -rwxr-xr-x 1 jim jim 2329065 May 17 16:31 000.jpg
  
  -rwxr-xr-x 1 jim jim  716563 May 17 16:31 001.jpg
  
  -rwxr-xr-x 1 jim jim  716626 May 17 16:31 002.jpg
  
  -rwxr-xr-x 1 jim jim  726686 May 17 16:31 003.jpg
  
  -rwxr-xr-x 1 jim jim  739312 May 17 16:31 004.jpg
  
  -rwxr-xr-x 1 jim jim  720249 May 17 16:31 005.jpg
  
  -rwxr-xr-x 1 jim jim  666757 May 17 16:31 006.jpg
  
  -rwxr-xr-x 1 jim jim  656259 May 17 16:31 007.jpg
  
  -rwxr-xr-x 1 jim jim  664960 May 17 16:31 008.jpg
  
  -rwxr-xr-x 1 jim jim  740801 May 17 16:31 009.jpg
  
  -rwxr-xr-x 1 jim jim  882502 May 17 16:31 010.jpg
  
  -rwxr-xr-x 1 jim jim  631117 May 17 16:31 011.jpg
  
  -rwxr-xr-x 1 jim jim  730331 May 17 16:31 018.jpg
  
  -rwxr-xr-x 1 jim jim  725132 May 17 16:31 019.jpg
  
  -rwxr-xr-x 1 jim jim  729626 May 17 16:31 020.jpg
  
  -rwxr-xr-x 1 jim jim  731980 May 17 16:31 021.jpg
  
  -rwxr-xr-x 1 jim jim  671597 May 17 16:31 022.jpg
  
  -rwxr-xr-x 1 jim jim  681978 May 17 16:31 023.jpg
  
  -rwxr-xr-x 1 jim jim  686600 May 17 16:31 024.jpg
  
  -rwxr-xr-x 1 jim jim  675316 May 17 16:31 025.jpg
  
  -rwxr-xr-x 1 jim jim  681826 May 17 16:31 026.jpg
  
  -rwxr-xr-x 1 jim jim  740998 May 17 16:31 027.jpg
  
  -rwxr-xr-x 1 jim jim  568480 May 17 16:31 028.jpg
  
  -rwxr-xr-x 1 jim jim  747400 May 17 16:31 029.jpg
  
  -rwxr-xr-x 1 jim jim  630995 May 17 16:31 030.jpg
  
  -rwxr-xr-x 1 jim jim  689926 May 17 16:31 031.jpg
  
  -rwxr-xr-x 1 jim jim  685054 May 17 16:31 032.jpg
  
  -rwxr-xr-x 1 jim jim  710620 May 17 16:31 033.jpg
  
  -rwxr-xr-x 1 jim jim  658365 May 17 16:31 034.jpg
  
  -rwxr-xr-x 1 jim jim  657037 May 17 16:31 035.jpg
  
  -rwxr-xr-x 1 jim jim  772135 May 17 16:31 036.jpg
  
  -rwxr-xr-x 1 jim jim  741759 May 17 16:31 037.jpg
  
  -rwxr-xr-x 1 jim jim  807470 May 17 16:31 038.jpg
  
  -rwxr-xr-x 1 jim jim  748423 May 17 16:31 039.jpg
  
  -rwxr-xr-x 1 jim jim  712377 May 17 16:31 040.jpg
  
  -rwxr-xr-x 1 jim jim  715804 May 17 16:31 041.jpg
  
  -rwxr-xr-x 1 jim jim  701025 May 17 16:31 042.jpg
  
  -rwxr-xr-x 1 jim jim  759446 May 17 16:31 043.jpg
  
  -rwxr-xr-x 1 jim jim  621801 May 17 16:31 044.jpg
  
  -rwxr-xr-x 1 jim jim  720843 May 17 16:31 045.jpg
  
  -rwxr-xr-x 1 jim jim  704002 May 17 16:31 046.jpg
  
  -rwxr-xr-x 1 jim jim  696075 May 17 16:31 047.jpg
  
  -rwxr-xr-x 1 jim jim  723685 May 17 16:31 048.jpg
  
  -rwxr-xr-x 1 jim jim  732332 May 17 16:31 049.jpg
  
  -rwxr-xr-x 1 jim jim  747235 May 17 16:31 050.jpg
  
  -rwxr-xr-x 1 jim jim  883655 May 17 16:31 051.jpg
  
  -rwxr-xr-x 1 jim jim 1750723 May 17 16:31 052.jpg
  
  -rwxr-xr-x 1 jim jim 1002588 May 17 16:31 053.jpg
  
  -rwxr-xr-x 1 jim jim  540666 May 17 16:31 054.jpg
  
  -rwxr-xr-x 1 jim jim 1876002 May 17 16:31 055.jpg
  
  -rwxr-xr-x 1 jim jim 1893761 May 17 16:31 056.jpg
  
  -rwxr-xr-x 1 jim jim 1979442 May 17 16:31 057.jpg
  
  -rwxr-xr-x 1 jim jim 1766249 May 17 16:31 058.jpg
  
  -rwxr-xr-x 1 jim jim 2085989 May 17 16:31 059.jpg
  
  -rwxr-xr-x 1 jim jim  883871 May 17 16:31 060.jpg
  
  -rwxr-xr-x 1 jim jim  755714 May 17 16:31 061.jpg
  
  -rwxr-xr-x 1 jim jim  797146 May 17 16:31 062.jpg
  
  -rwxr-xr-x 1 jim jim 2431531 May 17 16:31 065.jpg
  
  -rwxr-xr-x 1 jim jim 2413333 May 17 16:31 066.jpg
  
  -rwxr-xr-x 1 jim jim 2449278 May 17 16:31 067.jpg
  
  -rwxr-xr-x 1 jim jim 2458183 May 17 16:31 068.jpg
  
  -rwxr-xr-x 1 jim jim 2514419 May 17 16:31 069.jpg
  
  -rwxr-xr-x 1 jim jim 2477737 May 17 16:31 070.jpg
  
  -rwxr-xr-x 1 jim jim 2471347 May 17 16:31 071.jpg
  
  -rwxr-xr-x 1 jim jim 2384936 May 17 16:31 072.jpg
  
  -rwxr-xr-x 1 jim jim 2459983 May 17 16:31 073.jpg
  
  -rwxr-xr-x 1 jim jim 2501286 May 17 16:31 074.jpg
  
  -rwxr-xr-x 1 jim jim 2367710 May 17 16:31 075.jpg
  
  -rwxr-xr-x 1 jim jim 2455564 May 17 16:31 076.jpg

  FFMPEG Command and OUTPUT :

  ffmpeg -v verbose -f image2 -r .1 -i %03d.jpg -r 25 test.avi
  
  ffmpeg version 0.8.1-4:0.8.1-0ubuntu1, Copyright (c) 2000-2011 the Libav developers
  
    built on Mar 22 2012 05:09:06 with gcc 4.6.3
  
    configuration: --extra-version='4:0.8.1-0ubuntu1' --arch=amd64 --prefix=/usr --libdir=/usr/lib/x86_64-linux-gnu --enable-vdpau --enable-bzlib --enable-libgsm --enable-libschroedinger --enable-libspeex --enable-libtheora --enable-libvorbis --enable-pthreads --enable-zlib --enable-libvpx --enable-runtime-cpudetect --enable-libfreetype --enable-vaapi --enable-gpl --enable-postproc --enable-swscale --enable-x11grab --enable-libdc1394 --shlibdir=/usr/lib/x86_64-linux-gnu --enable-shared --disable-static
  
    avutil      configuration: --extra-version='4:0.8.1ubuntu1+medibuntu1' --arch=amd64 --prefix=/usr --libdir=/usr/lib/x86_64-linux-gnu --enable-vdpau --enable-bzlib --enable-libgsm --enable-libschroedinger --enable-libspeex --enable-libtheora --enable-libvorbis --enable-pthreads --enable-zlib --enable-libvpx --enable-runtime-cpudetect --enable-libopencore-amrnb --enable-libopencore-amrwb --enable-version3 --enable-libfreetype --enable-vaapi --enable-libopenjpeg --enable-libfaac --enable-nonfree --enable-gpl --enable-postproc --enable-swscale --enable-x11grab --enable-libdirac --enable-libmp3lame --enable-librtmp --enable-libx264 --enable-libxvid --enable-libopencore-amrnb --enable-version3 --enable-libopencore-amrwb --enable-version3 --enable-libvo-aacenc --enable-version3 --enable-libvo-amrwbenc --enable-version3 --enable-libdc1394 --shlibdir=/usr/lib/x86_64-linux-gnu --enable-shared --disable-static
  
    avcodec     configuration: --extra-version='4:0.8.1ubuntu1+medibuntu1' --arch=amd64 --prefix=/usr --libdir=/usr/lib/x86_64-linux-gnu --enable-vdpau --enable-bzlib --enable-libgsm --enable-libschroedinger --enable-libspeex --enable-libtheora --enable-libvorbis --enable-pthreads --enable-zlib --enable-libvpx --enable-runtime-cpudetect --enable-libopencore-amrnb --enable-libopencore-amrwb --enable-version3 --enable-libfreetype --enable-vaapi --enable-libopenjpeg --enable-libfaac --enable-nonfree --enable-gpl --enable-postproc --enable-swscale --enable-x11grab --enable-libdirac --enable-libmp3lame --enable-librtmp --enable-libx264 --enable-libxvid --enable-libopencore-amrnb --enable-version3 --enable-libopencore-amrwb --enable-version3 --enable-libvo-aacenc --enable-version3 --enable-libvo-amrwbenc --enable-version3 --enable-libdc1394 --shlibdir=/usr/lib/x86_64-linux-gnu --enable-shared --disable-static
  
    libavutil    51. 22. 1 / 51. 22. 1
  
    libavcodec   53. 35. 0 / 53. 35. 0
  
    libavformat  53. 21. 0 / 53. 21. 0
  
    libavdevice  53.  2. 0 / 53.  2. 0
  
    libavfilter   2. 15. 0 /  2. 15. 0
  
    libswscale    2.  1. 0 /  2.  1. 0
  
    libpostproc  52.  0. 0 / 52.  0. 0
  
  This program is not developed anymore and is only provided for compatibility. Use avconv instead (see Changelog for the list of incompatible changes).
  
  [image2 @ 0x1bef9c0] max_analyze_duration reached
  
  
  
  Seems stream 0 codec frame rate differs from container frame rate: 0.10 (1/10) -> 0.50 (1/2)
  
  Input #0, image2, from '%03d.jpg':
  
    Duration: 00:02:00.00, start: 0.000000, bitrate: N/A
  
      Stream #0.0: Video: mjpeg, yuvj440p, 1920x2560, 0.10 fps, 0.50 tbr, 0.10 tbn, 0.10 tbc
  
  File 'test.avi' already exists. Overwrite ? [y/N] y
  
  Incompatible pixel format 'yuvj440p' for codec 'mpeg4', auto-selecting format 'yuv420p'
  
  [buffer @ 0x1bf0100] w:1920 h:2560 pixfmt:yuvj440p
  
  [avsink @ 0x1bf13c0] auto-inserting filter 'auto-inserted scaler 0' between the filter 'src' and the filter 'out'
  
  [scale @ 0x1bf1ae0] w:1920 h:2560 fmt:yuvj440p -> w:1920 h:2560 fmt:yuv420p flags:0x4
  
  Output #0, avi, to 'test.avi':
  
    Metadata:
  
      ISFT            : Lavf53.21.0
  
      Stream #0.0: Video: mpeg4, yuv420p, 1920x2560, q=2-31, 200 kb/s, 25 tbn, 25 tbc
  
  Stream mapping:
  
    Stream #0.0 -> #0.0
  
  Press ctrl-c to stop encoding
  
  [buffer @ 0x1bf0100] Changing frame properties on the fly is not supported.
  
      Last message repeated 10 times
  
  frame=    1 fps=  0 q=5.7 Lsize=      97kB time=0.04 bitrate=19823.2kbits/s    
  
  video:91kB audio:0kB global headers:0kB muxing overhead 6.106282%

  Now test.avi is only 99Kbs and doesn't play anything.

• so Confused, why my build of libffmpeg.so > 17M ?

  24 février 2012, par ghost

  I did build ffmpeg for Android in winxp and scientific linux , ffmpeg is in dolphin player — an open source video player (http://code.google.com/p/dolphin-player/), and i just build the ffmpeg, its seems like the same as in rockplayer 1.7.0, they all use build_andriod.sh below， it worked in both winxp and linux,
  and all successfully got bin/ffmpeg (less than 5MB), but libffmpeg.so ( > 17MB)， when put libffmpeg.so in dolphin-player libs , player can't work, the size 17MB is too large, the original libffmpeg.so in olphin-player libs is less than 5MB, please give some advice.

  #!/bin/bash
  
  
  
  ######################################################
  
  # FFmpeg builds script for Android+ARM platform
  
  #
  
  # This script is released under term of 
  
  #   CDDL (http://www.opensource.org/licenses/cddl1) 
  
  # Wrote by pinxue (~@gmail.com) from RockPlayer.com
  
  #                                   2010-8 ~ 2011-4
  
  ######################################################
  
  
  
  ######################################################
  
  # Usage:
  
  #   put this script in top of FFmpeg source tree
  
  #   ./build_android
  
  #
  
  # It generates binary for following architectures:
  
  #     ARMv6 
  
  #     ARMv6+VFP 
  
  #     ARMv7+VFM-ïd16 (Tegra2) 
  
  #     ARMv7+Neon (Cortex-A8)
  
  #
  
  # Customizing:
  
  # 1. Feel free to change ./configure parameters for more features
  
  # 2. To adapt other ARM variants
  
  #       set $CPU and $OPTIMIZE_CFLAGS 
  
  #       call build_one
  
  ######################################################
  
  
  
  export TMPDIR=D:/tmp/android
  
  export NDK=D:/android-ndk-r4
  
  
  
  #PLATFORM=$NDK/build/platforms/android-8/arch-arm/
  
  PLATFORM=$NDK/build/platforms/android-8/arch-arm
  
  #PREBUILT=$NDK/build/prebuilt/darwin-x86/arm-eabi-4.4.0
  
  PREBUILT=$NDK/build/prebuilt/windows/arm-eabi-4.4.0
  
  
  
  function build_one
  
  {
  
  
  
  # -fasm : required. Android header file uses asm keyword instead of __asm__ , but most of c dialect (like ansi,c99,gnu99) implies -fno-asm.
  
  #   ~/android/android-ndk-r4/build/platforms/android-5/arch-arm//usr/include/asm/byteorder.h: In function '___arch__swab32':
  
  #   ~/android/android-ndk-r4/build/platforms/android-5/arch-arm//usr/include/asm/byteorder.h:25: error: expected ')' before ':' token
  
  
  
  # -fno-short-enums : optimized.  Else FFmpeg obj will generate a huge number of warning for variable-size enums, 
  
  #   though we may suppress them by --no-enum-size-warning, it would be better to avoid it.
  
  #   .../ld: warning: cmdutils.o uses variable-size enums yet the output is to use 32-bit enums; use of enum values across objects may fail
  
  
  
  # --extra-libs="-lgcc" : required. Else cannot solve some runtime function symbols
  
  #   ... undefined reference to `__aeabi_f2uiz'
  
  
  
  # --enable-protocols : required. Without this option, the file open always fails mysteriously.
  
  #   FFmpeg's av_open_input_file will invoke file format probing functions, but because most of useful demuxers has flag of zero 
  
  #   which cause them are ignored during file format probling and fall to url stream parsing,
  
  #   if protocols are disabled, the file:// url cannot be opened as well.
  
  
  
  # $PREBUILT/bin/arm-eabi-ar d libavcodec/libavcodec.a inverse.o : required.
  
  #   FFmpeg includes two copies of inverse.c both in libavutil and libavcodec for performance consideration (not sure the benifit yet)
  
  #   Without this step, final ld of generating libffmpeg.so will fail silently, if invoke ld through gcc, gcc will collect more reasonable error message.
  
  
  
  # -llog: debug only, FFmpeg itself doesn't require it at all.
  
  #   With this option, we may simply includes "utils/Log.h" and use LOGx() to observe FFmpeg's behavior
  
  #   PS, it seems the toolchain implies -DNDEBUG somewhere, it would be safer to use following syntax
  
  #    #ifdef NDEBUG
  
  #        #undef NDEBUG
  
  #        #define HAVE_NDEBUG
  
  #    #endif
  
  #    #include "utils/Log.h"
  
  #    #ifdef HAVE_NDEBUG
  
  #        #define NDEBUG
  
  #        #undef HAVE_NDEBUG
  
  #    #endif
  
  
  
  # --whole-archive : required. Else ld generate a small .so file (about 15k)
  
  
  
  # --no-stdlib : required. Android doesn't use standard c runtime but invited its own wheal (bionic libc) because of license consideration.
  
  
  
  # space before \ of configure lines: required for some options. Else next line will be merged into previous lines's content and cause problem.
  
  #   Especially the --extra-cflags, the next line will pass to gcc in this case and configure will say gcc cannot create executable.
  
  
  
  # many options mentioned by articles over internet are implied by -O2 or -O3 already, need not repeat at all.
  
  
  
  # two or three common optimization cflags are omitted because not sure about the trade off yet. invoke NDK build system with V=1 to find them.
  
  
  
  # -Wl,-T,$PREBUILT/arm-eabi/lib/ldscripts/armelf.x mentioned by almost every articles over internet, but it is not required to specify at all.
  
  
  
  # -Dipv6mr_interface=ipv6mr_ifindex : required. Android inet header doesn't use ipv6mr_interface which is required by rfc, seems it generate this user space header file directly from kernel header file, but Linux kernel has decided to keep its own name for ever and ask user space header to use rfc name.
  
  
  
  # HAVE_SYS_UIO_H : required. Else:
  
  # In file included from ~/android/android-ndk-r4/build/platforms/android-5/arch-arm//usr/include/linux/socket.h:29,
  
  #                 from ~/android/android-ndk-r4/build/platforms/android-5/arch-arm//usr/include/sys/socket.h:33,
  
  #                 from libavformat/network.h:35,
  
  #                 from libavformat/utils.c:46:
  
  #~/android/android-ndk-r4/build/platforms/android-5/arch-arm//usr/include/linux/uio.h:19: error: redefinition of 'struct iovec'
  
  #
  
  
  
  # --disable-doc : required because of strange bug of toolchain.
  
  #
  
  
  
  #
  
  #--extra-ldflags=-Wl,-T,$PREBUILT/arm-eabi/lib/ldscripts/armelf.x -Wl,-rpath-link=$PLATFORM/usr/lib -L$PLATFORM/usr/lib -nostdlib $PREBUILT/lib/gcc/arm-eabi/4.4.0/crtbegin.o $PREBUILT/lib/gcc/arm-eabi/4.4.0/crtend.o -lc -lm -ldl"
  
  #
  
  
  
  ./configure --target-os=linux \
  
      --prefix=$PREFIX \
  
      --enable-cross-compile \
  
      --extra-libs="-lgcc" \
  
      --arch=arm \
  
      --cc=$PREBUILT/bin/arm-eabi-gcc \
  
      --cross-prefix=$PREBUILT/bin/arm-eabi- \
  
      --nm=$PREBUILT/bin/arm-eabi-nm \
  
      --sysroot=$PLATFORM \
  
     --extra-cflags=" -O3 -fpic -DANDROID -DHAVE_SYS_UIO_H=1 -Dipv6mr_interface=ipv6mr_ifindex -fasm -Wno-psabi -fno-short-enums  -fno-strict-aliasing -finline-limit=300 $OPTIMIZE_CFLAGS " \
  
      --disable-shared \
  
      --enable-static \
  
      --extra-ldflags="-Wl,-rpath-link=$PLATFORM/usr/lib -L$PLATFORM/usr/lib  -nostdlib -lc -lm -ldl -llog" \
  
      --enable-parsers \
  
      --disable-encoders  \
  
      --enable-decoders \
  
      --disable-muxers \
  
      --enable-demuxers \
  
      --enable-swscale  \
  
      --disable-ffplay \
  
      --disable-ffprobe \
  
      --disable-ffserver \
  
      --enable-network \
  
      --enable-indevs \
  
      --disable-bsfs \
  
      --disable-filters \
  
      --enable-protocols  \
  
      --enable-asm \
  
      --disable-doc \
  
      $ADDITIONAL_CONFIGURE_FLAG
  
  
  
  ##make clean
  
  make  -j4 install
  
  
  
  
  
  $PREBUILT/bin/arm-eabi-ar d libavcodec/libavcodec.a inverse.o
  
  
  
  $PREBUILT/bin/arm-eabi-ld -rpath-link=$PLATFORM/usr/lib -L$PLATFORM/usr/lib  -soname libffmpeg.so -shared -nostdlib  -z,noexecstack -Bsymbolic --whole-archive --no-undefined -o $PREFIX/libffmpeg.so libavcodec/libavcodec.a libavformat/libavformat.a libavutil/libavutil.a  -lc -lm -lz -ldl -llog  --warn-once  --dynamic-linker=/system/bin/linker $PREBUILT/lib/gcc/arm-eabi/4.4.0/libgcc.a
  
  
  
  }
  
  
  
  
  
  #arm v6
  
  CPU=armv6
  
  OPTIMIZE_CFLAGS="-marm -march=$CPU"
  
  PREFIX=./android/$CPU 
  
  ADDITIONAL_CONFIGURE_FLAG=
  
  build_one
  
  
  
  #arm v7vfpv3
  
  CPU=armv7-a
  
  OPTIMIZE_CFLAGS="-mfloat-abi=softfp -mfpu=vfpv3-d16 -marm -march=$CPU "
  
  PREFIX=./android/$CPU 
  
  ADDITIONAL_CONFIGURE_FLAG=
  
  build_one
  
  
  
  #arm v7vfp
  
  CPU=armv7-a
  
  OPTIMIZE_CFLAGS="-mfloat-abi=softfp -mfpu=vfp -marm -march=$CPU "
  
  PREFIX=./android/$CPU-vfp
  
  ADDITIONAL_CONFIGURE_FLAG=
  
  build_one
  
  
  
  #arm v7n
  
  CPU=armv7-a
  
  OPTIMIZE_CFLAGS="-mfloat-abi=softfp -mfpu=neon -marm -march=$CPU -mtune=cortex-a8"
  
  PREFIX=./android/$CPU-neon 
  
  ADDITIONAL_CONFIGURE_FLAG=--enable-neon
  
  build_one
  
  
  
  #arm v6+vfp
  
  CPU=armv6
  
  OPTIMIZE_CFLAGS="-DCMP_HAVE_VFP -mfloat-abi=softfp -mfpu=vfp -marm -march=$CPU"
  
  PREFIX=./android/${CPU}_vfp 
  
  ADDITIONAL_CONFIGURE_FLAG=
  
  build_one

• Simply beyond ridiculous

  7 mai 2010, par Dark Shikari — H.265, speed

  For the past few years, various improvements on H.264 have been periodically proposed, ranging from larger transforms to better intra prediction. These finally came together in the JCT-VC meeting this past April, where over two dozen proposals were made for a next-generation video coding standard. Of course, all of these were in very rough-draft form ; it will likely take years to filter it down into a usable standard. In the process, they’ll pick the most useful features (hopefully) from each proposal and combine them into something a bit more sane. But, of course, it all has to start somewhere.

  A number of features were common : larger block sizes, larger transform sizes, fancier interpolation filters, improved intra prediction schemes, improved motion vector prediction, increased internal bit depth, new entropy coding schemes, and so forth. A lot of these are potentially quite promising and resolve a lot of complaints I’ve had about H.264, so I decided to try out the proposal that appeared the most interesting : the Samsung+BBC proposal (A124), which claims compression improvements of around 40%.

  The proposal combines a bouillabaisse of new features, ranging from a 12-tap interpolation filter to 12thpel motion compensation and transforms as large as 64×64. Overall, I would say it’s a good proposal and I don’t doubt their results given the sheer volume of useful features they’ve dumped into it. I was a bit worried about complexity, however, as 12-tap interpolation filters don’t exactly scream “fast”.

  I prepared myself for the slowness of an unoptimized encoder implementation, compiled their tool, and started a test encode with their recommended settings.

  I waited. The first frame, an I-frame, completed.

  I took a nap.

  I waited. The second frame, a P-frame, was done.

  I played a game of Settlers.

  I waited. The third frame, a B-frame, was done.

  I worked on a term paper.

  I waited. The fourth frame, a B-frame, was done.

  After a full 6 hours, 8 frames had encoded. Yes, at this rate, it would take a full two weeks to encode 10 seconds of HD video. On a Core i7. This is not merely slow ; this is over 1000 times slower than x264 on “placebo” mode. This is so slow that it is not merely impractical ; it is impossible to even test. This encoder is apparently designed for some sort of hypothetical future computer from space. And word from other developers is that the Intel proposal is even slower.

  This has led me to suspect that there is a great deal of cheating going on in the H.265 proposals. The goal of the proposals, of course, is to pick the best feature set for the next generation video compression standard. But there is an extra motivation : organizations whose features get accepted get patents on the resulting standard, and thus income. With such large sums of money in the picture, dishonesty becomes all the more profitable.

  There is a set of rules, of course, to limit how the proposals can optimize their encoders. If different encoders use different optimization techniques, the results will no longer be comparable — remember, they are trying to compare compression features, not methods of optimizing encoder-side decisions. Thus all encoders are required to use a constant quantizer, specified frame types, and so forth. But there are no limits on how slow an encoder can be or what algorithms it can use.

  It would be one thing if the proposed encoder was a mere 10 times slower than the current reference ; that would be reasonable, given the low level of optimization and higher complexity of the new standard. But this is beyond ridiculous. With the prize given to whoever can eke out the most PSNR at a given quantizer at the lowest bitrate (with no limits on speed), we’re just going to get an arms race of slow encoders, with every company trying to use the most ridiculous optimizations possible, even if they involve encoding the frame 100,000 times over to choose the optimal parameters. And the end result will be as I encountered here : encoders so slow that they are simply impossible to even test.

  Such an arms race certainly does little good in optimizing for reality where we don’t have 30 years to encode an HD movie : a feature that gives great compression improvements is useless if it’s impossible to optimize for in a reasonable amount of time. Certainly once the standard is finalized practical encoders will be written — but it makes no sense to optimize the standard for a use-case that doesn’t exist. And even attempting to “optimize” anything is difficult when encoding a few seconds of video takes weeks.

  Update : The people involved have contacted me and insist that there was in fact no cheating going on. This is probably correct ; the problem appears to be that the rules that were set out were simply not strict enough, making many changes that I would intuitively consider “cheating” to be perfectly allowed, and thus everyone can do it.

  I would like to apologize if I implied that the results weren’t valid ; they are — the Samsung-BBC proposal is definitely one of the best, which is why I picked it to test with. It’s just that I think any situation in which it’s impossible to test your own software is unreasonable, and thus the entire situation is an inherently broken one, given the lax rules, slow baseline encoder, and no restrictions on compute time.