Sur d’autres sites (11962)

• FFmpeg library is detecting pixel format of VP9 video stream not correctly

  6 avril 2023, par Pikachu

  I am using a C code to detect pixel format of VP9 video stream in a WebM container. FFmpeg version 6.0, full shared library build, downloaded from official website. Operating system is Windows 10. I feed the library with a VP9 video encoded with alpha channel, pixel format is YUVA420p. It detects pixel format as YUV420p.

  


  I have found a similar question on StackOverflow.com, Is there a way to force FFMPEG to decode a video stream with alpha from ​a WebM video encoded with libvpx-vp9 ?, but it does not actually help.

  


  When I override the decoder with a libvpx, it continues to detect the pixel format as YUV420p instead of YUVA420p.

  


  C code is following. Note that error handling in code is omitted here for StackOverflow question to be shorter.

  


  AVFormatContext *fmt_ctx = NULL;
int err = avformat_open_input(&fmt_ctx, infp, NULL, NULL);
err = avformat_find_stream_info(fmt_ctx, NULL);
int stream = av_find_best_stream(fmt_ctx, AVMEDIA_TYPE_VIDEO, -1, -1, NULL, 0);
AVCodecParameters *codecpar = fmt_ctx->streams[stream]->codecpar;

const AVCodec *codec = NULL;
if (codecpar->codec_id == AV_CODEC_ID_VP9) {
    codec = avcodec_find_decoder_by_name(CODEC_LIBVPX_VP9);
} else {
    codec = avcodec_find_decoder(codecpar->codec_id);
}

AVCodecContext *ctx = avcodec_alloc_context3(codec);
err = avcodec_parameters_to_context(ctx, codecpar);
av_log(NULL, AV_LOG_DEBUG, "Pixel format: %d.\n", ctx->pix_fmt); //TODO:DEBUG.
err = avcodec_open2(ctx, codec, NULL);


  


  The program tells Pixel format: 0., which means AV_PIX_FMT_YUV420P, not the AV_PIX_FMT_YUVA420P !

  


  If I override pixel format manually, I am able to decode video with alpha channel and to see the transparent background, but it breaks the logics, because when a real YUV420p pixel format comes in and gets overridden by YUVA420p, this will be a problem.

  


  if (codecpar->codec_id == AV_CODEC_ID_VP9) {
    if (strcmp(codec->name, CODEC_LIBVPX_VP9) == 0) {
        if (ctx->pix_fmt == AV_PIX_FMT_YUV420P) {
            ctx->pix_fmt = AV_PIX_FMT_YUVA420P;
        }
    }
}


  


  At the same time ffmpeg tool started from command line with libvpx decoder tells that my video has YUVA420p pixel format. Output is following.

  


  D:\Temp\4>ffmpeg -c:v libvpx-vp9 -i yuva.webm
ffmpeg version 6.0-full_build-www.gyan.dev Copyright (c) 2000-2023 the FFmpeg developers
  built with gcc 12.2.0 (Rev10, Built by MSYS2 project)
  configuration: --enable-gpl --enable-version3 --enable-shared --disable-w32threads --disable-autodetect --enable-fontconfig --enable-iconv --enable-gnutls --enable-libxml2 --enable-gmp --enable-bzlib --enable-lzma --enable-libsnappy --enable-zlib --enable-librist --enable-libsrt --enable-libssh --enable-libzmq --enable-avisynth --enable-libbluray --enable-libcaca --enable-sdl2 --enable-libaribb24 --enable-libdav1d --enable-libdavs2 --enable-libuavs3d --enable-libzvbi --enable-librav1e --enable-libsvtav1 --enable-libwebp --enable-libx264 --enable-libx265 --enable-libxavs2 --enable-libxvid --enable-libaom --enable-libjxl --enable-libopenjpeg --enable-libvpx --enable-mediafoundation --enable-libass --enable-frei0r --enable-libfreetype --enable-libfribidi --enable-liblensfun --enable-libvidstab --enable-libvmaf --enable-libzimg --enable-amf --enable-cuda-llvm --enable-cuvid --enable-ffnvcodec --enable-nvdec --enable-nvenc --enable-d3d11va --enable-dxva2 --enable-libvpl --enable-libshaderc --enable-vulkan --enable-libplacebo --enable-opencl --enable-libcdio --enable-libgme --enable-libmodplug --enable-libopenmpt --enable-libopencore-amrwb --enable-libmp3lame --enable-libshine --enable-libtheora --enable-libtwolame --enable-libvo-amrwbenc --enable-libilbc --enable-libgsm --enable-libopencore-amrnb --enable-libopus --enable-libspeex --enable-libvorbis --enable-ladspa --enable-libbs2b --enable-libflite --enable-libmysofa --enable-librubberband --enable-libsoxr --enable-chromaprint
  libavutil      58.  2.100 / 58.  2.100
  libavcodec     60.  3.100 / 60.  3.100
  libavformat    60.  3.100 / 60.  3.100
  libavdevice    60.  1.100 / 60.  1.100
  libavfilter     9.  3.100 /  9.  3.100
  libswscale      7.  1.100 /  7.  1.100
  libswresample   4. 10.100 /  4. 10.100
  libpostproc    57.  1.100 / 57.  1.100
[libvpx-vp9 @ 000001ecdf6002c0] v1.13.0-71-g45dc0d34d
    Last message repeated 1 times
Input #0, matroska,webm, from 'yuva.webm':
  Metadata:
    ENCODER         : Lavf60.3.100
  Duration: 00:00:05.55, start: 0.000000, bitrate: 227 kb/s
  Stream #0:0: Video: vp9 (Profile 0), yuva420p(tv, progressive), 1920x1080, SAR 1:1 DAR 16:9, 60 fps, 60 tbr, 1k tbn
    Metadata:
      alpha_mode      : 1
      ENCODER         : Lavc60.3.100 libvpx-vp9
      DURATION        : 00:00:05.550000000
At least one output file must be specified


  


  Here is my YUVA420p in the first video stream at the end of the console output :

  


  Stream #0:0: Video: vp9 (Profile 0), yuva420p(tv, progressive), 1920x1080, SAR 1:1 DAR 16:9, 60 fps, 60 tbr, 1k tbn


  


  


  The questions are following.

  


  

  1. How to detect real pixel format of VP9 video with FFmpeg library in C code reliably ?

  


  2. Why is the C code not detecting the actual pixel format even with codec overriden to libvpx ?

  


  


  Thank you.

  


• Export frames/images from compressed video

  24 mars 2015, par Jan Viehweger

  I have a compressed movie (mp4) and I want to extract every single frame / image from it. I know that each individual frame of the video only contains the changed pixels regarding to the last keyframe, because of the video compression. But that is exactly what I want. I just want to see those differences. I want to visualy see how the compressor works.

  Is there some tool like imagemagick out there what can things like that ?

• Feeding a series of images to ffmpeg as each image is created [closed]

  5 février 2013, par Mark Schneider

  I'm trying to use ffmpeg to build a 1280x720 slide-show from a sequence of pictures and videos, but I have concerns about potential disk I/O bottleneck.

  I expect a typical slide-show to have about 50 pictures and 2-3 videos (10-15 seconds each at 30 fps). I would like to show each picture for 3-4 seconds (possibly with a
  Ken Burns effect) with a smooth 2 second crossfade between each set of pictures (or for pictures adjacent to videos - between the picture and the first/last frame of the video).

  Given about 50 pictures, the crossfades alone would amount to about 3,000 images (50 transitions x 2 secs/transition x 30 fps). And I suppose if I implement a Ken Burns effect during each picture's 3-4 second showing, I'd have to provide ffmpeg with individual images for each of those frames. (I'm writing a script in Ruby that will pull a list of images from a database and in turn call ImageMagick to create the individual images for each frame. As I understand it, the RMagick library interfaces with ImageMagick such that the output images come back as in-memory objects without needing to write to disk. FWIW, I'm developing in Windows 8 and will deploy to Heroku.)

  All of the slideshow examples I've found online feed ffmpeg a set of images which have already been created. However, in an effort to avoid waiting on considerable disk I/O, I'd like to feed each image to ffmpeg as the image is created rather than create them all in advance.

  Is there a way to send each image file to ffmpeg on the fly as the file is created in memory ?