Recherche avancée

Sur d’autres sites (5510)

• How extract JPEG image from H264 stream in constant time

  27 août 2021, par Ross Gardiner

  I want to extract a JPEG frame from a H264 stream on disk. The extraction needs to be as fast as possible for my real-time requirements.

  


  Until now I have been using ffmpeg-python lib which is just a python wrapper for ffmpeg. Here is a code snippet :

  


  out, _ = (
    ffmpeg
    .input('./5sec.h264')
    .filter('select', 'gte(n,{})'.format(144))
    .output('pipe:', vframes=1, format='image2', vcodec='h264')
    .run(capture_stdout=True)
)


  


  This outputs the jpeg to stdout, with some effort I could read this into my program.

  


  However, as I use larger and larger stream files the extraction time to grab the JPEG increases. I thought lookup time would be constant as ffmpeg is highly optimised ?

  


  Is there a constant time solution to lookup and return a frame from a h264 (or even mjpeg) format stream on disk ?

  


  Edit :
Heres the command I use without the python wrapper :
ffmpeg -i 5sec.h264 -frames:v 1 -filter:v "select=gte(n\,25)" -f image2 frame.jpg

  


  here's output :

  


  ffmpeg version 4.1.6-1~deb10u1+rpt2 Copyright (c) 2000-2020 the FFmpeg developers
  built with gcc 8 (Raspbian 8.3.0-6+rpi1)
  configuration: --prefix=/usr --extra-version='1~deb10u1+rpt2' --toolchain=hardened --incdir=/usr/include/arm-linux-gnueabihf --enable-gpl --disable-stripping --enable-avresample --disable-filter=resample --enable-avisynth --enable-gnutls --enable-ladspa --enable-libaom --enable-libass --enable-libbluray --enable-libbs2b --enable-libcaca --enable-libcdio --enable-libcodec2 --enable-libflite --enable-libfontconfig --enable-libfreetype --enable-libfribidi --enable-libgme --enable-libgsm --enable-libjack --enable-libmp3lame --enable-libmysofa --enable-libopenjpeg --enable-libopenmpt --enable-libopus --enable-libpulse --enable-librsvg --enable-librubberband --enable-libshine --enable-libsnappy --enable-libsoxr --enable-libspeex --enable-libssh --enable-libtheora --enable-libtwolame --enable-libvidstab --enable-libvorbis --enable-libvpx --enable-libwavpack --enable-libwebp --enable-libx265 --enable-libxml2 --enable-libxvid --enable-libzmq --enable-libzvbi --enable-lv2 --enable-omx --enable-openal --enable-opengl --enable-sdl2 --enable-omx-rpi --enable-mmal --enable-neon --enable-rpi --enable-vout-drm --enable-v4l2-request --enable-libudev --enable-libdc1394 --enable-libdrm --enable-libiec61883 --enable-chromaprint --enable-frei0r --enable-libx264 --enable-shared --libdir=/usr/lib/arm-linux-gnueabihf --cpu=arm1176jzf-s --arch=arm
  libavutil      56. 22.100 / 56. 22.100
  libavcodec     58. 35.100 / 58. 35.100
  libavformat    58. 20.100 / 58. 20.100
  libavdevice    58.  5.100 / 58.  5.100
  libavfilter     7. 40.101 /  7. 40.101
  libavresample   4.  0.  0 /  4.  0.  0
  libswscale      5.  3.100 /  5.  3.100
  libswresample   3.  3.100 /  3.  3.100
  libpostproc    55.  3.100 / 55.  3.100
Input #0, h264, from '5sec.h264':
  Duration: N/A, bitrate: N/A
    Stream #0:0: Video: h264 (High), yuv420p(progressive), 640x480, 25 fps, 25 tbr, 1200k tbn, 50 tbc
Stream mapping:
  Stream #0:0 -> #0:0 (h264 (native) -> mjpeg (native))
Press [q] to stop, [?] for help
[swscaler @ 0x1a25390] deprecated pixel format used, make sure you did set range correctly
Output #0, image2, to 'frame.jpg':
  Metadata:
    encoder         : Lavf58.20.100
    Stream #0:0: Video: mjpeg, yuvj420p(pc), 640x480, q=2-31, 200 kb/s, 25 fps, 25 tbn, 25 tbc
    Metadata:
      encoder         : Lavc58.35.100 mjpeg
    Side data:
      cpb: bitrate max/min/avg: 0/0/200000 buffer size: 0 vbv_delay: -1
frame=    1 fps=0.4 q=6.8 Lsize=N/A time=00:00:01.04 bitrate=N/A speed=0.467x    
video:63kB audio:0kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead: unknown


  


  Note, achieved FPS is 0.4. When I increase the requested frame to be the 125th frame rather than the 25th, the FPS goes down to 0.1.

  


• LibVLC : Is a filter like v360:fisheye from ffmpeg available in LibVLC ?

  8 septembre 2021, par Craab

  I'm having problem playing fisheye(180) videos panoramically on the fly
using LibVlcSharp in a WinForms appliction.

  


  More specifically : what I want to do is filtering video with libvlcsharp in a WinForms application(c#) with the same effect like

  


  -vf v360=fisheye:e:ih_fov=180:iv_fov=180:pitch=90


  


  from ffmpeg.

  


  FYI : this will not make the video tagged with equirectanglar projection, the v360=fisheye filter dewarps the video and the video stays rectanglar.

  


  The input video is like this :

  


  

  


  And the the expected output should be like :

  


  

  


  This can be achieved by :

  


  ./ffmpeg -i input.mov  -vf v360=fisheye:e:ih_fov=180:iv_fov=180:pitch=90,crop=in_w:in_h/2:in_w:in_h/2 output.mov


  


  but I need to do this with LibVLC and on the fly.

  


  Can I achieve this with LibVLC's video filter ?

  


  Or I need to develop my own filter and make a wrapper for c# ?

  


• Trying to save frames as colored image using Ffmpeg in C++

  2 septembre 2021, par Tolga

  I am new to FFmpeg and I am trying to save the video frames as colored images. I have achieved saving them as grayscale using Netpbm, however, I need to save the frames as colored. I have tried implementing the code in this link.

  


  However, I get an error :

  


  'Exception thrown at 0x00E1FC4F (swscale-5.dll) in VideoDecoding2.exe:
 0xC0000005: Access violation writing location 0xCCCCCCCC.'


  


  Is there any way to improve this code or another way to save frames as colored ?

  


  Here is my code below.

  


  

  • src_pix_fmt is AV_PIX_FMT_YUV420p.

  


  • dst_pix_fmt is AV_PIX_FMT_RGB24.

  


  


  src_pix_fmt = avcc->pix_fmt;

src_width = avcc->width;
src_height = avcc->height;

dst_width = src_width;
dst_height = src_height;

numBytes = av_image_get_buffer_size(dst_pix_fmt, dst_width, dst_height, 0);

buffer = (uint8_t*)av_malloc(numBytes);

if ((ret = av_image_alloc(src_data, src_linesize, src_width, src_height, src_pix_fmt, 16)) < 0)
{
    printf("Couldn't allocate source image.\n");
    return 0;
}

av_image_fill_arrays(frameRGB->data, frameRGB->linesize, buffer, dst_pix_fmt, dst_width, dst_height, 0);

while (av_read_frame(avfc, packet) >= 0)
{
    ret = avcodec_send_packet(avcc, packet);
    if (ret < 0)
    {
        printf("Packets could not supplied to decoder.\n");
        return -1;
    }

    ret = avcodec_receive_frame(avcc, frame);
    printf("%d", ret);

    if (packet->stream_index == videoStream)
    {
        sws_ctx = sws_getContext(src_width, src_height, src_pix_fmt,
            dst_width, dst_height, dst_pix_fmt,
            SWS_BILINEAR, NULL, NULL, NULL);

        if (!sws_ctx)
        {
            printf("Cannot create scale context for conversion\n"
                "fmt:%s s:%dx%d --> fmt:%s s:%dx%d\n",
                av_get_pix_fmt_name(src_pix_fmt), src_width, src_height,
                av_get_pix_fmt_name(dst_pix_fmt), dst_width, dst_height);
            return 0;
        }

        sws_scale(sws_ctx, (const uint8_t* const*)frame->data, frame->linesize, 0, frame->height, dst_data, dst_linesize);

        FILE* f;
        char szFilename[32];
        int y;

        snprintf(szFilename, sizeof(szFilename), "frame%d.ppm", avcc->frame_number);
        fopen_s(&f, szFilename, "wb");
        
        if (f == NULL)
        {
            printf("Couldn't open file.\n");
            return 0;
        }
        
        fprintf(f, "P6\n%d %d\n255\n", dst_width, dst_height);

        for (y = 0; y < dst_height; y++)
            fwrite(frameRGB->data[0] + y * frameRGB->linesize[0], 1, dst_width * 3, f);
        
        fclose(f);
    }
}