Recherche avancée

Sur d’autres sites (7008)

• How to convert from GL_RGB to AVFrame

  11 avril 2022, par benny b

  For my project I need to convert a RGB (GL_RGB) Image generated by glReadPixels into a AVFrame. I've googled it and found just examples of the other way around. But in this case I need from GL_RGB to AVFrame.

  


  Here's my code :

  


  Here's how I set my codec :

  


  /* Allocate resources and write header data to the output file. */
void ffmpeg_encoder_start(AVCodecID codec_id, int fps, int width, int height) {
    const AVCodec *codec;
    int ret;
    codec = avcodec_find_encoder(codec_id);
    if (!codec ) {
        std::cerr << "Codec not found" << std::endl;
        exit(1);
    }
    c = avcodec_alloc_context3(codec);
    if (!c) {
        std::cerr << "Could not allocate video codec context" << std::endl;
        exit(1);
    }
    c->bit_rate = 400000;
    c->width = width;
    c->height = height;
    c->time_base.num = 1;
    c->time_base.den = fps;
    c->keyint_min = 600;
    c->pix_fmt = AV_PIX_FMT_YUV420P;
    if (avcodec_open2(c, codec, NULL) < 0) {
        std::cerr << "Could not open codec" << std::endl;
        exit(1);
    }
    frame = av_frame_alloc();
    if (!frame) {
        std::cerr << "Could not allocate video frame" << std::endl;
        exit(1);
    }
    frame->format = c->pix_fmt;
    frame->width  = c->width;
    frame->height = c->height;
    ret = av_image_alloc(frame->data, frame->linesize, c->width, c->height, c->pix_fmt, 32);
    if (ret < 0) {
        std::cerr << "Could not allocate raw picture buffer" << std::endl;
        exit(1);
    }
}


  


  Fetching the pixels and setting the new frame :

  


  BYTE* pixels = new BYTE[3 * DEFAULT_MONITOR.maxResolution.width * DEFAULT_MONITOR.maxResolution.height];

glReadPixels(0, 0, DEFAULT_MONITOR.maxResolution.width, DEFAULT_MONITOR.maxResolution.height, GL_RGB, GL_UNSIGNED_BYTE, pixels);
screenSrc->setNextFrame(pixels, DEFAULT_MONITOR.maxResolution.width, DEFAULT_MONITOR.maxResolution.height);


  


  And the function that I have for the conversion :

  


  static void ffmpeg_encoder_set_frame_yuv_from_rgb(uint8_t *rgb) {
    const int in_linesize[1] = { 3 * c->width };
    sws_context = sws_getCachedContext(sws_context,
            c->width, c->height, AV_PIX_FMT_RGB24,
            c->width, c->height, AV_PIX_FMT_YUV420P,
            0, 0, 0, 0);
    sws_scale(sws_context, (const uint8_t * const *)&rgb, in_linesize, 0,
            c->height, frame->data, frame->linesize);
}


  


  All the code can be found here
Here is the line that results in segmentation fault.

  


  Unfortunately, the function gives me a segmentation fault. Do you have an idea how to solve this problem ?

  


• How to Transcode ALL Audio streams from input to output using ffmpeg ?

  24 novembre 2022, par user1940163

  I have an input MPEG TS file 'unit_test.ts'. This file has following content (shown by ffprobe) :

  


  Input #0, mpegts, from 'unit_test.ts':
  Duration: 00:00:57.23, start: 73674.049844, bitrate: 2401 kb/s
  Program 1
    Metadata:
      service_name    : Service01
      service_provider: FFmpeg
    Stream #0:0[0x31]: Video: h264 (Main) ([27][0][0][0] / 0x001B), yuv420p(progressive), 852x480 [SAR 640:639 DAR 16:9], Closed Captions, 59.94 fps, 59.94 tbr, 90k tbn, 119.88 tbc
    Stream #0:1[0x34](eng): Audio: ac3 ([129][0][0][0] / 0x0081), 48000 Hz, 5.1(side), fltp, 448 kb/s
    Stream #0:2[0x35](spa): Audio: ac3 ([129][0][0][0] / 0x0081), 48000 Hz, stereo, fltp, 192 kb/s


  


  I want to convert it into another MPEG TS file. Requirement is that the Video stream of the input should be directly copied to the output whereas ALL the audio streams should be transcoded "aac" format.

  


  I tried this command :

  


  ffmpeg -i unit_test.ts  -map 0 -c copy -c:a aac maud_test.ts

  


  It converted it into 'maud_test.ts' with following contents (shown by ffprobe)

  


  Input #0, mpegts, from 'maud_test.ts':
  Duration: 00:00:57.25, start: 1.400000, bitrate: 2211 kb/s
  Program 1
    Metadata:
      service_name    : Service01
      service_provider: FFmpeg
    Stream #0:0[0x100]: Video: h264 (Main) ([27][0][0][0] / 0x001B), yuv420p(progressive), 852x480 [SAR 640:639 DAR 16:9], Closed Captions, 59.94 fps, 59.94 tbr, 90k tbn, 119.88 tbc
    Stream #0:1[0x101](eng): Audio: aac (LC) ([15][0][0][0] / 0x000F), 48000 Hz, 6 channels, fltp, 391 kb/s
    Stream #0:2[0x102](spa): Audio: aac (LC) ([15][0][0][0] / 0x000F), 48000 Hz, stereo, fltp, 133 kb/s


  


  So it appeared as if the command worked....However when I play the maud_test.ts file in vlc player I can see both audio streams listed in the menu ; but Stream 1 (eng) remains silent............whereas Stream 2 (spa) has proper audio. (Original TS file has both audio streams properly audible)

  


  I have tried this with different input files and have seen that same problem occurs in each case.

  


  What that I am doing is not right ?

  


  How should I get this done ? (I can write explicit stream by stream map and channel arguments to get that done ; however I want the command line to be generic, in that the input file could be having any configuration with one Video and several Audios with different formats. The configuration will not be known beforehand.)

  


• Error submitting the frame for encoding when submitting NV12 texture

  6 mai 2022, par Caio Augusto

  I'm trying to encode D3D11 NV12 Texture on QSV encoder but getting [h264_qsv @ 00000244ce6f50c0] Error submitting the frame for encoding.

  


  Main :

  


  int width = 1920;
int height = 1080;
FILE* outfile;

fopen_s(&outfile, "D:\\Sources\\D3D11QSV\\x64\\Debug\\outfile.264", "wb");

const AVCodec* codec = avcodec_find_encoder_by_name("h264_qsv");
AVCodecContext* ctx = avcodec_alloc_context3(codec);

ctx->width = width;
ctx->height = height;
ctx->time_base = AVRational{ 1, 60 };
ctx->framerate = AVRational{ 60, 1 };
ctx->slices = 1;

ctx->sw_pix_fmt = AV_PIX_FMT_NV12;
ctx->pix_fmt = AV_PIX_FMT_NV12;
ctx->bit_rate = 400000;
ctx->gop_size = 10;
ctx->max_b_frames = 1;

auto status = avcodec_open2(ctx, codec, NULL);
if (status < 0) {
    std::cout << "Open codec error!\n";
}

AVFrame* sw_frame = av_frame_alloc();
sw_frame->format = ctx->sw_pix_fmt;
sw_frame->width = ctx->width;
sw_frame->height = ctx->height;
status = av_frame_get_buffer(sw_frame, 0);

fill_frame(sw_frame, ctx);


  


  Filling the frame :

  


  auto ret = 0;

if (ret < 0) {
    fprintf(stderr, "Could not allocate the video frame data\n");
    exit(1);
}

int i, y, x, c = 0;
for (i = 0; i < 60; i++) {
    fflush(stdout);

    ret = av_frame_make_writable(frame);
    
    auto texture = create_texture();
    auto desc = (AVD3D11FrameDescriptor*)frame->buf[0]->data;
    desc->texture = (ID3D11Texture2D*)texture;
    desc->index = 0;

    frame->data[0] = (std::uint8_t*)texture;
    frame->data[1] = 0;
    frame->linesize[0] = width * 4;

    frame->pts = i;

    encode(frame, ctx);
}


  


  Creating Texture :

  


  D3D11_TEXTURE2D_DESC const desc = CD3D11_TEXTURE2D_DESC(
    DXGI_FORMAT_NV12,           // HoloLens PV camera format, common for video sources
    width,                    // Width of the video frames
    height,                   // Height of the video frames
    1,                          // Number of textures in the array
    1,                          // Number of miplevels in each texture
    D3D11_BIND_SHADER_RESOURCE, // We read from this texture in the shader
    D3D11_USAGE_DYNAMIC,        // Because we'll be copying from CPU memory
    D3D11_CPU_ACCESS_WRITE      // We only need to write into the texture
);

ID3D11Device* pd3dDevice = create_d3d11_device();

ID3D11Texture2D* pTexture = NULL;
HRESULT err = pd3dDevice->CreateTexture2D(&desc, nullptr, &pTexture);


if (SUCCEEDED(err)) {
    D3D11_SHADER_RESOURCE_VIEW_DESC SRVDesc = CD3D11_SHADER_RESOURCE_VIEW_DESC(
        pTexture,
        D3D11_SRV_DIMENSION_TEXTURE2D,
        DXGI_FORMAT_R8_UNORM
    );

    ID3D11ShaderResourceView* texSRV = NULL;

    err = pd3dDevice->CreateShaderResourceView(pTexture,
        &SRVDesc, &texSRV);

    D3D11_SHADER_RESOURCE_VIEW_DESC const chrominancePlaneDesc = CD3D11_SHADER_RESOURCE_VIEW_DESC(
        pTexture,
        D3D11_SRV_DIMENSION_TEXTURE2D,
        DXGI_FORMAT_R8G8_UNORM
    );

    ID3D11ShaderResourceView* m_chrominanceView = NULL;

    err = pd3dDevice->CreateShaderResourceView(pTexture,
        &chrominancePlaneDesc, &m_chrominanceView);

}

if (FAILED(err))
{
    fprintf(stderr, "Error creating texture\n");
    exit(1);
}

return pTexture;


  


  Creating D3D11 device :

  


      ID3D11Device* dev11 = NULL;
    ID3D11DeviceContext* devcon11 = NULL;

    D3D_FEATURE_LEVEL featureLevels[]{
   D3D_FEATURE_LEVEL_11_1,
   D3D_FEATURE_LEVEL_11_0,
   D3D_FEATURE_LEVEL_10_1,
   D3D_FEATURE_LEVEL_10_0,
   D3D_FEATURE_LEVEL_9_3,
   D3D_FEATURE_LEVEL_9_2,
   D3D_FEATURE_LEVEL_9_1
    };


    int err = D3D11CreateDevice(
        nullptr,
        D3D_DRIVER_TYPE_HARDWARE,
        nullptr,
        D3D11_CREATE_DEVICE_VIDEO_SUPPORT,
        featureLevels, sizeof(featureLevels) / sizeof(D3D_FEATURE_LEVEL),
        D3D11_SDK_VERSION,
        &dev11,
        nullptr,
        &devcon11);

    return dev11;


  


  Encoding :

  


  auto status = 0;

status = avcodec_send_frame(ctx, frame); //error happening here

AVPacket* pkt;

pkt = av_packet_alloc();

if (status < 0) {
    fprintf(stderr, "Error sending a frame for encoding\n");
    exit(1);
}

while (status >= 0) {
    status = avcodec_receive_packet(ctx, pkt);
    if (status == AVERROR(EAGAIN) || status == AVERROR_EOF)
        return;
    else if (status < 0) {
        fprintf(stderr, "Error during encoding\n");
        exit(1);
    }

    printf("Write packet \n", pkt->pts, pkt->size);
    fwrite(pkt->data, 1, pkt->size, outfile);
    av_packet_unref(pkt);
}


  


  Everything runs well until encoding the frame. I have tried sending a dummy nv12 data (not a d3d11 texture) and it works well.