Recherche avancée

Sur d’autres sites (11427)

• x264 : Map color parameters

  13 avril 2015, par Luca Barbato

  x264 : Map color parameters
  

  • [DBH] libavcodec/libx264.c

• Encoding YUV420 frames with x264 in C++

  2 avril 2020, par Eoin McBennett

  Im trying to use libx264 to convert YUV 420 frames to H264, as per this question : (How does one encode a series of images into H264 using the x264 C API ?) I have got my encoder parameters set up and my encoder initialised, the part that I am confused about is how do I give the encoder the YUV420 data I have ? The question above shows how the data would be passed in and converted from RGB to YUV420.

  



  I am also confused about what the stride, planes and nalls are within the x264 library any explanation of these would be greatly appreciated. I am very new to C++ and video processing in general.

  



  The code I have currently is posted below.
The YUV420 data I have is an unsigned char* called data that is declared outside of the function.
Also the current resolution of the frames being captured is 640x480

  



  void encode(){
    x264_param_t param;
    x264_param_default_preset(&param,"veryfast","zerolatency");
    param.i_threads = 1;
    param.i_width = camera.getWidth();
    param.i_height = camera.getHeight();
    param.i_fps_num = 30;
    param.i_fps_den = 1;
// Intra refres:
    param.i_keyint_max = 30;
    param.b_intra_refresh = 1;
//Rate control:
    param.rc.i_rc_method = X264_RC_CRF;
    param.rc.f_rf_constant = 25;
    param.rc.f_rf_constant_max = 35;
//For streaming:
    param.b_repeat_headers = 1;
    param.b_annexb = 1;
    x264_param_apply_profile(&param, "baseline");

    x264_t *encoder = x264_encoder_open(&param); //H.264 encoder object
    x264_picture_t pic_in, pic_out;
    x264_picture_alloc(&pic_in, X264_CSP_I420,camera.getWidth(), camera.getHeight());

    x264_nal_t *nals; //What is the purpose of nals?
    int i_nals; 

    //Im constantly getting frames captured from the camera that I need to encod
    while(true){

        //Im assuming the data would have to be placed within the pic_in pointer some how?

        int frame_size = x264_encoder_encode(encoder, &nals, &i_nals, &pic_in, &pic_out);
        if (frame_size >= 0) {
            std::cout<<"Encoded frame to h264!" << std::endl;
        }
    }
}


  



  Also if anyone could comment on the viability of this set up for video streaming, the end result of this project is to stream the captured and converted h264 frames over UDP for anything on the network to read and display.

  



  Any help would be greatly appreciated !

  


• speeding up x264 encoding (C++ code with libavcodec)

  20 décembre 2012, par Hrishikesh_Pardeshi

  I am trying to capture windows screen (continuous screen shots) and encode them into x264. For that I am using avcodec_encode_video2 function available with libavcodec. However, it takes a huge amount of time. The time fluctuates between 25 – 1800 milliseconds for encoding individual frames.

  I tried tried both 1080p and 720p with video recording on screen.

  These are the settings I am using. This was tested on Windows 7, win32 release build with 4 GB of RAM.

  bit_rate = 2000, width = 1920, height = 1080
  qmin = 0, qmax = 0, max_b_frames = 0, frame_rate = 25, pixel_format = YUV 4:4:4.
  The remaining settings are default which are fetched using avcodec_get_context_defaults3().

  Sample data(in milliseconds) for 20 frames (consecutive and chosen randomly) in a set of 250 frames.
  121, 106, 289, 126, 211, 30, 181, 58, 213, 34, 245, 50, 56, 364, 247, 171, 254, 83, 82, 229

  For the application it is a must that it captures at least at 15 fps. Can someone help out to tell whether any options can be used to improve the frame rate. I need to encode lossless, but I am open to some file size increase.

  Thanks in advance.