Recherche avancée

Sur d’autres sites (10763)

• avcodec/vvc : simplify priority logical to improve performance for 4K/8K

  24 septembre 2024, par Nuo Mi

  avcodec/vvc : simplify priority logical to improve performance for 4K/8K
  For 4K/8K video processing, it's possible to have over 1,000 tasks pending on the executor.
  
  In such cases, O(n) and O(log(n)) insertion times are too costly.
  
  Reducing this to O(1) will significantly decrease the time spent in critical sections
  clip                                                        | before | after  | delta
  
  |--------|--------|-------
  
  VVC_HDR_UHDTV2_OpenGOP_7680x4320_50fps_HLG10.bit            |    24  |   27   |  12.5%
  
  VVC_HDR_UHDTV2_OpenGOP_7680x4320_50fps_HLG10_HighBitrate.bit|    12  |   17   |  41.7%
  
  tears_of_steel_4k_8M_8bit_2000.vvc                          |    34  |  102   | 200.0%
  
  VVC_UHDTV1_OpenGOP_3840x2160_60fps_HLG10.bit                |   126  |  128   |   1.6%
  
  RitualDance_1920x1080_60_10_420_37_RA.266                   |   350  |  378   |   8.0%
  
  NovosobornayaSquare_1920x1080.bin                           |   341  |  369   |   8.2%
  
  Tango2_3840x2160_60_10_420_27_LD.266                        |    69  |   70   |   1.4%
  
  RitualDance_1920x1080_60_10_420_32_LD.266                   |   243  |  259   |   6.6%
  
  Chimera_8bit_1080P_1000_frames.vvc                          |   420  |  392   |  -6.7%
  
  BQTerrace_1920x1080_60_10_420_22_RA.vvc                     |   148  |  144   |  -2.7%
  

  • [DH] libavcodec/executor.c
  • [DH] libavcodec/executor.h
  • [DH] libavcodec/vvc/thread.c

• FFmpeg encoding live audio to aac issue

  12 juillet 2015, par Ruurd Adema

  I’m trying to encode live raw audio coming from a Blackmagic Decklink input card to a mov file with AAC encoding.

  The issue is that the audio sounds distorted and plays to fast.

  I created the software based on a couple of examples/tutorials including the Dranger tutorial and examples on Github (and of course the examples in the FFmpeg codebase).

  Honestly, at this moment I don’t exactly know what the cause of the problem is. I’m thinking about PTS/DTS values or a timebase mismatch (because of the too fast playout), I tried a lot of things, including working with an av_audio_fifo.

  • When outputting to the mov file with the AV_CODEC_ID_PCM_S16LE codec, everything works well
  • When outputting to the mov file with the AV_CODEC_ID_AAC codec, the problems occur
  • When writing RAW audio VLC media info shows :
    Type : Audio, Codec : PCM S16 LE (sowt), Language : English, Channels : Stereo, Sample rate : 48000 Hz, Bits per sample.
  • When writing with AAC codec VLC media info shows :
    Type : Audio, Codec : MPEG AAC Audio (mp4a), Language : English, Channels : Stereo, Sample rate : 48000 Hz.

  Any idea(s) of what’s causing the problems ?

  Code

  // Create output context
  
  
  
  output_filename = "/root/movies/encoder_debug.mov";
  
  output_format_name = "mov";
  
  if (avformat_alloc_output_context2(&output_fmt_ctx, NULL, output_format_name, output_filename) < 0)
  
  {
  
      printf("[ERROR] Unable to allocate output format context for output: %s\n", output_filename);
  
  } 
  
  
  
  // Create audio output stream
  
  
  
  static AVStream *encoder_add_audio_stream(AVFormatContext *oc, enum AVCodecID codec_id)
  
  {
  
      AVCodecContext  *c;
  
      AVCodec         *codec;
  
      AVStream        *st;
  
  
  
      st = avformat_new_stream(oc, NULL);
  
      if (!st) 
  
      {
  
          printf("[ERROR] Could not allocate new audio stream!\n");
  
          exit(-1);
  
      }
  
  
  
      c                 = st->codec;
  
      c->codec_id       = codec_id;
  
      c->codec_type     = AVMEDIA_TYPE_AUDIO;
  
      c->sample_fmt     = AV_SAMPLE_FMT_S16;
  
      c->sample_rate    = decklink_config()->audio_samplerate;
  
      c->channels       = decklink_config()->audio_channel_count;
  
      c->channel_layout = av_get_default_channel_layout(decklink_config()->audio_channel_count);
  
      c->time_base.den  = decklink_config()->audio_samplerate;
  
      c->time_base.num  = 1;
  
  
  
      if (codec_id == AV_CODEC_ID_AAC)
  
      {
  
          c->bit_rate       = 96000;  
  
          //c->profile = FF_PROFILE_AAC_MAIN; //FIXME Generates error: "Unable to set the AOT 1: Invalid config"
  
          // Allow the use of the experimental AAC encoder
  
          c->strict_std_compliance = FF_COMPLIANCE_EXPERIMENTAL; 
  
      }
  
  
  
      // Some formats want stream headers to be seperate (global?)
  
      if (oc->oformat->flags & AVFMT_GLOBALHEADER)
  
      {
  
          c->flags |= CODEC_FLAG_GLOBAL_HEADER;
  
      }
  
  
  
      codec = avcodec_find_encoder(c->codec_id);
  
      if (!codec)
  
      {
  
          printf("[ERROR] Audio codec not found\n");
  
          exit(-1);
  
      }
  
  
  
      if (avcodec_open2(c, codec, NULL) < 0)
  
      {
  
          printf("[ERROR] Could not open audio codec\n");
  
          exit(-1);
  
      }
  
  
  
      return st;
  
  }
  
  
  
  // En then, at every incoming frame this function gets called:
  
  
  
  void encoder_handle_incoming_frame(IDeckLinkVideoInputFrame *videoframe, IDeckLinkAudioInputPacket *audiopacket)
  
  {
  
      void *pixels = NULL;
  
      int   pitch = 0;
  
      int got_packet = 0;
  
  
  
      void *audiopacket_data          = NULL;
  
      long  audiopacket_sample_count  = 0;
  
      long  audiopacket_size          = 0;
  
      long  audiopacket_channel_count = 2;
  
  
  
      if (audiopacket)
  
      {  
  
          AVPacket pkt = {0,0,0,0,0,0,0,0,0,0,0,0,0,0};
  
          AVFrame *frame;
  
          BMDTimeValue audio_pts;
  
          int requested_size;
  
          static int last_pts1, last_pts2 = 0;
  
  
  
          audiopacket_sample_count  = audiopacket->GetSampleFrameCount();
  
          audiopacket_channel_count = decklink_config()->audio_channel_count;
  
          audiopacket_size          = audiopacket_sample_count * (decklink_config()->audio_sampletype/8) * audiopacket_channel_count;
  
  
  
          audiopacket->GetBytes(&audiopacket_data);
  
  
  
          av_init_packet(&pkt);    
  
  
  
          printf("\n=== Audiopacket: %d ===\n", audio_stream->codec->frame_number);
  
  
  
          if (AUDIO_TYPE == AV_CODEC_ID_PCM_S16LE)
  
          {
  
              audiopacket->GetPacketTime(&audio_pts, audio_stream->time_base.den);
  
  
  
              pkt.pts          = audio_pts;
  
              pkt.dts          = pkt.pts; 
  
              pkt.flags       |= AV_PKT_FLAG_KEY;                 // TODO: Make sure if this still applies
  
              pkt.stream_index = audio_stream->index;
  
              pkt.data         = (uint8_t *)audiopacket_data;
  
              pkt.size         = audiopacket_size;
  
  
  
              printf("[PACKET] size:              %d\n", pkt.size);
  
              printf("[PACKET] pts:               %li\n", pkt.pts);
  
              printf("[PACKET] pts delta:         %li\n", pkt.pts - last_pts2);
  
              printf("[PACKET] duration:          %d\n", pkt.duration);
  
              last_pts2 = pkt.pts;
  
  
  
              av_interleaved_write_frame(output_fmt_ctx, &pkt);
  
          } 
  
          else if (AUDIO_TYPE == AV_CODEC_ID_AAC)
  
          {
  
              frame = av_frame_alloc();
  
              frame->format = audio_stream->codec->sample_fmt;
  
              frame->channel_layout = audio_stream->codec->channel_layout;
  
              frame->sample_rate = audio_stream->codec->sample_rate;
  
              frame->nb_samples = audiopacket_sample_count;
  
  
  
              requested_size = av_samples_get_buffer_size(NULL, audio_stream->codec->channels, audio_stream->codec->frame_size, audio_stream->codec->sample_fmt, 1);
  
  
  
              audiopacket->GetPacketTime(&audio_pts, audio_stream->time_base.den);
  
  
  
              printf("[DEBUG] Sample format:      %d\n", frame->format);
  
              printf("[DEBUG] Channel layout:     %li\n", frame->channel_layout);
  
              printf("[DEBUG] Sample rate:        %d\n", frame->sample_rate);
  
              printf("[DEBUG] NB Samples:         %d\n", frame->nb_samples);
  
              printf("[DEBUG] Datasize:           %li\n", audiopacket_size);
  
              printf("[DEBUG] Requested datasize: %d\n", requested_size);
  
              printf("[DEBUG] Too less/much:      %li\n", audiopacket_size - requested_size);
  
              printf("[DEBUG] Framesize:          %d\n", audio_stream->codec->frame_size);
  
              printf("[DEBUG] Audio pts:          %li\n", audio_pts);
  
              printf("[DEBUG] Audio pts delta:    %li\n", audio_pts - last_pts1);
  
              last_pts1 = audio_pts;
  
  
  
              frame->pts = audio_pts;
  
  
  
              if (avcodec_fill_audio_frame(frame, audiopacket_channel_count, audio_stream->codec->sample_fmt, (const uint8_t *)audiopacket_data, audiopacket_size, 0) < 0)
  
              {
  
                  printf("[ERROR] Filling audioframe failed!\n");
  
                  exit(-1);
  
              }
  
  
  
              got_packet = 0;
  
              if (avcodec_encode_audio2(audio_stream->codec, &pkt, frame, &got_packet) != 0)
  
              {
  
                  printf("[ERROR] Encoding audio failed\n");
  
              }
  
  
  
              if (got_packet) 
  
              {
  
                  pkt.stream_index = audio_stream->index;
  
                  pkt.flags       |= AV_PKT_FLAG_KEY; 
  
  
  
                  //printf("[PACKET] size:              %d\n", pkt.size);
  
                  //printf("[PACKET] pts:               %li\n", pkt.pts);
  
                  //printf("[PACKET] pts delta:         %li\n", pkt.pts - last_pts2);
  
                  //printf("[PACKET] duration:          %d\n", pkt.duration);
  
                  //printf("[PACKET] timebase codec:    %d/%d\n", audio_stream->codec->time_base.num, audio_stream->codec->time_base.den);
  
                  //printf("[PACKET] timebase stream:   %d/%d\n", audio_stream->time_base.num, audio_stream->time_base.den);
  
                  last_pts2 = pkt.pts;
  
  
  
                  av_interleaved_write_frame(output_fmt_ctx, &pkt);
  
              }
  
              av_frame_free(&frame); 
  
          }
  
  
  
          av_free_packet(&pkt);
  
      }
  
      else
  
      {
  
          printf("[WARNING] No audiopacket received!\n");
  
      }
  
  
  
      static int count = 0;
  
      count++;
  
  }

• libavformat : Improve ff_configure_buffers_for_index for excessive deltas

  21 mars 2023, par Martin Storsjö

  libavformat : Improve ff_configure_buffers_for_index for excessive deltas
  Previously, the ff_configure_buffers_for_index function had
  
  upper sanity limits of 16 MB (1<<24) for buffer_size and
  
  8 MB (1<<23) for short_seek_threshold.
  However, if the index contained entries with a much larger
  
  delta, setting pos_delta to a value larger than the sanity
  
  limit, we would end up not increasing the buffer size at all.
  Instead, ignore the individual deltas that are excessive, but
  
  increase the buffer size based on the deltas that are below the
  
  sanity limit.
  Only count deltas that are below 1<<23, 8 MB ; pos_delta gets doubled
  
  before setting the buffer size - this matches the previous maximum
  
  buffer size of 1<<24, 16 MB.
  This can happen e.g. with a mov file with some tracks containing
  
  some samples that belong in the start of the file, at the end of
  
  the mdat, while the rest of the file is mostly reasonably interleaved ;
  
  previously those samples caused the maximum pos_delta to skyrocket,
  
  skipping any buffer size enlargement.
  Signed-off-by : Martin Storsjö <martin@martin.st>
  

  • [DH] libavformat/seek.c