Recherche avancée

Sur d’autres sites (8563)

• With ffmpeg's image to movie feature, is it possible to pass in frames over a period of time versus all at once ?

  2 novembre 2013, par Zack Yoshyaro

  For the purpose of making a time lapse recording of desktop activity, it is possible so "stream" the frame list to ffmpeg over time, rather than all at once in the beginning.

  Currently, it is a two step process.

  1. save individual snapshots to disc

     im = ImageGrab.grab()
     im.save("frame_%s.jpg" % count, 'jpg')

  2. compile those snapshots with ffmpeg via

     ffmpeg -r 1 -pattern_type glob -i '*.jpg' -c:v libx264 out.mp4

  It would be nice if there were a way to merge the two steps so that I'm not flooding my hard drive with thousands of individual snapshots. Is it possible to do this ?

• libswresample : swr_convert() returns empty buffer

  11 septembre 2019, par Герман Лиманський

  I try to convert audio in format AV_SAMPLE_FMT_S32. I use swr_convert(), but out buffer still empty.

  // frame is decoded frame, rframe - is empty frame(out buffer)

      if (!main_context->resampler) {
  
                      main_context->resampler = 
  
                      swr_alloc_set_opts(main_context->resampler,
  
                      AV_CH_LAYOUT_STEREO,                    // output
  
                      AV_SAMPLE_FMT_S32,                      // output
  
                      44100,                                  // output
  
                      audio_codec_context->channel_layout,    // input
  
                      audio_codec_context->sample_fmt,        // input
  
                      audio_codec_context->sample_rate,       // input
  
                      0,
  
                      nullptr);
  
                  swr_init(main_context->resampler);
  
  
  
              }
  
  
  
              //int in_samples = frame->nb_samples;
  
              int out_samples = av_rescale_rnd(swr_get_delay(
  
                  main_context->resampler, 44100) + 44100,
  
                  44100,
  
                  44100,
  
                  AV_ROUND_UP);
  
  
  
              size_t buffSize = av_samples_alloc(rframe->data, NULL,audio_codec_context->channels, out_samples, AV_SAMPLE_FMT_S32, 0);
  
              int len = swr_convert(main_context->resampler, rframe->data, frame->nb_samples, (const uint8_t * *)frame->data, frame->nb_samples);
  
  //here.. rframe->data should have some data, but its empty
  
              while (len > 0)
  
              {
  
                  size_t size_ = rframe->nb_samples * av_get_bytes_per_sample(AV_SAMPLE_FMT_S32);
  
                  main_context->audio_buf.write(rframe->data[0], size_, 1);
  
  
  
                  len = swr_convert(main_context->resampler, rframe->data, frame->nb_samples, NULL, NULL);
  
              }

• Using ffmpeg to record screen returns corrupted file

  8 août 2020, par odddollar

  I'm using ffmpeg and python to record my desktop screen. I've got it working so that when I input a shortcut, it starts recording. Then I use .terminate() on the subprocess to stop recording. When outputting to an mp4, this corrupts the file and makes it unreadable. I can output the file as an flv or avi and it doesn't get corrupted, but then the video doesn't contain time/duration data, something I need.

  


  Is there a way I can gracefully stop the recording when outputting an mp4 ?
Or is there a way I can include the time/duration data in the flv/avi ?

  


  import keyboard
import os
from subprocess import Popen

class Main:
    def __init__(self):
        self.on = False

    def main(self):
        if not self.on:
            if os.path.isfile("output.mp4"):
                os.remove("output.mp4")

            self.process = Popen('ffmpeg -f gdigrab -framerate 30 -video_size 1920x1080 -i desktop -b:v 5M output.mp4')
            self.on = True
        else:
            self.process.terminate()

            self.on = False

run = Main()
keyboard.add_hotkey("ctrl+shift+g", lambda:run.main())

keyboard.wait()