Recherche avancée

Sur d’autres sites (3088)

• Use NVIDIA GPU to speed up drawtext with ffmpeg

  28 février 2020, par Garrett

  I use the following ffmpeg command add text to a video, which takes longer than I would like.

  ffmpeg -i input.mp4 -vf "[in] drawbox= x=iw*0.75:y=ih*0.88:w=iw*0.25:h=ih*0.12:color=black@0.7:t=fill, drawtext=fontsize=20:fontcolor=White:text='Test Text Row 1':x=(w*.75)+15:y=(h*.88)+15, drawtext=fontsize=20:fontcolor=White:text='Test Text Row 2':x=(w*.75)+15:y=((h*.88)+25)+15, drawtext=fontsize=20:fontcolor=White:text='Test Text Row 3':x=(w*.75)+15:y=((h*.88)+50)+15, drawtext=fontsize=20:fontcolor=White:text='Test Text Row 4':x=(w*.75)+15:y=((h*.88)+75)+15 [out]" -y out.mp4

  I believe I have successfully configured ffmpeg to use the GPU on my laptop by following NVIDIA’s guide, which can be found https://developer.nvidia.com/ffmpeg

  However I am having trouble integrating the commands to utilize the GPU with my command above. I tested the following command from NVIDIA’s site to test configuration and it worked without error.

  ffmpeg -vsync 0 -c:v h264_cuvid -i  -f rawvideo 

  I have not found any examples yet of the integration of drawtext and GPU acceleration so any guidance appreciated.

• couldn't render an audio after extracting the audio from the video

  3 juin 2022, par Rama Ranneh

  hi … I am trying to use FFmpegFrameRecorder to extract an audio from a video using this code

  


          FFmpegFrameGrabber frameGrabber = new FFmpegFrameGrabber(mp4Path);
        frameGrabber.start();
        FFmpegFrameRecorder recorder = new FFmpegFrameRecorder(new 
        File("C:/Users/ASUS/Desktop/audio.aac"),audioChannels);
        recorder.setAudioMetadata(frameGrabber.getAudioMetadata());
        recorder.setFrameRate(frameGrabber.getFrameRate());
        recorder.setTimestamp(frameGrabber.getTimestamp());
        recorder.start(frameGrabber.getFormatContext());

        Frame f=null;
        //get audio sample and record it
        while ((f = frameGrabber.grabSamples()) != null) {
            recorder.record(f);
        }
        frameGrabber.stop();
        recorder.stop();
        recorder.close();
        LOGGER.info("C:/Users/ASUS/Desktop/audio.aac");


  


  but when I run it it gives me in terminal :

  


  Input #0, mov,mp4,m4a,3gp,3g2,mj2, from 'C:/Users/ASUS/Desktop/Lou.mp4':
Metadata:
major_brand     : isom
minor_version   : 512
compatible_brands: isomiso2avc1mp41
encoder         : Lavf58.76.100
Duration: 00:00:39.10, start: 0.000000, bitrate: 116 kb/s
Stream #0:0[0x1](und): Video: h264 (Main) (avc1 / 0x31637661), yuv420p(tv, bt709, 
progressive), 256x144 [SAR 1:1 DAR 16:9], 63 kb/s, 23.98 fps, 23.98 tbr, 90k tbn (default)
Metadata:
  handler_name    : VideoHandler
  vendor_id       : [0][0][0][0]
  Stream #0:1[0x2](und): Audio: aac (HE-AAC) (mp4a / 0x6134706D), 44100 Hz, stereo, fltp, 46 kb/s (default)
Metadata:
  handler_name    : SoundHandler
  vendor_id       : [0][0][0][0]
 [aac @ 000000001afd7080] noise_facs_q 31 is invalid
 [aac @ 000000001afd7080] noise_facs_q 31 is invalid
 [aac @ 000000001afd7080] noise_facs_q 31 is invalid
 [aac @ 000000001afd7080] noise_facs_q 31 is invalid
 [aac @ 000000001afd7080] noise_facs_q 31 is invalid
 [aac @ 000000001afd7080] noise_facs_q 31 is invalid
 [aac @ 000000001afd7080] noise_facs_q 31 is invalid
 [h264 @ 000000001ad8bc40] mmco: unref short failure
 [h264 @ 000000001ad8bc40] mmco: unref short failure
 [h264 @ 000000001ad8bc40] mmco: unref short failure
 [h264 @ 000000001ad8bc40] mmco: unref short failure
 [h264 @ 000000001ad8bc40] mmco: unref short failure
 and so on ...
  Output #0, adts, to 'C:\Users\ASUS\Desktop\aud.aac':
 Metadata:
 encoder         : Lavf59.16.100
  Stream #0:0: Audio: aac (HE-AAC) (mp4a / 0x6134706D), 44100 Hz, stereo, fltp, 46 kb/s


  


  and when I try to run the audio it is not running
what's the wrong thing I did ?

  


• How to mirror swscale PIX_FMT_YUYV422

  31 janvier 2012, par superg

  I'm trying to mirror libswscale PIX_FMT_YUYV422-type image horizontally. Using simple loop for each line with 16-bits per pixel results in having colors wrong, for example blue objects are orange. Here is my code :

     typedef unsigned short YUVPixel; // 16 bits per pixel
  
      for (int y = 0; y < outHeight; y++)
  
      {
  
          YUVPixel *p1 = (YUVPixel*)pBits + y * outWidth;
  
          YUVPixel *p2 = p1 + outWidth - 1;
  
          for (int x = 0; x < outWidth/2; x++) // outWidth is image width in pixels
  
          {
  
              // packed YUV 4:2:2, 16bpp, Y0 Cb Y1 Cr
  
              unsigned short tmp;
  
              tmp = *p1;
  
              *p1 = *p2;
  
              *p2 = tmp;
  
          }
  
      }

  Then I tried redefining YUVPixel as 32-bit type and modifying my loop accordingly. This results in correct colors, but looks like neighboring pixels are swapped. Any ideas, I'm totally lost with this ?