Sur d’autres sites (8539)

• FFMPEG : Add a fixed size image on a video, regardless of the video width & height (resolution)

  11 mars 2017, par Drupalist

  This is my code that adds an image to videos, running via PHP :

  exec('ffmpeg -i input.mp4 -i logo.png -filter_complex 
  
  "[0:v][1:v] overlay=10:10"  -pix_fmt yuv420p -c:a copy output.mp4');

  It works well but the problem is, the image is scaled down or up, up on the video resolution. For example in the following images the logo width is 50px but videos resolution are different :

  

  and this one

  

  How can I prevent the image from scaling down/up ?

  ---

  Update

  Thanks to Mulvya, he proposed these codes

  ffmpeg -i input.mp4 -i logo.png -filter_complex 
  
  "[1:v][0:v]scale2ref=(W/H)*ih/8:ih/8[wm][base];[base][wm]overlay=10:10"
  
  -pix_fmt yuv420p -c:a copy output.mp4

  and

  ffmpeg -i input.mp4 -i logo.png -filter_complex 
  
  "[1:v][0:v]scale2ref=(W/H)*ih/8:ih/8[wm][base];[wm]setsar=1[wmsar];
  
  [base][wmsar]overlay=10:10"
  
  -pix_fmt yuv420p -c:a copy output.mp4

  that works very well, but it doesn’t keep the logo aspect ratio.
  I tried this code on two videos with different resolution and this is the result

  

  and this one

  

  Is it possible to improve this solution ?

• FFMPEG merge two .mp4 videos - resolution distorted

  9 juin 2016, par Misha Moryachok

  I am trying to merge two .mp4 videos, and in some cases the seconds video part is distorted in the output video. Providing an example below :
  https://www.youtube.com/watch?v=wWMNTBWJ37A

  The real video is : https://www.youtube.com/watch?v=ASio-j-Epi8

  As you can see, we added intro before the real content, however, the real content is stretched.
  In my opinion it happens because first video is 1280x720 and the second is 460x720.

  Providing commands for merging videos :

  *1st step (convert the videos from .mp4 to .ts)

   ffmpeg -i videoPathMP4 -c copy -bsf:v h264_mp4toannexb -f mpegts videoPathTS

  *2nd step (merge videos)

  ffmpeg -i "concat:$video1 | $video2" -c copy -bsf:a aac_adtstoasc $meagePathMP4

  Video output is like you saw in provided videolink on youtube.
  I also tried to change the first video resolution to be like the second video :

  ffmpeg -i inputVideo.mp4 -s 460x720 outputVideo.mp4

  However it doesn’t helped.
  Is anyone know how to solve this ?
  Thanks

• FFmpeg sws_scale crash at certain resolution

  23 mai 2016, par Tamás Szobonya

  I’m having a weird issue with sws_scale. The problem is, that at certain resolutions i got an Access violation reading location exception. Resolutions like 1920x1080, 1600x900 works, but 1280x720 doesn’t ? This happens in a c++ cli code which is called from c#. Every project is x64 build (no Any CPU) on a Win7 x64.

  c++ cli code :

  void FFmpegWrapper::Codec::E(int width, int height, IntPtr dataIn, [Out] IntPtr %dataOut)
  
  {
  
      int ret;
  
      AVFrame *f, *fIn, *fOut;
  
      f = av_frame_alloc();
  
      fIn = av_frame_alloc();
  
      fOut = av_frame_alloc();
  
  
  
      fIn->format = AV_PIX_FMT_RGB24;
  
      fIn->width = width;
  
      fIn->height = height;
  
      ret = av_image_alloc(fIn->data, fIn->linesize, width, height, AV_PIX_FMT_RGB24, 32);
  
  
  
      f->format = AV_PIX_FMT_YUV420P;
  
      f->width = width;
  
      f->height = height;
  
      ret = av_image_alloc(f->data, f->linesize, width, height, AV_PIX_FMT_YUV420P, 32);
  
  
  
      fOut->format = AV_PIX_FMT_RGB24;
  
      fOut->width = width;
  
      fOut->height = height;
  
      ret = av_image_alloc(fOut->data, fOut->linesize, width, height, AV_PIX_FMT_RGB24, 32);
  
  
  
  
  
      uint8_t *data = (uint8_t*)dataIn.ToPointer();
  
      fIn->data[0] = data;
  
  
  
      //with or without struct no difference
  
      /*struct */SwsContext *convertCtx = sws_getContext(width, height, AV_PIX_FMT_RGB24, width, height, AV_PIX_FMT_YUV420P, 0, NULL, NULL, NULL);
  
  
  
      // CRASH here
  
      sws_scale(convertCtx, fIn->data, fIn->linesize, 0, height, f->data, f->linesize);
  
  
  
      convertCtx = sws_getContext(width, height, AV_PIX_FMT_YUV420P, width, height, AV_PIX_FMT_RGB24, 0, NULL, NULL, NULL);
  
  
  
      sws_scale(convertCtx, f->data, f->linesize, 0, height, fOut->data, fOut->linesize);
  
  
  
      dataOut = (IntPtr)fIn->data[0];
  
  
  
  }

  And its called from c# like this :

  FFmpegWrapper.Codec test = new FFmpegWrapper.Codec();
  
  
  
  Bitmap image = new Bitmap(w, h, PixelFormat.Format24bppRgb);
  
  
  
  // Get a screenshot from the desktop
  
  Screen.Capture(w, h, image, PixelFormat.Format24bppRgb);
  
  
  
  Rectangle rec = new Rectangle(0, 0, image.Width, image.Height);
  
  BitmapData bitmapData = image.LockBits(rec, ImageLockMode.ReadWrite, image.PixelFormat);
  
  
  
  IntPtr ptr = bitmapData.Scan0;
  
  
  
  IntPtr testptr1;
  
  
  
  test.E(w, h, ptr, out testptr1);
  
  
  
  // We never reach this with 1280x720 resolution
  
  Bitmap bmp = new Bitmap(w, h, w * 3, PixelFormat.Format24bppRgb, testptr1);
  
  
  
  bmp.Save(@"H:\sajt1.bmp", ImageFormat.Bmp);

  What i don’t understand is, how can it work with certain resolutions and crash with others ?
  Using 20160512-git-cd244fa-win64 version of ffmpeg.

  Edit :
  It seems, that changing AV_PIX_FMT_RGB24 to AV_PIX_FMT_BGR24 fixes it, but I’m not sure why. I know that .Net stores the pixels in bgr, but why does wrong format crashes it ? And only at some resolutions ?