Recherche avancée

Sur d’autres sites (8939)

• FFmpeg transparent PNG black outline issue

  5 février 2024, par kklmarvelmedia

  I'm encoding a video with a transparent PNG using ffmpeg. I noticed there's a slight black outline surrounding the image. Is there any way to remove it ?

  



  Output image :


  



  Transparent PNG sample :


  



  My ffmpeg command

  



  ffmpeg -hide_banner -y -ss 0.0 -t 8.5 -i C:\Users\Admin\Desktop\test_movies\6.mp4 -i C:\Users\Admin\Desktop\test_movies\text_and_emoji.png -filter_complex [0:v]setpts=PTS-STARTPTS,scale=640:640:force_original_aspect_ratio=decrease,pad=640:640:(ow-iw)/2:(oh-ih)/2:color=#18ffff[0v];[1:v]scale=556.24744:141.41884[1v];[0v][1v]overlay=(W-w)/2-(W/2-325.33328):(H-h)/2-(H/2-567.7075):enable='between(t,0.0,8.5)' -ac 2 -ar 44100 -vcodec libx264 -g 75 -r 20 -preset ultrafast -strict experimental C:\Users\Admin\Desktop\test_movies\test.mp4


  



  Last edit 1 :

  



  I tried using without [1:v]scale=556.24744:141.41884[1v], the output still have the slight outline

  



  Sample output :


  



  Sample code :

  



  ffmpeg -hide_banner -y -ss 0.0 -t 8.5 -i C:\Users\Admin\Desktop\test_movies\white.mp4 -i C:\Users\Admin\Desktop\test_movies\text_and_emoji.png -filter_complex [0:v]scale=640:640:force_original_aspect_ratio=decrease,pad=640:640:(ow-iw)/2:(oh-ih)/2:color=#18ffff[0v];[0v][1:v]overlay=(W-w)/2-(W/2-325.33328):(H-h)/2-(H/2-567.7075):enable='between(t,0.0,8.5)' -ac 2 -ar 44100 -vcodec libx264 -preset ultrafast -strict experimental C:\Users\Admin\Desktop\test_movies\test.mp4


  



  Last edit 2 :

  



  I tried another one with added alpha=premultiplied with latest ffmpeg version. It somehow removed the outline, but the quality of the picture reduced alot till it seems like it's pixelated. Plus. there's another unknown white layer at the back of the image.

  



  Output video


  



  Sample code :

  



  C:\Users\Admin\Downloads\ffmpeg-20180102-57d0c24-win64-static\bin\ffmpeg -y -ss 0.0 -t 8.5 -i C:\Users\Admin\Desktop\test_movies\white.mp4 -i C:\Users\Admin\Desktop\test_movies\text_and_emoji.png -filter_complex [0:v]scale=640:640:force_original_aspect_ratio=decrease,pad=640:640:(ow-iw)/2:(oh-ih)/2:color=#00ffff[0v];[1:v]scale=480:120[1v];[0v][1v]overlay=(W-w)/2-(W/2-325.33328):(H-h)/2-(H/2-567.7075):alpha=premultiplied:enable='between(t,0.0,8.5)' -ac 2 -ar 44100 -vcodec libx264 C:\Users\Admin\Desktop\test_movies\test.mp4


  



  Latest edit 3 :

  



  As suggested by @Mulvya, I combined his code with alpha=premultiplied and it seems alot better now, with very slight black outline (almost not visible)

  



  Output video :


  



  Sample code :

  



  C:\Users\Admin\Downloads\ffmpeg-20180102-57d0c24-win64-static\bin\ffmpeg -y -ss 0.0 -t 8.5 -i C:\Users\Admin\Desktop\test_movies\white.mp4 -i C:\Users\Admin\Desktop\test_movies\text_and_emoji.png -filter_complex [0:v]setpts=PTS-STARTPTS,scale=640:640:force_original_aspect_ratio=decrease,pad=640:640:(ow-iw)/2:(oh-ih)/2:color=#18ffff[0v];[1:v]premultiply=inplace=1,scale=480:120[1v];[0v][1v]overlay=(W-w)/2-(W/2-325.33328):(H-h)/2-(H/2-567.7075):alpha=premultiplied:enable='between(t,0.0,8.5)':format=rgb,format=yuv420p -ac 2 -ar 44100 -vcodec libx264 C:\Users\Admin\Desktop\test_movies\test.mp4


  


• FFMPEG How to resize image to 1280 x 720 but with aspect ratio - if ratio not exact how to add black background [on hold]

  26 septembre 2019, par Aster

  FFMPEG How to resize image to 1280 x 720 but with aspect ratio.
  If ratio not exact how to add black background ?

  windows 10
  
  
  
  ffmpeg -y -i c:/img/image1.jpg -vf scale=1280 x 720 c:/img/image2.jpg

• getting black and white image after encoding

  16 mai 2012, par user1310596

  I am trying to encode image using ffmpeg library and objective c. I am using following code but I am getting black and white image as the result. Is there something to do with pixel format(PIX_FMT) ? Please help me so that I can get colored image.

  av_register_all();
  
  avcodec_init();
  
  avcodec_register_all();
  
  avformat_alloc_context();
  
  
  
  AVCodec *codec;
  
  AVCodecContext *ctx= NULL;
  
  int out_size, size, outbuf_size;
  
  AVFrame *picture;
  
  uint8_t *outbuf;
  
  unsigned char *flvdata = malloc(sizeof(unsigned char) * 30);
  
  
  
  
  
  outbuf_size = 200000;
  
  outbuf = malloc(outbuf_size);
  
  
  
  
  
  printf("Video encoding\n");
  
  
  
  codec = avcodec_find_encoder(CODEC_ID_FLV1);
  
  if (!codec) {
  
          fprintf(stderr, "codec not found\n");
  
          exit(1);
  
  }
  
  
  
  ctx= avcodec_alloc_context();
  
  picture= avcodec_alloc_frame();
  
  
  
  
  
  ctx->width = 320;
  
  ctx->height = 240;
  
  ctx -> sample_rate = 11025;
  
  ctx -> time_base.den = 1000;
  
  ctx -> time_base.num = 23976;
  
  ctx -> codec_id = CODEC_ID_FLV1;
  
  ctx -> codec_type = CODEC_TYPE_VIDEO;
  
  ctx->pix_fmt = PIX_FMT_YUV420P;
  
  
  
  if (avcodec_open(ctx, codec) < 0) {
  
          fprintf(stderr, "could not open codec\n");
  
          exit(1);
  
  }
  
  
  
  outbuf_size = 100000;
  
  outbuf = malloc(outbuf_size);
  
  size = ctx->width * ctx->height;
  
  
  
  AVFrame* outpic = avcodec_alloc_frame();
  
  int nbytes = avpicture_get_size(PIX_FMT_YUV420P, ctx->width, ctx->height);
  
  
  
  uint8_t* outbuffer = (uint8_t*)av_malloc(nbytes);
  
  
  
  fflush(stdout);
  
  
  
  int numBytes = avpicture_get_size(PIX_FMT_YUV420P, ctx->width, ctx->height);
  
  
  
  UIImage *image = [UIImage imageNamed:[NSString stringWithFormat:@"0.jpg"]];
  
  CGImageRef newCgImage = [image CGImage];
  
  
  
  CGDataProviderRef dataProvider = CGImageGetDataProvider(newCgImage);
  
  CFDataRef bitmapData = CGDataProviderCopyData(dataProvider);
  
  long dataLength = CFDataGetLength(bitmapData);
  
  
  
  uint8_t *buffer = (uint8_t *)av_malloc(dataLength);
  
  buffer = (uint8_t *)CFDataGetBytePtr(bitmapData);
  
  
  
  for(int i = 0; i < dataLength; i++)
  
  {
  
          if((i + 1) % 16 == 1 && i != 1)
  
                  printf("\n");
  
          printf("%X\t",buffer[i]); // getting something different than the     actual hex value of the image
  
  }
  
  
  
  
  
  
  
  outpic -> pts = 0;        
  
  
  
  avpicture_fill((AVPicture*)picture, buffer, PIX_FMT_RGB8, ctx->width, ctx->height);
  
  
  
  avpicture_fill((AVPicture*)outpic, outbuffer, PIX_FMT_YUV420P, ctx->width, ctx->height);
  
  
  
  struct SwsContext* fooContext = sws_getContext(ctx->width, ctx->height, 
  
                                                         PIX_FMT_RGB8, 
  
                                                         ctx->width, ctx->height, 
  
                                                         PIX_FMT_YUV420P, 
  
                                                         SWS_FAST_BILINEAR, NULL, NULL, NULL);
  
  
  
  sws_scale(fooContext, picture->data, picture->linesize, 0, ctx->height, outpic->data, outpic->linesize);
  
  
  
  printf("abcdefghijklmnop");
  
  out_size = avcodec_encode_video(ctx, outbuf, outbuf_size, outpic);
  
  printf("\n\n out_size %d   outbuf_size %d",out_size,outbuf_size);

  Thanks in advance