Recherche avancée

Sur d’autres sites (3410)

• Why ffmpeg print SAR instead of PAR ?

  10 juillet 2016, par Jofsey

  If my knowledge is correct, SAR (Storage Aspect Ratio) is the ratio of pixel dimensions.
  For example, a 640 × 480 video has a SAR of 640/480 = 4:3.

  Whereas PAR (Pixel Aspect Ratio) is ratio of pixel height and width, and most of modern videos have square PAR (1:1).

  But when I tried to check it with ffmpeg -i I got square SAR instead of square PAR for all test videos.

  Stream #0:0(und): Video: h264 (High) (avc1 / 0x31637661), yuv420p, 1280x720 [SAR 1:1 DAR 16:9], 1758 kb/s, 24.99 fps, 25 tbr, 25 tbn, 50 tbc
  
  Stream #0:0(und): Video: h264 (Constrained Baseline) (avc1 / 0x31637661), yuv420p, 540x360  [SAR 1:1 DAR 3:2], 386 kb/s, 25 fps, 25 tbr, 25 tbn, 50 tbc
  
  Stream #0:0(und): Video: h264 (Constrained Baseline) (avc1 / 0x31637661), yuv420p,  450x360 [SAR 1:1 DAR 5:4], 328 kb/s, 25 fps, 25 tbr, 25 tbn, 50 tbc

  Is it ffmpeg bug or my mistake ?

• Merge commit ’5372cda67109848d22146289e401669266217e80’

  19 septembre 2013, par Michael Niedermayer

  Merge commit ’5372cda67109848d22146289e401669266217e80’
  * commit ’5372cda67109848d22146289e401669266217e80’ :
  
    rv10 : Validate the dimensions set from the container
  This check could be replaced by a width==0||height==0
  Merged-by : Michael Niedermayer <michaelni@gmx.at>
  

  • [DH] libavcodec/rv10.c

• How to find the actual size occupied by an AVFrame ?

  23 septembre 2013, par jsp99

  The main task at hand is to find the actual size of a decoded AVFrame.

  Here is some background information about the problem -
  I wrote a program which takes as input a media file and displays all the video frames of the file, on the screen. My Code (written in C using SDL and ffmpeg) worked perfectly with .wmv files. When it came to .mkv files, it failed (SIGSEGV) due to the following reason :

  Example : Suppose I have a decoded frame from my video stream of file (filename.mkv) in AVFrame *pFrame. The frame has a pix_fmt (pFrame->pix_fmt) of yuv420p. It has dimensions 640 * 346. The Y-plane has it's linesize (pFrame->linesize[0]) as 672.

  I think you get the point here. When I use avpicture_get_size() to get the size of the frame, it calculates a wrong size. It takes pix_fmt, width and height as arguments. Nowhere does it consider 672 instead of 640.

  avpicture_get_size() : Calculates the size in bytes that a picture of the given width and height would occupy if stored in the given picture format.

  I got around this problem by adding some yuv420p frame specific code, which can find the size of the frame in situations like above.

  • Can someone explain why are most of the decoded frames (from video streams of most of the formats) this way ? And why is it not this way with the wmv format ?

  But if I have to find the actual size occupied by an AVFrame of another format (Eg: yuv410p), I have to write the format specific code again. Coming to the main problem,

  • How can I find the actual size occupied the decoded AVFrame (size in bytes occupied by the data of the AVFrame) ?

  • I tried using av_image_get_buffer_size(), but I couldn't understand how to use the align parameter. Can someone explain the usage of the parameter align ? Tried using many other functions in pixdesc.c, but I am missing something.