[Libav-user] Is a buffer required while streaming using ffmpeg code?

[Navin]

As per the ffmpeg tutorial code, for code like this:/
while(av_read_frame_proc(pFormatCtx, &packet) >= 0)
   {
     if (packet.stream_index == videoStream)
     {
       avcodec_decode_video2_proc(pCodecCtx, pFrame, &frameFinished, 
&packet);// Decode video frame
       if (frameFinished)
       {
         sws_scale_proc(sws_ctx, (uint8_t const * const *)pFrame->data, 
pFrame->linesize, 0, pCodecCtx->height, pFrameRGB->data, 
pFrameRGB->linesize );/

As a general question, if I'm receiving realtime webcam frames or from a 
remote video file from a very fast network and I receive frames faster 
than the framerate of the video, I'd need to buffer the frames in my own 
datastructure. But if my datastructure is small, then won't I lose a lot 
of frames that didn't get buffered? I'm asking because I read somewhere 
that ffmpeg buffers a video internally. How does this buffering happen? 
How would that buffering be any different than if I implemented my own 
buffer? The chances that I'd lose frames is very real, isn't it? Any 
chance I could read up or see some source code about this?
Or does all this depend on the streaming protocol being used?

-- 
Navin

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://ffmpeg.org/pipermail/libav-user/attachments/20121128/9c7b8efd/attachment.html>