Does the video decode need to decode something each time?
I'm writing the wrapper between MPlayer and the video decoder such like
I have some questions about the buffer and flow control. Say if the vdo
decoder can only start to decode after he gots more than one NAL frame's
bitstreams. What does the wrapper need to do?
Can it be done that wrapper returns something that can tell mplayer to keep
going to put more streams down to wrapper?
In dev_video.c::decode_video(), it seems always wants wrapper to return mpi
back to it.
mpi = mpvdec->decode(sh_video, start, in_size, drop_frame);
What can I do to make Mplayer know that I need more NAL streams downto
wrapper so that decoder can start to decode. It doesn't seem drop_fram or
skip frame situation.
Thanks a lot!
---------- Forwarded message ----------
From: boku <[hidden email]>
Date: Dec 28, 2007 3:02 PM
Subject: What if different colorspace for codec and display output?
To: FFmpeg user questions and RTFMs <[hidden email]>, "MPlayer
usage questions, feature requests, bug reports" <[hidden email]>
I have some questions about the video output and video filters (vf_vo.c).
1) Can a decoder decode input bit streams in different kinds of color
space format such as YUV420P, IYUV or UYVY422?
Or it only accepts one format such as YUV420? Also, where it can be defined?
2) Say if my specific mpeg decoder will decode YUV420P bit stream and
can only output one format of UYVY422 packed frame data,
how can I write my wrapper, vd_xxx.c in libmpcodecs to let video
output layer or the video filter change to the color space for the
3) For FFMPEG wrapper into mplayer, vd_ffmpeg.c, I checked out the
match mechanism in ::init() as folows:
lavc_codec = (AVCodec
it means mplayer uses "dll" field specified in codecs.conf to match
However, I checked the document and it seems "dll" field is not for
this definition or purpose. Am I misunderstanding?