[vdr] Approach for frame exact A/V cutting (not only I-frames)

["Dr. Werner Fink" <werner@suse.de>]

Hi,

      just for protocol due to lack of time.  IMHO it is
possible to modify VDR in such way that a) the marks saved
within marks.vdr could also add the numbers of frames/slices
after the marked I-frame within the comment and b) modify
cutter.c in such way that it access the data of the I-frame
before and after the CutIn/CutOut I-frames (six I-frames).

To get the audio data in sync with the video data the audio
packages of the the audio data has to be sorted with regards
to the PTS delay of the CutIn/CutOut switch.  For real audio
this could be done in ms resolution if the few audio frames
will be decoded into PCM, appropriate cut, and encoded back
into mpeg2 audio frames. For DD the 32ms frames delay as
resolution has to be enough.  This something like partly
(re)multiplexed.

To get the video frames after the CutIn/CutOut I-frame the
video data of the CutIn/CutOut I-frame may converted
into a sequence of still pictures (e.g. by the usage of GrabImage()
or a software decoder like libmpeg2/libmpeg3) to get all
informations even if the CutIn or CutOut is placed on a B-frame.
With this pictures, and the offset information within the
comments of the marks found in marks.vdr, one or two new
I-frames together with the new audio frames can replace the
former I-frames around the CutIn/CutOut switch.

IMHO this could be done ;)

        Werner


-- 
Info:
To unsubscribe send a mail to ecartis@linuxtv.org with "unsubscribe vdr" as subject.