Re: Chapter marks not where specified.

On Fri, Mar 30, 2012 at 3:25 PM, Tom Brock <tomb-yZYHXxUkMT10YYv+nEap2lpr/1R2p/[email protected]> wrote:
>> Because remember what I said
>> elsewhere about VOBU boundaries, and checking some FFmpeg-converted
>> MPEG footage I happen to have lying handy seems to indicate that VOBUs
>> are typically about 0.5 seconds in duration.
>
> Hold the bus!  All the I've read says that chapter marks must be I-frames,
> with no mention of also having to coincide with the start of VOBUs; indeed,
> how would one be supposed to ensure this?.  I realize that VOBUs are the only
> addressable objects on a DVD, and had assumed (there I go again) that when
> dvdauthor said it was "fixing VOBUs" that it was ensuring that the a new VOBU
> starts at each chapter mark.  If this is not the case, then, surely, trying to
> specify chapter marks with any precision is entirely futile unless ffmpeg
> implement a -force_vobu_boundary' option instead of '-force_key_frames', which
> is rendered useless.
>
> Clarification would be *extremely* welcome.

Chapter (really, any) boundary must be on a VOBU boundary.  VOBU
boundary should be at an I frame.  There shouldn't be more than one I
frame per VOBU, if so, your multiplexer sucks (use mplex from
mjpegtools -- the one I always tested/developed with).

Other "shoulds" include 2 B frames per P frame, minimum GOP length of
(I think) 9 frames, and max of (I think) 18 frames -- though that
might be NTSC dependent.

(having the right number of B frames helps with hardware decoders that
won't FF/REW well if you only have I/P frames)
-- 
        Scott

------------------------------------------------------------------------------
This SF email is sponsosred by:
Try Windows Azure free for 90 days Click Here 
http://p.sf.net/sfu/sfd2d-msazure