|
This new core mode is for NTSC sources that otherwise fail to
transcode with other strategies and yield bad audio/video
synchronization results. It also only works with MPEG
program streams in connection with
import module "-x vob". In general, it's a good idea to
move the various chunks of the big program stream, we
previously ripped from a DVD, into a
directory, we denote as vob_dir. Please see the
DVD backup guide for details. We
will need to put the "*.VOB" files into this extra directory
to make the directory mode of tccat and transcode
work. Please keep the vob_dirclean of other files.
Restrictions: Options -c /
-M3 / -M4 / -L / -S do not work for this mode because
transcode
sets some of the parameters internally for navigation.
The first step is to create the navigation logfile
nav_log containing the frame and related group of picture (GOP) list with file offsets:
|
tccat -i vob_dir | tcdemux -W >nav_log |
A typical output of the command looks as follows:
Sisko:cluster>tccat -i vob_dir | tcdemux -W >nav_log
(seqinfo.c) 284 video frame(s) in unit 0 detected
(seqinfo.c) 413 video frame(s) in unit 1 detected
(seqinfo.c) 3506 video frame(s) in unit 2 detected
(seqinfo.c) 150662 video frame(s) in unit 3 detected
This result shows, that the entire stream is made up of 4
program stream units (PSU), that are defined by a reset of the
presentation time stamp (PTS) in the stream pack header. This
situation may require a re-synchronization at the beginning of
each PSU, which transcode can only handle in this new core mode.
This trick is to restart import at every PSU. We need to provide only
two flags (long options) to make it work:
|
transcode [...] -i vob_dir --psu_mode movie --nav_seek nav_log |
This will eventually produce the 4 output files:
movie-psu00.avi, movie-psu01.avi, movie-psu02.avi, movie-psu03.avi.
transcode does not switch off the audio or video
encoder. However, a video keyframe is forced at the beginning of
each output file so each file played individually. This will make
the chunks perfect candidates for avimerge. The above
example shows small units before the main movie. In most cases, these
junk trailers or film company/distributor intro logos and can be
skipped with the additional option:
--psu_chunks 2-4
Simlar to the "-c" option, only units 2 and 3 but not 4 are
encoded.
Other movies are even more complicated and have a
chapter-like/episodes superstructure imposed with a large number
of PSUs. In this case, you will need to use this new core mode.
If you prefer 2-pass encoding, latest
0.6.0pre5 snapshots provide another long option
--no_split
that will generate only a single output file
movie-psu.avi for proper 2-pass bitrate estimation, valid
for the entire movie.
Notes:I have noticed that some
frames are dropped at the end of a unit, which does not break AV sync
but lead to a jump in the audio track. I will try to
improve this issue, but somehow audio is short in frames.
|