I think I finally found the best of both worlds...

[Roger Merchberger]

I'm trying something new that _should_ increase the quality of my
SVCD's and decrease their encoding time... lemme explain:

I'd seen somewhere that someone else was using DVD2AVI to edit
commercials out of video clips - so I started tinkering with that,
and much to my chagrin, it won't work for what I'm using it for. ISTR
that they were converting movies to high-quality AVI's which were
taking ~ 10G of space each [more space than I have to spare], then
using those clipping out the commercials so each segment was a
"project" then batch encoding the segments in TMPGenc... which is
good for a few things:

1) TMPGenc's cut/merge options are flaky, at best -- reducing the
reliance on that will improve "glitches" in the show (I have a few
shows that are "more than a few" seconds off...)

2) The old way, you need to encode the whole show, then piece it
apart -- requiring more disk space. Having the 5 seperate segments
without the commercials in-between means you don't have to allocate
extra disk space for 2 "copies" of a show.

3) TMPGenc doesn't have to encode the commercials, saving some
serious CPU time. -- Yes, even on my Dual Athlons, saving time is a
Good Thing (TM).

The reason DVD2AVI won't work for me is I'm not encoding the entire
show to AVI - just "enough" so that NanDub can show me exactly where
the commercials are. DVD2AVI cannot work with separate video / audio
streams. So when I set DVD2AVI to save a project as only one part of
a show, then open the full audio file in TMPGenc, the audio was
offset by the amount of the "clipped" video that wasn't being encoded
at the beginning of the show.

So, back to the drawing board... and this is what I came up with:

1) The usual - ES/sendstream -> nc -> splitstream -> WinAmp to .wav
the audio... this hasn't changed, so I won't get into too much
detail.

2) DVD2AVI the full video stream into a project, just like always.

3) VFAPI-Conv into the small AVI for Nandub, just like always.

[tangent] Here's where it changes -- I *used* to encode the entire
show in TMPGENC, then use NanDub to find the HH:MM:SS.TT of the
commercial breaks and use those times in the cut/merge function of
TMPGenc to break-apart the video into the final product, entering in
big-time errors... [end-tangent]

4) Now: Use NanDub next - and instead of HH:MM:SS.TT, record the
exact frame number of where you want the program segments to start &
stop. Jot/Notepad these down.

5) Open TMPGenc, and load all your other favorite encoding filters,
and when you get to the "Source Range" section put the audio offset
you got from splitstream (remember, if it's a negative number in SS,
it's a positive offset for TMPGenc) and use the start & end frame
numbers you got from NanDub for the first section, then save the
project as ShowName-1.tpr. Reopen the filters, and change the start &
end frames to the second section, ***Make sure you change the name of
the output file to ShowName-2.mpg, or else you'll overwrite the first
section*** and save as ShowName-2.tpr. Continue for each segment of
the show.

6) Go to "File", "Batch Encode" and load each of these segments, then
run the batch. Voila!!! :-)

I'm still running my first batch encoded this way, so I don't have a
full report yet, but it *looks* like TMPGenc is clipping the video on
the exact frame I want, not 2-3 seconds off like the cut/merge
does...

An added bonus is if you set up Nero to handle chapters, you can skip
chapter to chapter for each segment instead of relying solely on fast
forward & rewind.

Hope this helps! Happy incr(Year),
Roger "Merch" Merchberger

__________________________________________________
Do You Yahoo!?
Send your FREE holiday greetings online!
http://greetings.yahoo.com