Hi!

Sven Over wrote:

> I guess the treatment of the DVCBUT file by the script is correct (apart from 
> that off-by-one thing). I would guess the point is, that when re-reading the 
> mpg file into DVCBUT, it doesn't read it in full. When I remember correctly, 
> the indexer of DVBCUT stops at an I-Frame and ignores the last P/B-Frames in 
> the file. This is, I must admit, an unneccessary limitation which probably 
> originates from my usual usage of DVBCUT, using huge bulk TS-Files as source. 
> There you don't mind missing a few frames in the end... a future improved 
> implementation should not have this limitation.

The indexer contains some design flaws anyway. E.g. pictures are
currently stored in file (decoding) order. To access them in display
order, you have to go through index::indexnr() again and again. That
function in turn fails badly if a sequence header is missing due to
transport errors: There may be pairs of pictures with identical sequence
numbers, but indexnr() will always return the first one from the
sequence. With an index stored in display order, indexnr() would be a
simple subtraction that can't go wrong. Besides that it would enable
dvbcut to detect missing pictures and headers more easily: If the
sequence numbers aren't contiguous, something must be wrong with the stream.

Sequence number counting is slightly wrong, too. The code assumes that
sequence numbers are reset when a sequence header is encountered.
According to my documentation, it's reset after each GOP header.
Fortunately, in all DVB (and also DVD) material I've seen so far a GOP
header is immediately following the sequence header, so it doesn't make
a *big* difference.

When recutting, there are at least two problems: First, you can't
position the slider *past* the final frame (which would be required in
order to include it). Second, the last few frames are neither displayed
nor written. So every time you recut a file, it is shortened by at least
two frames - no matter if there are B frames at the end or not. I
suspect this is an indexing problem as well, and/or maybe an off-by-one
error somewhere.

Guess what my next project will be? Don't hold your breath, though -
that may take quite a while.
-- 
Michael "Tired" Riepe <[EMAIL PROTECTED]>
X-Tired: Each morning I get up I die a little

-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
_______________________________________________
DVBCUT-user mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/dvbcut-user

Reply via email to