Page 1 of 1

[RESOLVED] DGIndex vs. DgIndexNV d2v creation

Posted: Tue Nov 06, 2012 2:00 pm
by Selur
A user of Hybrid (small gui I wrote, that also has the possibility to use avisynth for filtering on windows systems) send me a two links to a file (http://aavg.net/g42ito8d63q2 and http://cyberlocker.ch/s7du6gmddiz2, both to the same file) which causes problems with DGIndex. To be more precise the problem is that DGIndex creates a .d2v and a d.d2v.bad file along with a fixed.txt and when using the .d2v file as source the video has 13064 instead of 11510 frames and therefore get's async during further processing. If the .d2v.bad file is used instead everything is fine.

A friend of mine tested the file with DgIndexNV and it handles the file without a problem (returns 11510), so my question is, is this due to a feature that is NVIDIA only or is it a feature that might get incorporated into DGIndex (not sure if it is still under development).

Thanks for reading. :)

Cu Selur

Re: DGIndex vs. DgIndexNV d2v creation

Posted: Sun Nov 11, 2012 12:38 pm
by admin
Sorry, can't look at it until I get home tomorrow evening.

Re: DGIndex vs. DgIndexNV d2v creation

Posted: Mon Nov 12, 2012 5:59 pm
by admin
Your links give me virus alerts and blockages, and they pop up crap on my screen.

Please give me a link to a safe reputable website if you want me to look at this.

Re: DGIndex vs. DgIndexNV d2v creation

Posted: Mon Nov 12, 2012 6:07 pm
by Selur
that were the links I got from the user, but for your convenience, if google is reputable enough I'll upload it to my google drive and send you a link

Re: DGIndex vs. DgIndexNV d2v creation

Posted: Tue Nov 13, 2012 7:54 am
by admin
It's a mix of hard and soft telecine. In the hard telecined parts there are LOTS of signalled field order transitions but they appear to be spurious. You should not use the "fixed" D2V in this case as the corrections add fields that produce the wrong frame count.

Bottom line: the stream is screwed up.

DGDecNV does not have the functionality to detect and correct field order transitions, so you don't see this.

Re: DGIndex vs. DgIndexNV d2v creation

Posted: Tue Nov 13, 2012 8:08 am
by Selur
It's kind of a bummer to encode a bunch of files and later detect that one should not have used the 'fixed' .d2v, but the 'bad' one, so is there a way to 'detect' such streams automatically by parsing the fixed.txt ?
(I could to the parsing myself, I just don't know what to look for)

Re: DGIndex vs. DgIndexNV d2v creation

Posted: Tue Nov 13, 2012 8:55 am
by admin
I'm really not interested in doing anything to support processing of severely broken streams. Hopefully, they will be rare.