[mythtv] Transcoding of DVB-T recording expands the file

Kenneth Aafløy ke-aa at frisurf.no
Sat May 22 00:16:40 EDT 2004


On Friday 21 May 2004 20:27, Geoffrey Hausheer wrote:
> On Fri, 21 May 2004 13:58:15 -0400 (EDT), "Daniel Thor Kristjansson" wrote:
> > From the reading I've been doing on MPEG-2... GOPs are not required for
> > a valid stream, they are only a requirement in MPEG-1. Which means we
> > can't depend on hardware encoders or ATSC/DVB output to always contain
> > GOPs.

Thanks for letting me know, section 6.1.1.7, 'I-pictures and group of pictures 
header' in the 13818-2 for anyone else that has not completly dived into the 
maze yet.

> You are right, and Myth doesn't rely on them.  However, the transcoder
> does rely on them, and fixing that isn't one of my priorities as it
> complicates the processing quite a bit (the whole mpeg2trans probably
> needs either a significant amount of work or to be rewritten).

Ahh, so to remove this limitation (and possible bug, as we have yet to receive 
aconfirmation of what caused the bug) essentially what needs to be done is if 
noe GOP is discovered before a I-Frame construct a virtual GOP, then put all 
pictures/frames in this GOP untill the next I-Frame, right? Does not sound 
hard at all, does it?

-- 
Kenneth *

* Looking forward to the day I can contribute to the Linux Community!


More information about the mythtv-dev mailing list