[XviD-devel] Bframes almost finished

[Christoph Lampert]

> >Hi,
> >so it looks a little garbled)... There is lot of debug output,
> >please look at the Time= N  values:
> >
> >They start fine, with 0(I) 3(P) 1(B) 2(B) 6(P) 4(B) 5(B) etc.
> >but then after Frame   28: and DecFrame   28: the P-frame(!) decoding
> >seems to "fail": it returns after 5 decoded bytes instead of 6 as usual
> >and the timestamp is 2 instead of 27. After that, bytestream seems to be
> >broken, the B-frame is identified correctly, but time_bp/time_pp is wrong
> >and the image gets buggy.

Hi,
I found the bug, but didn't attack it, yet. It's due to wrongly written
time_incr/time_allthestuff. 
internal P-frame numeration/timestamps gets out of sync by that (at least
with the current code in mbcoding.c ) and therefore B-frames don't know
the correct distance to next I/P-frame. 

Does anyone know enough to fix it "soon", or should I look into it
myself?

gruel