[XviD-devel] [TEST] Syskin's ME weird

[Edouard Gomez]

Michael Niedermayer (michaelni@gmx.at) wrote:
> hmm, can u output the modulo_time_base=time_incr too ...

I have no  modulo_time_base data, but i have a  time_base data that is
always equal  to 0 (zero).  Perhaps the stable version  discarded some
bframe information in decoding functions.

You'll find my matrix.gom2 file at :
http://perso.wanadoo.fr/ed.gomez/files/mp4ref/matrix.bin (1.8Mb)

Header : GOM2 or 2MOG (4b) (allows apps to detect endianess)

then frames :
Frame size (4b) (in arch specific endianness, mine is little endian)
Frame data ('Frame size' bytes)

I'm sorry  but i had  no time  to prepare a  nice patch for  both XviD
(allow  dynamic  debugging level)  and  xvid_decraw (container  format
support).

PS : i  forgot to mention i have encoded this  lobby matrix scene with
max_bframe == 2. (noticeable with the first 2 N_VOPs)

-- 
Edouard Gomez