PDA

View Full Version : Enabling BFRAME

CavalloPazzo
29th September 2002, 19:46
I've enable BFRAMES with latest source of XVID (26/9) (doing as Koepi said in this thread http://forum.doom9.org/showthread.php?s=&threadid=34550 )and I have two qustion...
It seems that it's not possible to change quantization mode when bframes are enabled. Is it normal?
Encoded clip can be decoded without errors only by xvid decoder and not by DIVX and FFDSHOW, even if DX-50 BVOP compatibility is enabled. Does the problem is in XVID encoded clip or in other decoders?
Koepi
29th September 2002, 20:09
That's why it's in the devel branch - it's not for testing and those issues are known and discussed amongst the developers. sysKin is investigating what went wrong with his ME merge.
SirDavidGuy
29th September 2002, 21:11
Encoded clip can be decoded without errors only by xvid decoder and not by DIVX and FFDSHOW, even if DX-50 BVOP compatibility is enabled

Other decoders have problems with custom quant matrices, B-Frames or not. Either they don't support them, or XviD is using an incorrect format for them.
-h
29th September 2002, 21:58
Other decoders have problems with custom quant matrices, B-Frames or not. Either they don't support them, or XviD is using an incorrect format for them.

They do? I tested the latest custom quant matrix code with libavcodec and DivX 5.0x, and both rendered correctly.

Didn't try B-frames of course.

-h
SirDavidGuy
30th September 2002, 00:30
?????

Weird.

I couldn't get either of them to work (Using the "Perfect Matrix").

Perhaps whichever matrix you used was so similiar to either Mpeg or h.263 that it didn't cause any problems when decoding, since the ouput coefficients from the inverse quantization would be so similiar. Unlikely, but possible. I'll test again, and see if I made a mistake somewhere.
-h
30th September 2002, 02:34
Perhaps whichever matrix you used was so similiar to either Mpeg or h.263 that it didn't cause any problems when decoding, since the ouput coefficients from the inverse quantization would be so similiar. Unlikely, but possible. I'll test again, and see if I made a mistake somewhere.

They were variations on MPEG now that you mention it. That wouldn't be good at all..

I'll have a play and see what happens.

Edit: libavcodec (recent version) works correctly, Divx 5 performs much less admirably. Bizarrely enough, Sigma's RMP4 fork doesn't work properly either.

-h