[XviD-devel] a bug, I think

[Christoph Lampert]

On Sat, 1 Mar 2003, Radek Czyz wrote:
> I found a bug. If adaptive quantization is active and b-frames are
> used, bframes have strange quantizers. For example two frames have an
> average quantizer of 4..5 while a bframe between them has quant 13,
> and that's with ratio 150% and offset 100...
> 
> I guess that lumimasking changes the quantizer for a frame to match
> quantizer of first macroblock. That would be logical, BUT if b-frames
> use this quantizer to compute their own, it's a bug.
> 
> Is it the case? I'll check that myslf tommorow, but if anyone can
> give me a good answer or tip how to fix it, I'll be grateful.

I remember that bug from an earlier post. Hasn't it been fixed a while
ago? 

gruel