[XviD-devel] better trellis?

[Christoph Lampert]

On Fri, 3 Sep 2004, Michael Militzer wrote:
> So probably, your lambda should also take motion or current avrg. texture 
> size into consideration. On the other hand: from a HVS perspective we may
> not like to have best quality in high motion scenes as the human eye may
> not be capable at all to perceive the extra quality here - oh well, just
> some thoughts...

Basically, chosing a variable lambda boils down to optimizing a different 
quality metric than SSD, hopefully HVS-based. There is of course prior 
work on that, e.g. (for the "reasonable" ones):

http://ise.stanford.edu/class/ee392j/projects/xiao_report.pdf

He classifies scenes as static or not, and then uses just the MPEG 
quantization matrix for static scenes, and a weighted power thereof for 
non-static. So, maybe using a similar approach based on a typical weight 
matrix would be a good starting point. 

chl