[XviD-devel] Quality optimization

[Christoph Lampert]

On 27 Feb 2003, skal wrote:

> 	In fact, having the Hadamard transform done doesn't mean
> 	you're off with the hard work. Taking the abs values with
> 	MMX is painful. For SSE, we have the mighty 'psadbw' instr.,
> 	but it works on 8bits data, whereas the output of Hadamard
> 	is 11bits (yes, it's scaled by 8, not 64!:). So? Should we
> 	descale 11->8bits? 

Sure... results are gonna be quantized anyway... or would it be possible
to do Hadamard in saturated 8 bit (7 bit+sign?), acting on the
residue? Because if value overflows, then we most likely won't use the
block anyway. 

>       Another idea would be to multiply the
> 	Hadamard output by a pseudo-quant matrix that mimics the real
> 	quantizers (and the missing cosines, maybe)... Dunno. 
> 	Anyway, here's an Hadamard_SAD for 8x8 or 16x16 byte input, 
> 	in replacement for SAD. I'm not quite satisfied with it
> 	mainly because of the above (and not just because it's
> 	8 times slower than pure SAD :)

Thanks! I'll check this weekend (no time before)

gruel