[XviD-devel] inter/intra screener

Andrew Voznytsa av at polynet.lviv.ua
Sat Dec 18 06:26:09 CET 2004


Hi,

you may try to use MB boundary error. See 
ftp://jvttest:tester@standards.polycom.com/2003_12_Waikoloa/JVT-J033r1.doc, 
section 3,*/ "Selective intra mode decision".
IMO this should work for MPEG-4 ASP too.
/*

Best regards,
Andrew Voznytsa



Radek Czyz wrote:

> Hi everyone,
>
> I'm trying to "screen" some macroblocks from R-D INTRA calculations - 
> most macroblocks will never be intra, or even close, and it's 
> unnecessary to calculate their intra R-D.
>
> This is quite important because INTRA R-D is very slow - it handles 
> large coefficients (cache misses in calc_bits() ), bits are always 
> added twice (with and without AC/DC prediction) and there is the extra 
> step of ac/dc prediction performed.
>
> My attempts with SAD vs DEV are a huge failure :/ even if I skip only 
> some macroblocks (speed gain less than 1%) I'm still loosing much of 
> psnr.
>
> I'll try different appoaches (R-D vs deviation, and a 8x8-based 
> deviation rather than 16x16-based) but perhaps you can suggest 
> something better? I'm not interested in decision itself (that will be 
> RD-based) but with some reliable method of measuring "best case 
> intra", to drop the calculations if "best case" is bad enough.
>
> Currently I'm looping through all blocks one-by-one, doing 
> transfer->dct->quantization->calculate_bits(1)->prediction->calculate_bits(2)->dequant->calculate_distortion. 
>
> And I'm checking for early skip after each block.
> Perhaps reorganizing it would be better..... Any ideas?
>
> Radek
> _______________________________________________
> XviD-devel mailing list
> XviD-devel at xvid.org
> http://list.xvid.org/mailman/listinfo/xvid-devel



More information about the XviD-devel mailing list