[FFmpeg-devel] dv_guess_dct_mode

Michael Niedermayer michaelni
Fri Feb 6 21:02:25 CET 2009


On Wed, Feb 04, 2009 at 04:41:18PM -0800, Roman V. Shaposhnik wrote:
> On Wed, 2009-02-04 at 23:00 +0100, Michael Niedermayer wrote:
> > > Comments are welcome.
> > 
> > see vsad_intra16_c() and ildct_cmp
> 
> Thanks, from the first glance these look exactly like what I need.
> Although I'm confused about one thing: ildct_cmp is only set
> in MPV_encode_init, shouldn't it be initialized for all the
> codecs?

i guess its so far only used by mpv


> 
> > these would need to be extended with a width=8 intra variant, arrays would
> > need to be enlarged to [6] ...
> 
> What's the convention for array indexes? IOW, how's one to know that
> a width=8 intra variant supposed to be @[5] ?

[a+1] is half the size w&h than [a] except when [a] is already the smallest
supported


> 
> Also, I'm a bit confused about one thing: if I only implement 
> width=8 intra for vsad/vsse case and the user picks a different
> class via -ildctcmp, should I just issue an error message
> "non implemented" or something?

yes, but its like just 3 lines to build a width=16 out of width=8


> 
> > also supporting full RD would be nice, that is encoding both and picking
> > the actually better instead of some heuristic
> 
> Sure. But wouldn't that be horrendously slow?

encoding a frame twice would be half the speed, same if each block is
encoded twice

[...]
-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

When you are offended at any man's fault, turn to yourself and study your
own failings. Then you will forget your anger. -- Epictetus
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/attachments/20090206/4adc0d94/attachment.pgp>



More information about the ffmpeg-devel mailing list