[FFmpeg-devel] [PATCH] mpeg2: fix block_last_index when mismatch control modifies last coeff

[Michael Niedermayer]

On Mon, Jun 21, 2010 at 07:33:26AM -0700, Jason Garrett-Glaser wrote:
> On Mon, Jun 21, 2010 at 7:23 AM, Michael Niedermayer <michaelni at gmx.at> wrote:
> > On Mon, Jun 21, 2010 at 04:17:58PM +0200, Michael Niedermayer wrote:
> > [...]
> >> either way, a (block[63]&7) == 4 test before the dc idct seems
> >
> > i meant (block[0]&7) == 4 of course
> 
> But of course that would be MPEG2-specific code inside generic code
> (in the mpegvideo decoder)... why slow down everything else for the
> sake of MPEG-2?

the (block[0]&7) == 4 check could also be put in the mpeg2 code
though that should happen only when a dc idct is actually enabled because
before it, it is just slowing mpeg2 down

also its all a tradeoff, placing the check closer to the idct would allow
us to also support the ==4 case for example and i think the check will
be more expensive in the mpeg2 code (must be run on non dc blocks and 
we dont have an existing branch like for the dc idct where it can be
merged into)
that is after a shallow look, i likely am missing options and details

btw, mpeg4-asp can use the mpeg2 unquantizer too, so its not just mpeg2
that needs to be changed. But its easy to skip (if wanted) as its a seperate
codepath and so the dc_idct() optimization could just be ommited from
it.
above is more based on memory then RTFS so things might not match completely
reality, if something looks wrong tell me and ill investigate it more
later

[...]

-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

I wish the Xiph folks would stop pretending they've got something they
do not.  Somehow I fear this will remain a wish. -- M?ns Rullg?rd
-------------- 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/20100621/47f7dd0b/attachment.pgp>