[MPlayer-dev-eng] [PATCH] mp_image flags in filters

Reimar Döffinger Reimar.Doeffinger at stud.uni-karlsruhe.de
Thu Jul 28 17:24:08 CEST 2005


Hi,
On Thu, Jul 28, 2005 at 10:57:26AM -0400, Rich Felker wrote:
> On Thu, Jul 28, 2005 at 01:56:30AM +0200, Reimar Döffinger wrote:
> > I forgot a few other things I saw that I think are wrong.
> > E.g.
> > vf_eq, vf_eq2 and vf_hue modify the original image and export them.
> 
> They do not modify the original last I checked. The export the
> original plane for unmodified planes, and an internal buffer for
> planes they modify. If they're modifying the original image, it's
> blatently wrong, and someone else must have introduced that
> brokenness..

No, it is okay as far as I can see. Hard to notice such subtleties that
late at night *g*

> > harddup: I reuses the same buffers, so shouldn't it set PRESERVE?
> 
> Huh? harddup never allocates a buffer, it just exports. There's no way
> it can tell the previous filter that the image contents must be
> preserved, BUT the key is that the previous filter will not be called
> before the duplicate control is called by mencoder.

Actually I was thinking about the case where somebody would stick a
filter _after_ it... Sure it doesn't allocate a buffer, but the export
buffer it passes on could have the preserve flag set...

> > qp: Shouldn't/Couldn't it use EXPORT buffers? After all it doesn't
> > change the _data_ at all, only the qscale values.
> 
> Yes. What does it use now?

It uses DR with TEMP buffers + memcpy as fallback.

Greetings,
Reimar Döffinger




More information about the MPlayer-dev-eng mailing list