Revised sub-rectangle scaling for use in Chromoting.

Linear sub-rectangle scaler for use in Chromoting.

This implementation re-uses the common row filter procedures, but is currently limited to a C horizontal interpolation procedure.

There's also plenty of scope for optimizing the new sub-rectangle scaler routine.

BUG=93451
TEST=media_unittests, remoting_unittests and manual verification of image quality of Chromoting sessions using fit-to-screen.


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=117748

[Wez, 2011-12-15 01:32:50]

scherkus: PTAL the changes under media/

This CL is based on Alpha's CL 7744002, and provides a way to convert and scale
only a sub-rectangle of a frame.

I've replaced his LinearScaleYUVToRGB32RowWithRange by re-using the
LinearScaleYUVToRGB32Row code with a slight tweak, making the latter a special
case of the former.  Ideally the linear-scale MMX and SSE implementations would
be updated to the more general form, so that the same functions could be used
for both full-frame and sub-rectangle scaling.

The ScaleYUVToRGB32WithRect method could be similarly replaced with a
generalized implementation of ScaleYUVToRGB32, but I've stopped short of that in
this CL. :)

[Wez, 2011-12-15 21:34:07]

fbarchard: FYI, this is our draft CL for sub-rectangle scaling for Chromoting.

hbono: PTAL at the media/base/ changes.  I'd also like to get the MMX & SSE code
updated so that LinearScaleYUVToRGB32WithRange() can be implemented for all
three, and then LinearScaleYUVToRGB32() is just a special-case call to that. 
I've not yet got that working, though. :(

[scherkus (not reviewing), 2011-12-15 21:52:34]

http://codereview.chromium.org/8954003/diff/1012/media/base/yuv_convert.h
File media/base/yuv_convert.h (right):

http://codereview.chromium.org/8954003/diff/1012/media/base/yuv_convert.h#new...
media/base/yuv_convert.h:73: void ScaleYUVToRGB32WithRect(const uint8* yplane,
thinking out loud... I wonder if it makes sense to go with gfx::Size / gfx::Rect
at some point

maybe one day :)

[Wez, 2011-12-19 20:56:41]

hclam: PTAL!

[Alpha Left Google, 2011-12-19 21:00:18]

LGTM. Thanks for picking this up!

[Hironori Bono, 2011-12-22 06:15:23]

LGTM2. (I'm not an owner, though.)
Thanks for your change.

Regards,

Hironori Bono

[Wez, 2012-01-12 02:29:48]

scherkus: Could you take a look at this CL & make sure you're happy with it;
it's based on Alpha's earlier implementation and has his blessing. ;)

http://codereview.chromium.org/8954003/diff/1012/media/base/yuv_convert.h
File media/base/yuv_convert.h (right):

http://codereview.chromium.org/8954003/diff/1012/media/base/yuv_convert.h#new...
media/base/yuv_convert.h:73: void ScaleYUVToRGB32WithRect(const uint8* yplane,
On 2011/12/15 21:52:35, scherkus wrote:
> thinking out loud... I wonder if it makes sense to go with gfx::Size /
gfx::Rect
> at some point
> 
> maybe one day :)

That would be awesome, but:
a) are those stable enough; other code seems to use skia equivalents?
b) I thought media/ specifically avoided that dependency, although I notice some
media/ code now uses gfx::Rect, so perhaps that's no longer true?

[scherkus (not reviewing), 2012-01-12 02:40:31]

LGTM

http://codereview.chromium.org/8954003/diff/1012/media/base/yuv_convert.h
File media/base/yuv_convert.h (right):

http://codereview.chromium.org/8954003/diff/1012/media/base/yuv_convert.h#new...
media/base/yuv_convert.h:73: void ScaleYUVToRGB32WithRect(const uint8* yplane,
On 2012/01/12 02:29:48, Wez wrote:
> On 2011/12/15 21:52:35, scherkus wrote:
> > thinking out loud... I wonder if it makes sense to go with gfx::Size /
> gfx::Rect
> > at some point
> > 
> > maybe one day :)
> 
> That would be awesome, but:
> a) are those stable enough; other code seems to use skia equivalents?
> b) I thought media/ specifically avoided that dependency, although I notice
some
> media/ code now uses gfx::Rect, so perhaps that's no longer true?

we have a bit of gfx::Size/Rect usage but no skia usage yet

I'd need to check with some graphics-y people to see if perhaps we should be
using skia primitives instead

[commit-bot: I haz the power, 2012-01-14 00:10:31]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/wez@chromium.org/8954003/22001

[commit-bot: I haz the power, 2012-01-14 01:21:20]

Change committed as 117748