Image filters: optimize crop rect application

Image filters: optimize crop rect application.

If a filter does not affect transparent black, there's no 
reason to allow the crop rect to expand beyond the optimal 
size determined by onFilterNodeBounds(). So don't enlarge 
the bounds unless the filter affects transparent black.

In order to determine which filters affect transparent
black, I've pulled the inverse of the canComputeFastBounds() 
logic into a non-recursive, affectsTransparentBlack() 
virtual, and left canComputeFastBounds() as a non-virtual,
recursive function that calls it.

BUG=599933
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1848953002
TBR=reed@google.com

Committed: https://skia.googlesource.com/skia/+/6db0a7bdceb6be85721bfb0db8dea7fd27db5970

[Stephen White, 2016-03-31 20:18:04]

Description was changed from

==========
Image filters: optimize crop rect application for the expand case.

If a filter does not affect transparent black, there's no reason to
allow the crop rect to expand beyond the optimal size determined by
onFilterNodeBounds(). So don't enlarge the bounds unless the filter
affects transparent black.

BUG=skia:
==========

to

==========
Image filters: optimize crop rect application for the expand case.

If a filter does not affect transparent black, there's no reason to
allow the crop rect to expand beyond the optimal size determined by
onFilterNodeBounds(). So don't enlarge the bounds unless the filter
affects transparent black.

BUG=skia:
GOLD_TRYBOT_URL=
https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&is...
==========

[Stephen White, 2016-03-31 21:22:39]

The CQ bit was checked by senorblanco@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2016-03-31 21:22:48]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1848953002/80001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1848953002/80001

[commit-bot: I haz the power, 2016-04-01 11:50:38]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-04-01 11:50:39]

Dry run: This issue passed the CQ dry run.

[Stephen White, 2016-04-01 13:53:46]

Description was changed from

==========
Image filters: optimize crop rect application for the expand case.

If a filter does not affect transparent black, there's no reason to
allow the crop rect to expand beyond the optimal size determined by
onFilterNodeBounds(). So don't enlarge the bounds unless the filter
affects transparent black.

BUG=skia:
GOLD_TRYBOT_URL=
https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&is...
==========

to

==========
Image filters: optimize crop rect application.

If a filter does not affect transparent black, there's no reason to
allow the crop rect to expand beyond the optimal size determined by
onFilterNodeBounds(). So don't enlarge the bounds unless the filter
affects transparent black.

BUG=skia:
GOLD_TRYBOT_URL=
https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&is...
==========

[Stephen White, 2016-04-01 18:59:34]

Description was changed from

==========
Image filters: optimize crop rect application.

If a filter does not affect transparent black, there's no reason to
allow the crop rect to expand beyond the optimal size determined by
onFilterNodeBounds(). So don't enlarge the bounds unless the filter
affects transparent black.

BUG=skia:
GOLD_TRYBOT_URL=
https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&is...
==========

to

==========
Image filters: optimize crop rect application.

If a filter does not affect transparent black, there's no reason to
allow the crop rect to expand beyond the optimal size determined by
onFilterNodeBounds(). So don't enlarge the bounds unless the filter
affects transparent black.

BUG=599933
GOLD_TRYBOT_URL=
https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&is...
==========

[Stephen White, 2016-04-01 18:59:34]

senorblanco@chromium.org changed reviewers:
+ robertphillips@google.com

[Stephen White, 2016-04-01 19:02:05]

The CQ bit was checked by senorblanco@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2016-04-01 19:02:07]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1848953002/100001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1848953002/100001

[Stephen White, 2016-04-01 19:04:33]

Description was changed from

==========
Image filters: optimize crop rect application.

If a filter does not affect transparent black, there's no reason to
allow the crop rect to expand beyond the optimal size determined by
onFilterNodeBounds(). So don't enlarge the bounds unless the filter
affects transparent black.

BUG=599933
GOLD_TRYBOT_URL=
https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&is...
==========

to

==========
Image filters: optimize crop rect application.

If a filter does not affect transparent black, there's no 
reason to allow the crop rect to expand beyond the optimal 
size determined by onFilterNodeBounds(). So don't enlarge 
the bounds unless the filter affects transparent black.

In order to determine which filters affect transparent
black, I've pulled the inverse of the canComputeFastBounds() 
logic into a non-recursive, affectsTransparentBlack() 
virtual, and left canComputeFastBounds() as a non-virtual,
recursive function that calls it.

BUG=599933
GOLD_TRYBOT_URL=
https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&is...
==========

[Stephen White, 2016-04-01 19:05:02]

Rob: PTAL. Thanks!

[commit-bot: I haz the power, 2016-04-01 19:23:58]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-04-01 19:23:58]

Dry run: This issue passed the CQ dry run.

[robertphillips, 2016-04-01 19:58:08]

lgtm + dox suggestions & a nit

https://codereview.chromium.org/1848953002/diff/100001/include/core/SkImageFi...
File include/core/SkImageFilter.h (right):

https://codereview.chromium.org/1848953002/diff/100001/include/core/SkImageFi...
include/core/SkImageFilter.h:93: *  edges which reduce the size of the bounds
will be used.
Maybe:
* This is to accommodate filter that affect transparent black. In such cases
(i.e.,
* when embiggen is set) the resulting rect must cover all the pixels

?

https://codereview.chromium.org/1848953002/diff/100001/include/core/SkImageFi...
include/core/SkImageFilter.h:97: */
overlength

https://codereview.chromium.org/1848953002/diff/100001/src/core/SkImageFilter...
File src/core/SkImageFilter.cpp (right):

https://codereview.chromium.org/1848953002/diff/100001/src/core/SkImageFilter...
src/core/SkImageFilter.cpp:72: bool embiggen,
cropped -> dst - as in header ?

https://codereview.chromium.org/1848953002/diff/100001/src/effects/SkMergeIma...
File src/effects/SkMergeImageFilter.cpp (right):

https://codereview.chromium.org/1848953002/diff/100001/src/effects/SkMergeIma...
src/effects/SkMergeImageFilter.cpp:88: // Apply the crop rect to the union of
the inputs' bounds.
// add a comment about why this is 'false' 

?

[Stephen White, 2016-04-01 21:26:35]

https://codereview.chromium.org/1848953002/diff/100001/include/core/SkImageFi...
File include/core/SkImageFilter.h (right):

https://codereview.chromium.org/1848953002/diff/100001/include/core/SkImageFi...
include/core/SkImageFilter.h:93: *  edges which reduce the size of the bounds
will be used.
OK, I tried to clarify it a bit.

https://codereview.chromium.org/1848953002/diff/100001/include/core/SkImageFi...
include/core/SkImageFilter.h:97: */
On 2016/04/01 19:58:08, robertphillips wrote:
> overlength

Fixed.

https://codereview.chromium.org/1848953002/diff/100001/src/core/SkImageFilter...
File src/core/SkImageFilter.cpp (right):

https://codereview.chromium.org/1848953002/diff/100001/src/core/SkImageFilter...
src/core/SkImageFilter.cpp:72: bool embiggen,
On 2016/04/01 19:58:08, robertphillips wrote:
> cropped -> dst - as in header ?

Changed the header back to cropped instead.

https://codereview.chromium.org/1848953002/diff/100001/src/effects/SkMergeIma...
File src/effects/SkMergeImageFilter.cpp (right):

https://codereview.chromium.org/1848953002/diff/100001/src/effects/SkMergeIma...
src/effects/SkMergeImageFilter.cpp:88: // Apply the crop rect to the union of
the inputs' bounds.
Changed it to pass this->affectsTransparentBlack() instead.

[Stephen White, 2016-04-01 21:28:23]

The CQ bit was checked by senorblanco@chromium.org

[Stephen White, 2016-04-01 21:28:24]

The patchset sent to the CQ was uploaded after l-g-t-m from
robertphillips@google.com
Link to the patchset: https://codereview.chromium.org/1848953002/#ps120001
(title: "Fixes per review remarks")

[commit-bot: I haz the power, 2016-04-01 21:28:27]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1848953002/120001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1848953002/120001

[commit-bot: I haz the power, 2016-04-01 21:29:56]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-04-01 21:29:57]

Try jobs failed on following builders:
  skia_presubmit-Trybot on client.skia.fyi (JOB_FAILED,
http://build.chromium.org/p/client.skia.fyi/builders/skia_presubmit-Trybot/bu...)

[Stephen White, 2016-04-01 21:30:49]

Description was changed from

==========
Image filters: optimize crop rect application.

If a filter does not affect transparent black, there's no 
reason to allow the crop rect to expand beyond the optimal 
size determined by onFilterNodeBounds(). So don't enlarge 
the bounds unless the filter affects transparent black.

In order to determine which filters affect transparent
black, I've pulled the inverse of the canComputeFastBounds() 
logic into a non-recursive, affectsTransparentBlack() 
virtual, and left canComputeFastBounds() as a non-virtual,
recursive function that calls it.

BUG=599933
GOLD_TRYBOT_URL=
https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&is...
==========

to

==========
Image filters: optimize crop rect application.

If a filter does not affect transparent black, there's no 
reason to allow the crop rect to expand beyond the optimal 
size determined by onFilterNodeBounds(). So don't enlarge 
the bounds unless the filter affects transparent black.

In order to determine which filters affect transparent
black, I've pulled the inverse of the canComputeFastBounds() 
logic into a non-recursive, affectsTransparentBlack() 
virtual, and left canComputeFastBounds() as a non-virtual,
recursive function that calls it.

BUG=599933
GOLD_TRYBOT_URL=
https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&is...
TBR=reed@google.com
==========

[Stephen White, 2016-04-01 23:40:05]

The CQ bit was checked by senorblanco@chromium.org

[commit-bot: I haz the power, 2016-04-01 23:40:17]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1848953002/120001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1848953002/120001

[commit-bot: I haz the power, 2016-04-01 23:41:16]

Description was changed from

==========
Image filters: optimize crop rect application.

If a filter does not affect transparent black, there's no 
reason to allow the crop rect to expand beyond the optimal 
size determined by onFilterNodeBounds(). So don't enlarge 
the bounds unless the filter affects transparent black.

In order to determine which filters affect transparent
black, I've pulled the inverse of the canComputeFastBounds() 
logic into a non-recursive, affectsTransparentBlack() 
virtual, and left canComputeFastBounds() as a non-virtual,
recursive function that calls it.

BUG=599933
GOLD_TRYBOT_URL=
https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&is...
TBR=reed@google.com
==========

to

==========
Image filters: optimize crop rect application.

If a filter does not affect transparent black, there's no 
reason to allow the crop rect to expand beyond the optimal 
size determined by onFilterNodeBounds(). So don't enlarge 
the bounds unless the filter affects transparent black.

In order to determine which filters affect transparent
black, I've pulled the inverse of the canComputeFastBounds() 
logic into a non-recursive, affectsTransparentBlack() 
virtual, and left canComputeFastBounds() as a non-virtual,
recursive function that calls it.

BUG=599933
GOLD_TRYBOT_URL=
https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&is...
TBR=reed@google.com

Committed:
https://skia.googlesource.com/skia/+/6db0a7bdceb6be85721bfb0db8dea7fd27db5970
==========

[commit-bot: I haz the power, 2016-04-01 23:41:17]

Committed patchset #7 (id:120001) as
https://skia.googlesource.com/skia/+/6db0a7bdceb6be85721bfb0db8dea7fd27db5970