Change the equation for computing sigma from blur radius

Change the equation for computing sigma from blur radius

In a previous CL: https://codereview.chromium.org/1502283002/, this equation
was changed and now when the blur radius is 2px the blur effect completely
disappears. This CL change the equation back to the original one but add
a special case handler so that when the blur radius is 0 there is no blur
effect at all.

Also, together with this change, a bunch of layout test will need to
rebaseline.

BUG=572822
Committed: https://crrev.com/c2a7942a1d42e2f77b1656f2e02550e4e865943a
Cr-Commit-Position: refs/heads/master@{#367188}

[xidachen, 2015-12-30 16:14:47]

xidachen@chromium.org changed reviewers:
+ junov@chromium.org

[xidachen, 2015-12-30 16:14:48]

PTAL

[xidachen, 2015-12-30 19:16:20]

Something interesting here:
Under SkBlurMask::ConvertRadiusToSigma, the equation is:
sigma = 0.57735f * radius + 0.5f;

But in the SkiaUtils::skBlurRadiusToSigma, the equation is:
sigma = 0.288675f * radius + 0.5f;

I am a bit confused why the scaled factor is divided by 2 here.

[Justin Novosad, 2015-12-30 20:32:53]

lgtm

On 2015/12/30 19:16:20, xidachen wrote:
> Something interesting here:
> Under SkBlurMask::ConvertRadiusToSigma, the equation is:
> sigma = 0.57735f * radius + 0.5f;
> 
> But in the SkiaUtils::skBlurRadiusToSigma, the equation is:
> sigma = 0.288675f * radius + 0.5f;
> 
> I am a bit confused why the scaled factor is divided by 2 here.

Yeah that is a bit weird. That filter is used for shadows too. If it ain't
broken, don't fix it.

[xidachen, 2015-12-30 20:33:43]

The CQ bit was checked by xidachen@chromium.org

[xidachen, 2015-12-30 20:34:41]

On 2015/12/30 20:32:53, Justin Novosad wrote:
> lgtm
> 
> On 2015/12/30 19:16:20, xidachen wrote:
> > Something interesting here:
> > Under SkBlurMask::ConvertRadiusToSigma, the equation is:
> > sigma = 0.57735f * radius + 0.5f;
> > 
> > But in the SkiaUtils::skBlurRadiusToSigma, the equation is:
> > sigma = 0.288675f * radius + 0.5f;
> > 
> > I am a bit confused why the scaled factor is divided by 2 here.
> 
> Yeah that is a bit weird. That filter is used for shadows too. If it ain't
> broken, don't fix it.

I think it is related to this bug:
https://code.google.com/p/chromium/issues/detail?id=179006, which says the blur
covers about half the expected distance.

[commit-bot: I haz the power, 2015-12-30 20:34:41]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1556523002/20001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1556523002/20001

[commit-bot: I haz the power, 2015-12-30 21:52:47]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2015-12-30 21:53:58]

Description was changed from

==========
Change the equation for computing sigma from blur radius

In a previous CL: https://codereview.chromium.org/1502283002/, this equation
was changed and now when the blur radius is 2px the blur effect completely
disappears. This CL change the equation back to the original one but add
a special case handler so that when the blur radius is 0 there is no blur
effect at all.

Also, together with this change, a bunch of layout test will need to
rebaseline.

BUG=572822
==========

to

==========
Change the equation for computing sigma from blur radius

In a previous CL: https://codereview.chromium.org/1502283002/, this equation
was changed and now when the blur radius is 2px the blur effect completely
disappears. This CL change the equation back to the original one but add
a special case handler so that when the blur radius is 0 there is no blur
effect at all.

Also, together with this change, a bunch of layout test will need to
rebaseline.

BUG=572822

Committed: https://crrev.com/c2a7942a1d42e2f77b1656f2e02550e4e865943a
Cr-Commit-Position: refs/heads/master@{#367188}
==========

[commit-bot: I haz the power, 2015-12-30 21:53:58]

Patchset 2 (id:??) landed as
https://crrev.com/c2a7942a1d42e2f77b1656f2e02550e4e865943a
Cr-Commit-Position: refs/heads/master@{#367188}