Re: [css3-background] vastly different takes on "blur"

On Tue, Jun 22, 2010 at 8:11 PM, Tab Atkins Jr. <jackalmage@gmail.com> wrote:
> On Tue, Jun 22, 2010 at 3:39 PM, Simon Fraser <smfr@me.com> wrote:
>> On Jun 22, 2010, at 3:20 PM, Tab Atkins Jr. wrote:
>>> If we don't want to change text-shadow (we don't), then we should go
>>> ahead and make the blur length specify the full size of the blur
>>> region, like Brad suggests.  While I do prefer the other way, I value
>>> consistency between nearly-identical properties more.
>>
>> That's not what I see. Don't forget that shadows larger than 8px are
>> buggy in WebKit. Here's a test with an 8px shadow:
>> <http://smfr.org/misc/shadow.html>
>>
>> and here's what it looks like in WebKit:
>> <http://smfr.org/misc/shadow-webkit.png>
>>
>> The box and text shadows are identical, and Pixie shows that the shadow extends out by 8px (for a total shadow transition of 16px).
>
> Hmm, my own somewhat unscientific test of taking a screenshot and
> checking pixel colors in GIMP doesn't agree.  I get 4px of blur
> extending out from the character.  (On one side there's a 5th layer of
> pixels that are #fefefe, but I think we can chalk that up to vagaries
> of gaussian blurs.)  Similarly, I have 4px of gradually solidifying
> color extending inwards (again, with one extra layer of #010101 on
> some sides, but again, that's okay).
>
> Perhaps platform differences are causing this?  I'm using Chrome5 and
> FF3.6 on Linux.

There must be. In Windows 7, both Firefox 3.6.4 and Safari 5 are
giving ~8px of blur from the original, non-blurred shadow edge out to
pure white. Chrome 6 and a Chromium nightly are giving half that,
however.

Firefox also gives me 91px in Simon's example set to a much larger
font size and a 100px blur radius. A Photoshop drop shadow effect on
the same shape, with a shadow "Size" of 100px, gives 94px from the
original edge out to pure white.

Given text-shadow's different implementations and it's somewhat
ambiguous spec ("The blur radius is a length value that indicates the
boundaries of the blur effect."), it doesn't seem like it should be
the canonical example.

Received on Wednesday, 23 June 2010 03:07:31 UTC