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

On Tue, Jun 22, 2010 at 1:55 PM, Brad Kemper <brad.kemper@gmail.com> wrote:
> On Jun 22, 2010, at 11:27 AM, fantasai <fantasai.lists@inkedblade.net>
> wrote:
>
>> On 06/21/2010 10:33 PM, Brad Kemper wrote:
>>>
>>> On Jun 21, 2010, at 10:23 PM, Robert O'Callahan wrote:
>>>
>>>> I'm also in favour of that.
>>>>
>>>> Apart from the other points made, I have another ... when there is no
>>>> shadow offset, the blurred area inside the shadow edge is not visible.
>>>> I expect authors will be surprised to find that the width of the
>>>> visible blur is only half the value they specified.
>>>
>>> Seriously? You want to optimize for those uthors that use shadows, but
>>> are surprised by what they see when they fill in those first 2 values
>>> of box-shadow with something other than zero? Maybe they would also be
>>> surprised by the fact that it is called 'box-shadow' and not 'box-glow'.
>>
>> Whatever it's called, it will be used for both shadow and glow effects,
>> and other effects we have not yet thought of. I see box-shadow less and
>> less as a shadow effect and more like a swiss-army-knife of graphical
>> border-edge tweaking.
>
> The point is, it it not reasonable to ignore the biggest existing usage of
> people that do use offsets and do perhaps want to actually see some of their
> shadow color at the opacity they specified  But we should pretend instead
> that the blur value does not have a measurable visible effect and is mainly
> just for growing the shadow?  That is not reasonable. The main purpose of
> spread is to grow the shadow and the main purpose of blur is to blur the
> shadow. Spread frees designers from having to hack blur for it's secondary
> effect, so they can grow the shadow without having to make it blurrier.
>
>>

Let's not lose sight of the fact that none of us are advocating a
different rendering approach or effect. Fundamentally, we're just
discussing whether a parameter should be specified with a number or
twice that same number.

Of course author intuition and expectations should be anticipated to
the best of our ability. But, *regardless* of which format is chosen
here, any confusion is going to last for about 2 seconds (cleared up
with some simple experimentation or a diagram in a blog).

It seems that, if
1) people are using existing, vendor prefixed box-shadow
implementations with no problem (though the effect is inconsistent
across browsers)
2) shadows are blurred with a radius-type parameter elsewhere (e.g.
canvas and Photoshop)
3) At least two browser implementations already exist, with at least
the intent to be parameterized with a radius value

Regardless of a hypothetical best option (which I don't think we'll
find through this discussion), clearly authors are creating shadows as
intended and with results satisfactory to them. Since that is the
case, I think it makes perfect sense to allow the momentum brought
about by (3) to define the way forward, as others have already said.

Again, this may or may not be the "best" choice, but we're in full
bike shed mode here.

Received on Tuesday, 22 June 2010 19:42:00 UTC