Re: [SVG1.1F2 LC] 15.17 feGaussianBlur (ISSUE-2345, ACTION-2822)

On Tue, 29 Jun 2010 16:00:47 +0200, Dr. Olaf Hoffmann  
<Dr.O.Hoffmann@gmx.de> wrote:

> Hello www-svg,
>
> it is nice to see, that the formula for the y dimension is
> added for the feGaussianBlur.
> Nice as well to get now the input image for standard deviation zero.
>
> Minor cosmetic questions/suggestions:
> a) Why is H(x) styled in black, I(y) in red?

Just an editing mishap, fixed.

> b) Why is the approximation formula only for s available?
> (should be sufficient to write, that the approximation applies
> for the t correspondingly)

I've added an informative note to that effect, see [1].

> c) Maybe worth to mention, that for something like stdDeviation="1 0"
> or stdDeviation="0 1" with '1' representing something different from
> 0, the mathematically expected result is not the input image.
> It is only a blur in one direction.

Fair enough, I've added explicit mention of this case, see [1].

> However, because the definition does not explictly mention
> this case, implementations will hopefully do it as mathematically
> expected.

Yes, one would hope so :)

I've added two testcases for the changes in question:

   http://dev.w3.org/SVG/profiles/1.1F2/test/svg/filters-gauss-02-f.svg
   http://dev.w3.org/SVG/profiles/1.1F2/test/svg/filters-gauss-03-f.svg

Please let us know if these changes satisfies your comment.
/Erik

[1]  
http://dev.w3.org/SVG/profiles/1.1F2/publish/filters.html#feGaussianBlurElement

-- 
Erik Dahlstrom, Core Technology Developer, Opera Software
Co-Chair, W3C SVG Working Group
Personal blog: http://my.opera.com/macdev_ed

Received on Tuesday, 13 July 2010 14:50:19 UTC