[fxtf-drafts] Clarify the size of the "input image"

mstange has just created a new issue for https://github.com/w3c/fxtf-drafts:

== Clarify the size of the "input image" ==
The definition of the `edgeMode` property refers to the "edges" of the input image, but I'm not sure if the spec sufficiently describes where those edges are, or, more specifically what the size of the input image is. Or even more specifically, what rectangle the input image occupies in filter space, and whether (and in what way) that rectangle is influenced by the element that the filter gets applied to, the filter region, and the primitive subregions.

Take this testcase:

<svg xmlns="http://www.w3.org/2000/svg" width="500" height="500">
  <filter id="f1" filterUnits="userSpaceOnUse" x="0" y="0" width="100" height="100">
    <feGaussianBlur stdDeviation="10" edgeMode="duplicate"/>
  <g filter="url(#f1)">
    <rect x="0" y="0" width="20" height="20"/>

Does the input image cover (0, 0, 20, 20), or does it cover the full filter region at (0, 0, 100, 100) with transparent padding to the right and the bottom? There's a test in Gecko that assumes the latter, for the purposes of `feConvolveMatrix` edge sampling: https://searchfox.org/mozilla-central/source/layout/reftests/svg/filters/feConvolveMatrix-2.svg
In that testcase, if it was the former, there would be no green lines.

Please view or discuss this issue at https://github.com/w3c/fxtf-drafts/issues/324 using your GitHub account

Received on Tuesday, 20 November 2018 00:42:30 UTC