On Jun 14, 2010, at 5:14 PM, "Robert O'Callahan" <robert@ocallahan.org> wrote: > On Tue, Jun 15, 2010 at 10:55 AM, Brad Kemper > <brad.kemper@gmail.com> wrote: > Gecko gives me a blur width that is about twice as wide as the > authored amount, so I still want to change that to be a more > straightforward 1:1 relationship. > > Yeah. Right now we make a blurred region whose width is twice the > "blur radius" and which is centered on the shadow edge. Cutting that > in half to match the spec could be done, but it seems a less > intuitive interpretation of "radius". Our implementation has the > property that the pixels affected by the blur are those whose > distance from the shadow edge is less than the "blur radius". Am I > alone in thinking that makes sense? It makes some sense if the width of the blurred region has to be called "blur radius" and then you wotlrk backwards from there to come up with a meaning to match the words. But if you instead work forwards from what would be the most intuitive to author, that is specifying the width of the part that is visibly different, then you can free yourself from the word "radius", and save that work for things that are actually based on a more obvious circle or ellipse or arc with that measurement as it's radius.
Received on Tuesday, 15 June 2010 02:34:23 UTC