Re[3]: [Gnash-dev] Re: point test

[Udo Giacomozzi]

Hello Bastiaan,

Monday, November 5, 2007, 10:33:55 AM, you wrote:
BJ> ...Anyway, I hope the concept is clear now.

Aha! Yes, it is now, thanks :)

So the normalization is a preprocessing step before going on to
tesselate/triangulate. I guess this is necessary for any hardware
renderer because they normally need simple shapes. The Flash format is
optimized for software rendering, that's why it contains two fill
styles. I would be surprised if any gfx hardware could render such
information without conversion.

Back on topic: The normalization would *not* be necessary for the
method I described because that method is specifically designed for
the two-fill-sides information, ie. it takes advantage of it.

If we go the renderer way we would probably still need the
normalization as it is part of the rendering process. Renderer-based
hit test is basically nothing else than rendering without actually
drawing (simplified, of course). Anyway, this is up to the renderer.
AGG would not need (not want) normalization for the renderer-based hit
test.

Udo