Re: [GD-DEVEL] An algorithm for rendering antialiased polygons

> I fully agree, I made similar experiments using the edges
> supersampling (2-32x), some results are visible here:
> 
> http://pierre.libgd.org/polygons/
> 
> I did not commit the code yet as I like to clean it up before (I
> focused on the results more than on the code readability :). However,
> I really like your implementation. I think it makes sense to use it
> instead of mine. The license being BSD, I don't see any problem to
> merge it into libgd.
> 
> I have a couple of questions (warning: I did not try intensively yet):
> 
> - does it support polygon with holes? Like the first two examples in
> my link (it is a simple rectangle, but it works with custom polygon as
> well)

It supports self-intersection and polygons with holes. Also two fill rules to determine
the "insideness" are supported (even-odd and non-zero winding). Basically it supports
everything SVG, PDF or PS rasterization would require.

> - does it depend on agg for the rendering or do you use agg only for the demo?

There is no dependency to AGG, it's just used in the performance comparisons.

> - Are you interested to port it to C and/or to gd?

I need to think about it. :) Separating the algorithm from the test framework and
converting to stand-alone C library should be fairly straightforward, as C++ features
are used very sparingly. However, I don't have much spare time for doing this, so
I can't really commit to any tight schedule...

> As shown in your demo and in my 3rd example, the gd arc and ellipse
> functions (as well as the curves functions) can and will use it
> (except for filled ellipse or circle, it may be much faster to use
> another implementation).
> 
> Cheers,
> --Pierre

Kiia