[cairo] anti-aliasing quality in glitz/OpenGL vs. Cairo native rasterization?

[Keith Packard]

Around 0 o'clock on Jun 11, Leon Woestenberg wrote:

> Is my assumption true that the typical OpenGL backend output is worse
> than native rasterization in the Cairo software quality-wise?

I'd rather use the term "less accurate" than "worse"; it's a time/
appearance tradeoff.

When the existing cairo software rasterizer is reimplemented (it's
currently irreparibly broken), I'd like to see the algorithm parameterized
so the number of discrete alpha values generated can vary (probably just as
a function of alpha depth ranging from 1 to 8).

Given the proposed new trapezoid rasterizer spec, this should be
straightforward.

> I may start a thesis project exploiting offloading parts of a rasterizer
> into logic (as GPL VHDL code) and Cairo may be an interesting candidate
> there.

It's the Render specification that's central here; cairo uses an identical 
spec for it's own rasterization.  But, that's also got the old broken
design in it still, so we'll need to rewrite that too...

It's all just point sampled alpha values (like super sampling); the only 
question is where you place the points.

-keith


-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 228 bytes
Desc: not available
Url : http://lists.freedesktop.org/archives/cairo/attachments/20040613/855d6997/attachment.pgp