Around 9 o'clock on Dec 19, "Ramshaw, Lyle" wrote: > So there is a small "reversed" triangle at the top. When > processing the remaining trapezoids in a tiling, we will > never know about both of the edges t.left and t.right > simultaneously; so we will never be able to correct for > this "reversed" triangle.
I don't think we need to correct for this; the tiling of this pixel "overfills" this alpha value. As the alpha channel saturates to 1, this won't cause any significant visible artifacts. The goal should be that a perfect tiling sum to one; imperfections in the tiling should show through to imperfections in the output; in this case a tiling including the triangle and the left and right trapezoid may have an alpha value one greater than it "should". Our area method subtracted out the upper triangle because it couldn't compute the lower triangle area exactly without using 192 bit arithmetic; we justified that hack using your tiling argument, but I think all acknowledged that it was just a hack. Keith Packard XFree86 Core Team HP Cambridge Research Lab _______________________________________________ Render mailing list [EMAIL PROTECTED] http://XFree86.Org/mailman/listinfo/render
