Open ehkropf opened 9 years ago
What complicates it is that you have to break the boundary at each circle and then hop from circle to circle, avoiding all crossings. Even so I wonder whether the seams of the process might be visible when they don't align with coordinate axes.
What is the objection to poking holes?
I just had an idea for an algorithm which would cut the domain in the right way. It needs more careful thought, but I think it might work.
Let's suppose one of us got a "cutting" algorithm to work. The seams wouldn't be a problem, because we could just draw the seams in the same color as the fill, making the line wide enough to cover what the fill doesn't. Then draw the outer boundaries on top of the fill.
Poking holes would work fine if it was an actual hole to the axis underneath, so to speak. If someone wants a grid under the circle domain plot, the holes might have the axis color, but you wouldn't see the grid through the holes, since they're really just painted on top of the domain fill.
Though having said that, maybe there could be some kind of alpha channel mask to get actual poked holes.
In the case of a multiply connected domain, the circle region will on occasion "poke holes" in a filled patch by painting background colour on top. Is there a way to create a multiply connected patch and just draw that?