Hello again. > It would be great if anyone has any better ideas for how to go about this. > To me it seems like the problem is that in the webrev I chop up the curve to > be > interpolated at equal intervals of the parameter.
I implemented another version that detects where either dx/dt or dy/dt is 0, and splits the curve there. This works pretty well for all the curves I've tested (except ones containing "cusps", or something close to a cusp, like p.moveTo(0,0); p.curveTo(100,100,0,100,100,0);). Best of all, this: > For example, the curve p.moveTo(0,0),p.curveTo(84.0, 62.0, 32.0, 34.0, 28.0, > 5.0) is no longer a problem. There is another problem with this method (other than the cusps, which can probably be handled easily as a special case): it is axis-dependent. Ideally, it shouldn't be because the optimal subdivisions of a curve are at values of t that do not change under rotations and translations, but with this method, the t values at which I split the curve do change. A better way would take into account curve flatness instead of changes in x or y direction. Thanks, Denis.
