Hi Michael, I noticed some artifacts on surfaces in some cases, but I think there is a bugs in the code. Yesterday it was worse than now. :) I've already found and eliminated several bugs, but I'm sure that some bugs I missed.
I implemented this algorithm:
http://www.cise.ufl.edu/research/SurfLab/papers/10fan_sbs.pdf
I think it really can produce perfect surfaces

Yes, I thought about non-uniform refinement, and used it in previous version of this script. But I wasn't satisfied with results. So I will play for a while with this algorithm, but if I'll still have problems with it - I'll bring non-uniform surfaces back.
Here is example of non-uniform surfaces:
http://moi3d.com/forum/index.php?webtag=MOI&msg=7038.12