In an earlier blog entry I mentioned how I was making a medal like object in software using two different sorts of geometry. I faffed around with a complex one, then settled on a simple one, but that simple one had consequencies. Consequencies I could have forseen if had more than three neurons in my brain.
The work was for PhotoToMesh V5, released last week...
...It creates bas relief files for 3D printers. There is a smoothing setting in the program, explained here, which makes the bas reliefs which the program creates less ragged.
When I finally settled on the "correct" medal geometry I'd forgotten about the smoothing. It works by taking the topologically (not neccessarily geometrically) nearest 3d point to the 3d point under consideration, and averages them.
With rectangular meshes topologically and geometrically are practically the same thing. With the medal mesh geometry which I'd decided on, they're not. This is because there are fewer distinct points at the edge of the mesh and many more near the center:
So when I smooth the bas reliefs I get a "pleasant"(?) concentric rings effect, noticeable at the edges of the medal:
You can get rid of the effect by having many more samples, but the resulting large file size can cause some 3d printers to choke.
Here's another example showing the same effect:
Again increasing the density removes the strange smoothing effect.
By the way...I'm waiting for some of the users of V5 to send me some of their 3d printed creations. I'd love to see something like this:
The partial cylinder and the hemisphere were specifically requested by users...