Anim8or Community
General Category => General Anim8or Forum => Topic started by: Old Codger on September 02, 2021, 02:05:48 pm

Can I ask why 32 is the max number of segments Anim8or will break anything into? Whether it be a cylinder, sphere or a "lathe" operation I have determined empirically that 32 is the max number. I suspect there is a deep reason why that is the limit. I'm simply curious as to the "why". There have been times when it would have been really handy to have a cylinder or sphere with more than 32 longitudinal elements.
Please note that I'm not complaining, merely curious. Unfortunately when the "question" flag in my skull asserts the only way to clear it is to ask the question. In a classroom the flag has a prettynear direct link to my right hand. "Hand up" state follows "question flag" assertion within a few milliseconds. It's one of my many faults. The instructor in the Pascal class I took darned near kicked me out of class for asking questions several minutes before she got to the explanation in her lecture. I tried to explain the link between the flag in my head and my right arm but she had never done either assembly language or C programming so had no idea what I was talking about with the term "flag". Come to think of it she wasn't all that good with Pascal either.

This isn't an answer to why lathing is limited to 32 steps, but you can create a cylinder with up to 100 sides by extruding a 100sided Ngon.

This isn't an answer to why lathing is limited to 32 steps, but you can create a cylinder with up to 100 sides by extruding a 100sided Ngon.
Interesting. Then I could  conceivably  spin (I think of lathing as "spinning" because that's what, in effect I'm doing) a cross section of that shape to produce a sphere with 25 latitude lines. In profile it might make for a more rounded sphere for backgrounds. As it is 32 segments yields 8 increments per quadrant which can sometimes show up on a close zoom.
Please note, Steven (sorry for the repeated misspellings previously) that I am not complaining. The fact that the rotations available in a lathe operation is a power of 2 tells me there might be some fundamental limitation at play here.

Yes, you can certainly lathe a 100sided NGon into a sphere shape, though you will end up with a doubling of the mesh since the object would actually be a torus with a diameter of zero. A simpler way is to start with the basic sphere primitive and subdivide it, either by converting to a subdivision object (Build > Convert to Subdivided) or converting it to a mesh (Build > Convert to Mesh) and using the Build > Smooth Object command with a tension of 0. In the latter case you can smooth as many times as you like, doubling the number of lat/long lines each time, and the result is an exportable mesh.
Subdivision objects use an algorithm to determine subdivisions between the basic mesh so an exported version of the object may only include the original lowres mesh, but this can be resolved using Build > Convert to Mesh. Subdivision objects are excellent for 'organic' models that will be morphanimated, since you only need to animate the coarse mesh while allowing the subdivision algorithm to smooth everything out.
One thing to note: due to its topology the sphere primitive may look a bit pinched at the 'poles' when subdivided. Subdividing the dodecahedron or icosahedron primitive results in something more like a geodesic object without polar distortions and approaches a spherical shape with a couple of subdivision stages.
EDIT: I momentarily forgot that the sphere object can be converted from lon/lat topology to geodesic by doubleclicking to open the object dialog, and the geodesic divisions can be defined (max 6). While not strictly a true geodesic it may subdivide more smoothly than the lon/lat topology.

It's an arbitrary limit, I agree, but you seldom really need that many faces. And allowing arbitrary limits can use an enormous amount of memory when building and rendering the model, etc.