Nice work!!! I like how the cone is able to handle varying base and top sizes which is far more useful than the existing add cone operator.

On the cylinder it says "vertices" which is a little bit misleading considering there are twice as many vertices actually produced. Also in practice a cylinder often needs the long strips of faces to be divided with edge loops in order to be useful for subdiv and displacement. Maybe the cylinder primitive could be made similarly to the UV sphere and have "segments" and "rings" as well?

In D10715#271375, @Kenzie (kenziemac130) wrote:

Nice work!!! I like how the cone is able to handle varying base and top sizes which is far more useful than the existing add cone operator.

On the cylinder it says "vertices" which is a little bit misleading considering there are twice as many vertices actually produced. Also in practice a cylinder often needs the long strips of faces to be divided with edge loops in order to be useful for subdiv and displacement. Maybe the cylinder primitive could be made similarly to the UV sphere and have "segments" and "rings" as well?

I believe on a cylinder verts should be called meridians and horizontal cuts should be parallels

In D10715#271375, @Kenzie (kenziemac130) wrote:

Nice work!!! I like how the cone is able to handle varying base and top sizes which is far more useful than the existing add cone operator.

On the cylinder it says "vertices" which is a little bit misleading considering there are twice as many vertices actually produced. Also in practice a cylinder often needs the long strips of faces to be divided with edge loops in order to be useful for subdiv and displacement. Maybe the cylinder primitive could be made similarly to the UV sphere and have "segments" and "rings" as well?

That gives me an idea-- the topology of the UV sphere and a cylinder / cone with rings would be very similar. I wonder if I could share the topology generation between the UV sphere, the cone, and the cylinder. Interesting! I'll have to think about that when I'm more awake.

Just FYI this was brought to my attention as something related to that: D10093

On the UV Sphere, when Rings or Segments are less than 3 it returns no Geometry. I suppose this is partly a UI issue but would it be better to clamp using min(x, 3)?

I'm not done with the review yet.

I think returning no geometry in the case when the number of vertices is too low is better. One can always add a clamp node before, but the opposite would not be easily possible.

Great, now, at least we can generate exact amount of points!
I hope indexes will be exposed as attribute.

Already tested.
IMO primitives and mesh operations (like subdivide etc.) should be separated. Now they all mixed and it is confusing a bit, if you are new user and use menu.

In D10715#272552, @Vyacheslav (hitrpr) wrote:

Great, now, at least we can generate exact amount of points!
I hope indexes will be exposed as attribute.

Already tested.
IMO primitives and mesh operations (like subdivide etc.) should be separated. Now they all mixed and it is confusing a bit, if you are new user and use menu.

IMO primitives and mesh operations (like subdivide etc.) should be separated.

Yes, definitely, like the comment in nodeitems_builtins.py says, we'll need to do some refactoring of the add menu before adding these to a sub-menu. But I hope to do that for 2.93.