Hi PaQ - definitely a quad topology is more suitable for catmull-clark subdivision like you are showing there.

Displacement-based modeling like ZBrush is as far as I know a completely separate mechanism from that kind of subdivision.

It is true that a displacement based approach wants to subdivide polygons to provide smaller ones to displace, but it is a kind of subdivision that is also oriented towards breaking polygons into smaller pieces, not on following catmull-clark subdivision smoothing rules. (there is a subdivide operation in ZBrush that allows you to do catmull-clark as an option, that is different than the kind of dynamic decimation that will happen for displacement).

As far as I can see ZBrush decimates things into triangles very quickly even if you start with quads.

But it definitely helps if your polygons are of all regular even size.

The example that you show there is all about doing catmull-clark subdivision, that is a totally different kind of subdivision than just direct polygon decimation, as far as I know anyway.

I guess this issue can get confused pretty quickly due to the use of the word "subdivision" in different contexts. It has been a normal thing since the early forms of displacement mapping that the kind of subdivision that happens for displacement is not the same kind of thing as catmull-clark subdivision.


I've been noticing a lot recently that people are tending to ask for "all quads" for many types of situations where the actual need is "evenly sized polygons".


Here's an example - take a large box in MoI and export it to OBJ with plain settings so that it generates an OBJ file with 6 quad faces. Then try to paint on this in ZBrush - it does not work well even though you have "all quad" polygons. For ZBrush's particular use of polygons it is not "quadness" that is so important but "even-ness".

I do not spend a lot of time with ZBrush so it is possible that I don't have the details right, I am just basing the above on what I have observed with some messing around and with what is the typical and normal approach for displacement type subdivision, which is that it is different than Catmull-Clark...

- Michael