I was wondering what is the difference between rendering a subdivided surface through a Subdivision SOP versus doing it at OBJ level.

I get great results using the Subd SOP, but when I try Subdivision at Object level, it gives me MUCH different results, such as streaking and distortion.

What are advantages and disadvantages to these approaches?

Subdivision surfaces should be superiour at the object level because then you can just send the low poly cage to the renderer. At render time, mantra will adaptively subdivide just the geometry that you see so it's a lot more efficient. Off the top of my head, it's not working as well for you because the quality is set too low? Just some other factors might account for differences, are the problem areas at uv seams? Are the problem areas where you have non-quads?

I managed to fix the streaking seen when Subdividing at Object level by tweaking the uv's.

I find it strange, however, that “Polygons as Subdivision Surfaces” and the Subdivision SOP yield completely different results using the exact, same UV map.

Do they use different algorithms for solving UVs?

Yes, they do use different algorithms for dealing with vertex uvs.

Specifically, “Smooth Vertex UVs” isn't used in Mantra. This should only be a difference with coarse levels of subdivision. You should get more similar behaviour if you turn off the Smooth Vertex UVs option.

Note also that creasing is not supported inside Mantra.

- Jeff Lait

Thanks Jeff!

Subdivision SOP and “Polygons as Subdivision Surfaces” DO have similar behaviour when Smooth Vertex UVs option is off.

I was wondering if you could explain what the difference is between a “rough” uv and a “smooth” UV. This would help me better understand the behaviour of the Smooth SOP when I'm smoothing texture UVs.

Thanks again!
Frankie

Yeah that's the bad thing about SubD rendering, you don't get the nice vertex smoothing. I made a post about it… http://www.sidefx.com/forum/viewtopic.php?t=2849&highlight= [sidefx.com]