PDA

View Full Version : Large number of polygons or less polygons with subpatch?



Sebasvideo
05-15-2014, 08:02 AM
I was wondering if for example you need to make a lake that is rather large, and you want to make the little waves typical of a lake, which option is less taxing on the system, especially in RAM, to make the plane for the lake with a large number of polygons, or to make it with a small number and then set it to subpatch with a 5 to 8 level.

Lake aside, what is the standard notion, more polygons and no subpatch, or less polygons and higher subpatch level?

Thanks

Sensei
05-15-2014, 09:13 AM
Sub-patches are frozen to regular triangles at the beginning of rendering every frame.

From renderer point of view there is no difference between 1000 sub-patches and 18,000 triangles (3x3 sub-patches = 9 quads = 18 triangles/sub-patch).

jeric_synergy
05-15-2014, 09:20 AM
hmmmm.... isn't there some sort of Adaptive Subdivision available?

JoePoe
05-15-2014, 09:23 AM
In terms of working with it..... (especially in Modeler with it's anemic one core capabilities) I'd much rather throw around a 10K poly subD object than the equivalent 90K (at level 3) faces. CCs are much "worse" (or better on your point of view :)). At level 3 10K gets you 640+K, at 2.... 160K. In Layout, where the object comes in subdivided it's a little more even, but with SubD you also have the added benefit of being able to lower the display level while working easing the burden on the system. IMHO.

At render though aren't you rendering twice the polys if it's SubD????? (vs an in-Modeler frozen version)

Same example. A 10 K SubD object frozen in Modeler is 90K. It renders at 90K. But a 10K subD object rendered as SubD renders 180K polys because it's been frozen AND tripled. If you triple your faces in Modeler it's a wash at render time.
______

In terms of your lake:
Don't you have to go SubD in order to get a nice deformation?

Sensei
05-15-2014, 09:44 AM
APS is frozen at beginning of rendering frame.

jeric_synergy
05-15-2014, 11:17 AM
I'm not familiar w/it, but it results in varying # of polys per frame, right? So that the animation overall might be rendered w/fewer polys, and hence faster?