Question: Polygon w/Smooth Proxy vs SubD

[ThirdEye]

subds are subds, they're not polygon based i think :shrug:

[Soeren Nielsen]

Some mapping methods are not available with subd's and it gets very heavy on dense objects.
As to what is really the best thing to use, I have yet to find out (pretty new in Maya)

[Modulok]

True SubDivision surfaces are just that: surfaces. They behave like NURBS in a rendering sense, but also share polygonal characteristics with regard to how they are modeled. In short, they are a third type of modeling paradigm, just as Polygons and NURBS are two other types of 3d modeling. In a way, they draw from the 'best of both worlds'. Still, that doesn't mean they're the best for every task. For example, modeling a submarine torpedo could probably be accomplished best with NURBS. On the other hand, a human face is best done with SubDs.

Because SubD's are curves (surfaces) and thus render like NURBS, when output to a RenderMan renderer edges are always perfectly smooth regardless of resolution or how close the camera gets. Unlike NURBS, however, SubDs can be modeled more like polygons.

Now, even Maya's built-in renderer doesn't render NURBS or SubD's as curves... Maya's renderer first converts the curves to polygons and then renders the resulting polygonal mesh. There are a few advantages with this method compared to having modeled with polygons in the first place, because you can set desired tessellation so that if you know the camera will get close or if you're rendering to high resolution, you can up the tessellation to generate a visibly smooth curve---but doing so can drastically slow render times. This is where the advantage of a RenderMan renderer which receives actual curve mathematics instead of tessellated polygons really shines.

As for SoftImage XSI, the only other application I know of that generates true SubDs, I am unsure whether or not MentalRay will receive SubD surface data as curves or pre-tessellated polys. In any event, there is currently no manner I am aware of in which XSI can render to RenderMan because SoftMan RIB translator works only with the old SoftImage 3D program.

The trick, of course, is ensuring a true SubD modeler and ability to output true curves to a renderer that will accept them. A second-best solution is a true-subD modeler with tessellated rendering. Generally the least useful is sub-divided faces.

Why is this all important? It's important foremost to anyone needing to render visibly smooth edges at film resolutions and for camera zooms. It is also useful to anyone generating still images where they can essentially benefit from what appear to be ultra-high "poly" images except with render times and memory usage similar to a low-poly model!



Source: Zaon.org, Justin Winters

Hope this clears some things up
-Modulok-

[ThirdEye]

quote:

"NURBS are rendered equally fast as SDS in PRMan. So are polygons. Everything gets converted to micrpolygons after gridding and before gridding everything has become bicubic or bilinar subpatches already.
The disadvance is that you can only approaximate curved shapes with polygons; and you'll need a lot more of them in such a case than with high-level primitives. Furthermore there will always be a pov imaginable where still the image will exhibit the faceted nature of the geometry depictured in it.
The rule of thumb is to avoid polygons at all costs, if using a RMan compliant renderer."

Source: http://www.cgtalk.com/showthread.php?s=&postid=422035#post422035

[playmesumch00ns]

SDS are rendered much faster than an equivalently-detailed polygonal surface (in PRMan at least) in Renderman. I was recently test-rendering for a project and had the 12,000-face mesh set to 3 subdivisions with MTOR SubDivs. Virtually no speed-hit at all! Doing the subdivisions with Maya's polysmooth however...

The important thing for Renderman is that while the rendertime may be roughly equivalent, the RIB-generation time for a SubDiv or NURBS, as opposed to an equivalent poly object, is minute.

[tomb]

There is a subtle difference between Polysmooth and Sub-Ds in Maya in terms of modeling.

With polys you can create a cage with a smooth proxy inside it.

With Sub-Ds you can do the same thing AND you can edit the smooth proxy "sub-dividing" the mesh to add in details while keeping the cage light.

A light cage has its advantages in animation and editing your model. You can also use this kind of setup for your characters:
http://www.androidblues.com/rigtut.html

[MasonDoran]

tomb, question about Steves Tut. how does one edit morph targets of the wrap deformer once it is bound to the skin? What i mean is...if i push vets in the skinned wrap deformer....things are bit un intuitive because the axis goes all funky due to the bones....and when i edit the actual blendshapes....its also unintuitive due to not knowing which vert to move to get the smooth deformations....

do i make sense?

[FordPrefect]

Thanks everybody. I understand now. Now its time to go try these SubD's out . . .

[tomb]

2byts: I'm not sure I understand. I think you are meant to edit the blend shapes in bind pose? I haven't tried this setup (I don't have Sub-Ds in my Maya :( ) so I can't comment. Try contacting Steven directly, he might be able to help you with this.

[MasonDoran]

yeah...its basically how to edit morph targets while in the bind pose.......and nothing to do with the sub ds itself....they are hidden away on another layer anyways and all edits are done with the wrap deformer and morph targets

[CGTalk Moderation]

This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.