Triangle Strips

[TheUsualAlex]

In what case scenerio would tristrip be used in a model or effects animation? What is its advantage over a quad strips (aside from much predictable deformation and subdividing)?

Thanks,

Alex

[miguel m]

Hi, I'd only use triangles for games geometry or some other realtime stuff that may require triangles for faster rendering. I believe prman for instance runs slower rendering a triangle mesh than the cuad counterpart.

As for subdivision, well i think that triangles always give more convoluted and irregular subdivided meshes.

Not sure, maybe I'm completely off.

[George]

As far as I know, tristrips are mostly advantageous for realtime rendering applications. Usually games will have models built out of tristrips. The advantage is that you can send down fewer vertices to the video card. The advantage over quadstrips is that video cards, in general, support triangles, while quads have to be triangulated before rendering. Hope this helps.

[TheUsualAlex]

So.... is it more efficient to convert a quad geometry to tri-strip before rendering? Or does it not really mattered?

[miguel m]

If your quads are mostly flat i think it's at least redundant to output them as triangles. I could be slower too for the renderer to parse the file since you're outputting double number of plygons

[George]

I believe that the advantage of converting quads to tristrips is only relevant when you are talking about hardware rendering. Sorry if that wasn't clear in my earlier post.

[danteA]

I think what BoBoFancy is really saying is that you just get faster tumbling and that's about it if you're not plugging into a games engine of some sort.

[TheUsualAlex]

Eh. Understood. Thank you.

[Jason]

Any renderer would prefer to render poly triangles than poly quads. It often makes the raytracing more robust too. Many renderers will internally tessellate your poly geometry to tri's anyway.

Meshes and parametric surfaces (NURBS, Bezier patches, etc), however, should *not* be made into poly's. Meshes/parametric surfaces provide very good derivatives which improve shading quality.

[TheUsualAlex]

Thank you, Jason.

So, I gather that it isn't really necessary to convert my character into a tristrip before rendering?

[miguel m]

prman must be the exception to the rule,

app. note 11.

PhotoRealistic RenderMan is designed and optimized to handle rectangular regions, and one of the results is that a triangle is rendered at the same speed as a rectangle which is twice the size. Looked at another way, triangles render at half the speed of rectangles.

[Jason]

Oh - that's a surprise. Well, in fact, not really I guess. Maybe I'm just not thinking clearly this morning... I guess I should've said "a raytracer".

I guess mantra AND prman would do better with rectangles when it comes to the primary dicing of the surfaces when creating the micropolys for shading. I guess this would be because the patches are split parameterically and splitting a rectangle down U or V would be more efficient than a triangle. However, when mantra is raytracing I'm pretty sure it'd prefer triangles because ray intersection code is most efficient and robust against tri's.

I hope I make sense now... :-)

J.

PS. also - i'm not sure about polygons with more than 4 points... but i'd guess those get less reliable..