Tris vs. quads
Author
I've heard that if you are modelling for games, then you should stick with triangles and not quads or n-gons, because DirectX/OpenGL's rendeing architecture renders triangles, and converting from quads to tris is non-deterministic and produces bad results. Not to mention the fact that you usually end up using more polys than you need to. Is this correct?
It's true that going from quads/n-gons to triangles is non-deterministic, but I wouldn't say it yields "bad" results. If you are modelling characters and such you might not care that your exporter triangulates the polygons slightly differently than Max/Maya as you probably won't even notice. If it's easier to model for you in quads or n-gons, I wouldn't worry about it too much, just make sure your exporter triangulates on the way out.
I am inclined to think that it is best to triangulate just before you export, i.e. let the modeling program do the triangulation.
Presumably it will have more consistent results than letting whichever exporter do the work, of course if the exporter in question uses the modeling’s triangulation then this point is moot.
However it is a good idea to see the triangulated version in modeler before you see it in game so you can presumably be in a better position to pick out differences/make sure the exporter/game load code worked as expected.
Presumably it will have more consistent results than letting whichever exporter do the work, of course if the exporter in question uses the modeling’s triangulation then this point is moot.
However it is a good idea to see the triangulated version in modeler before you see it in game so you can presumably be in a better position to pick out differences/make sure the exporter/game load code worked as expected.
Realisticaly, you are working with triangles anyway, the program just hides it from you. (Triangles are the only objects that can be rendered, even things like curves get converted before they are drawn to the screen.)
If you want to see what the triangles look like, use a modifier to convert it into triangles. (In max you would use the Tessalate modifier. Sorry, I don't know what the other programs would use)
If you want to see what the triangles look like, use a modifier to convert it into triangles. (In max you would use the Tessalate modifier. Sorry, I don't know what the other programs would use)
Quote:Original post by methinks
Realisticaly, you are working with triangles anyway, the program just hides it from you. (Triangles are the only objects that can be rendered, even things like curves get converted before they are drawn to the screen.)
exactly, and the reson for using triangles is you need to make every poly count so if you use quads you only get one poly to work with but are rendering two, so if you actully want a square surface its fine to use quads but if you want a smooth organic shape your better with 50 tris than 30 quads
In max, working in editible_poly is very similar to editible_mesh,
both you can have tri's/quads/n-gons and you can also triangulate (or turn edges in edit_mesh)
In max you always know your triangulation by looking at it - if it matters and you can retriangulate quads/n-gons.
methings - you would want to avoid using a tesslate modifier as that will increase you polygon count anyway. There are many other ways to view the tris.
Most people use a combination of tris and quads when modelling,
As people have said, it all gets turned into tri's
So it doesn't matter much, its visually useful to the artist to have a clean mesh.
both you can have tri's/quads/n-gons and you can also triangulate (or turn edges in edit_mesh)
In max you always know your triangulation by looking at it - if it matters and you can retriangulate quads/n-gons.
methings - you would want to avoid using a tesslate modifier as that will increase you polygon count anyway. There are many other ways to view the tris.
Most people use a combination of tris and quads when modelling,
As people have said, it all gets turned into tri's
So it doesn't matter much, its visually useful to the artist to have a clean mesh.
since graphics cards can/will only render tri's wouldnt it create a very, very slight jump in performance if you strictly used triangles?
When modeling for games, you model in whatever way works best for you.
I'd say most if not all of us work mainly with quads, it's a lot easier to get a clean topology. If you can solve something with a tri and save a poly, you just build a tri instead of a quad.
Working with quads doesn't mean that you won't use tris, if a tri is all you need in that spot.
You convert to tris when done and manually go over the trippled mesh and spin tris that didn't tripple in a decent way.
You can do the UVs before trippling (less polygons to move around) and just tripple the areas necessary to get the seams the way you want.
That's the most common and standard way to build game models.
I'd say most if not all of us work mainly with quads, it's a lot easier to get a clean topology. If you can solve something with a tri and save a poly, you just build a tri instead of a quad.
Working with quads doesn't mean that you won't use tris, if a tri is all you need in that spot.
You convert to tris when done and manually go over the trippled mesh and spin tris that didn't tripple in a decent way.
You can do the UVs before trippling (less polygons to move around) and just tripple the areas necessary to get the seams the way you want.
That's the most common and standard way to build game models.
This topic is closed to new replies.
Advertisement
Popular Topics
Advertisement
