Tri vs quad modeling questions

Grunternz

I'm really interested to clear this up .

At school in my 3d animation class we've been taught to model using quads, and to eliminate as many triangles as we can, even to the point where we have none on whatever our model may be. Then I look around places, and people specifically try to only model using tris :poly141:.

What are the advantages of using one vs the other?
Is it wrong to mix tris and quads on the same model?

What is more commonly used in games and animated films?

Thanks to anyone who can help me clear this up :thumbup:.

pior

Hehe the good old one :P

There is many different situations, asking for different approaches.

- basemesh to be subdivided and sculpted in Z or mud, not to be animated : preferably quad only. Some apps even crash when subdividing tris, ngons or star shaped topoogy. but some app do fine.

- Ingame mesh that actually appear on screen and is animated : can be anything, as the engine chew on tris anyways. Nice quad/loops around the joints and face is handy to have tho, since it makes the rigging process much easier and cleaner. But tris only is fine, as it might even give you a tighter silhouette, matching your basemesh better.

- Film model that appear on screen and is animated : will most likey use displacement maps for the details, hence require a nice clean flow for subdivision and displacement. Quads mostly.

Tricky huh!

Grunternz

Thanks Ill bookmark this for future reference.

Mark Dygert

Short answer: For final in game models use mostly quads, except where it makes sense to use tris.

Grunternz wrote: »

At school in my 3d animation class we've been taught to model using quads, and to eliminate as many triangles as we can, even to the point where we have none on whatever our model may be. Then I look around places, and people specifically try to only model using tris :poly141:.

Games use tris. Always count tri's not polys. Polys can be made up of any number of tris.


You can feed a game a quad but it will break it down into tri strips. you thought you had 4 polys but one of them was a 5 tri poly the rest where 2 tri so you ended up with 11.

Its perfectly fine to use tris when needed for final in game models. However quads are easier to work with and they do make skin weighting characters a little easier. Mostly because quads preserve loop and rings where tris create poles.

The hidden edges in quads can deform oddly and need to sometimes be flipped or reenforced. Maya is especially bad about hidden edges, it recalcuates them all the time, flipping them the way it chooses, which leads to a lot more tris than is nessesary. Max allows you to flip hidden edges and they stay flipped without having to make them visible. Preserves loops and rings which is great.

Sticking to all quads can cause tri counts to sky rocket. You end up with a lot of edges runing wild all over your mesh because of a few details on the dark side of the moon. There is also a strong chance that a lot of edges will be packed in together tightly making it a mess to skin weight.

Evenly sized quads(more or less) with a few tris to control try count is probably just about right.

Grunternz

Thank you both very much. Would you say I'd be at a disadvantage learning Maya if I was interested in the gaming industry, instead of Max? Or are they coming closer and closer together seeing as they are both owned by Autodesk now?

Could you possibly give me a few examples of games created with Maya / Max that might be well known today?

praxedes

I happen to know a guy who coded on 50 Cent 2 and their art dept apparently uses just Maya, his previous place did just Max. I'm on Maya cos it came with my computer, but I have Max trial ready to install.

I have been told that it's less about software and more about core skills, in that if you can model well, you can learn the toolset (but of course, the tricks of each softwareare another matter )

~P~

Blaizer

Games uses tris, but being in animation classes is quite normal of professors to ask for what it's really needed, Quads.

Models for animation should avoid triangles at all costs, due to bad deformations and bad smooth results when turning on the subdivision. I think it's great to have a 100% quad mesh (with the needed level of detail), it's not as hard to do as too many think.

In the early stages of modelling a model for animation, tris and pentagons are used in some topologies. So there's no need to worry about it very much, because the trend is to make a very low poly model to define loops, and then subdivide x1, and later, if needed.. it's common to optimize the result removing loops of edges, etc. Loops and topologies are very important for good deformations. Due to all this, we hear: "suitable model for animation".

I think it's better said with an image:

In this pic you can see tris and pentagons. Understanding the subdiv technique, you can make models very very fast, look how easy to do are hands. This one was done in more than a work day and it was done for a tutorial, for helping a friend of mine to understand my workflow. When you say.. 1 day.. people call you liar, and the worse, your firends too . too much ignorancy.

Hope this helps.

dragonwithin15

Blaizer wrote: »

This one was done in more than a work day and it was done for a tutorial, for helping a friend of mine to understand my workflow.

did you document the tut? if so i would love to take a look at it if you don't mind

-angie