i am not a professional modeler at all…for the time being it’s just learning/hobby (although i would like to get professional). most of the time i find myself modeling objects i have around, or maybe cars…everything not organic for now just for practice you know. i often read people say “no never use tris in your models!” “ngons are bad!!” “your model MUST be made up of quads!!”. this seems to be connected to the fact that if you are going to send your model to someone who need to deform it for animation, it would cause problems here and there around the topology. but what about hard surfaces? for example, i have a car model i would like to sell, and the model have some tris. it is that bad? for my experience, triangles cause problems if i have to apply a subsurf modifier, or i have to smooth it. but what about if i model a car surface that it is not meant to be subdivided , and the topology has triangles that i need? i try to understand what the professional point of view is about this subject, and actually if there is one really.
(as usual bad english, sorry!)
Moved from “General Forums > Blender and CG Discussions” to “Support > Modeling”
As a general rule of thumb, topology tends to matter less if the mesh isn’t going to be deformed (e.g. rigged for animation).
mainly quadrilaterals because those make a readable, workable and deformable topology. triangles are ok in right places and fully triangulated mesh is sometimes a requirement (export file format, game engines). n-gons are used to help to getting to quads and tris. for selling your models the sites may have specific requirements for the geometry, like https://support.turbosquid.com/entries/20203606-CheckMate-Pro-Specification which doesn’t list anything extraordinary. basically says “keep your models error free and don’t model like a muppet”. you don’t need triangles if the model is using quads otherwise. that just means you don’t want to bother making it a quad which may be the right course of action, but you don’t actually need a triangle, even if you’re modeling a triangle.
The problem with triangles and subdivision isn’t the triangles themselves, it’s the poles they create. Topologically, you can not have a mesh without poles, unless you’re modeling something like a Torus. You need to learn how to work with poles - if you can keep them in flat (and undeformed) areas, they don’t cause problems.
After the first subdivision step, all triangles will turn into quads anyway - you can actually use this to your advantage by keeping a simpler control mesh through strategically placed triangles. This is probably regarded as bad practice in some places that worship quad-only meshes.
NGONs are mostly there to help you modeling. You’re basically dealing with an arbitrary triangulation, which will cause trouble unless the entire polygon is flat. If your NGON is flat, I would keep it in the final mesh - it makes it simpler.
Whether a mesh is good for animation depends more on edge flow - another thing where triangles may cause trouble. Still, the solution is not to blindly eliminate triangles, but to understand the flow and deformation. The actual deformation algorithm may work on triangles anyway.
If you just have a static mesh and don’t use subdivision, just do whatever keeps your mesh manageable.
if you need only quads then works with Nurbs sufaces
happy bl
For pure modeling things can be approached differently.
For animation there are preferred ways to do it and it is true that quads work best because you prevent
pinching which can also impact deformations for the geometry and textures.
For object modeling you can basically mix and match the modeling approaches based on what you are the most
comfortable with.
Tris as BeerBaron said are not bad - the question is more to you want something pinched or smooth.
So for smooth obviously Quads and NGONS work better.
I use sub-d quite a lot for concept modeling and mix quads and ngons.
If you are dealing with a smooth surface lets say a car door side then you want to have a topology that
allows the subdivision surface modifier to smoothly skin it. Adding tris there could show up in highlights.
So in that situation you might want not to use it. If in other cases a triangle will describe the shape
you want use it.
But the rule is really the less points you have with sub-d modeling the easier it will be to maintain smoothness.
The problem gets hard when you want to mix smooth and sharp structures with just Sub-D.
Thats then an area also like Ricky mentioned where you mix Sub-D and NURBS in workflow where via polygons
you build your surfaces as separate objects but use NURBS tools to create the transitions.
Example: low poly Sub-D sphere that as a triangular opening.
I hope this explains this a little more.
with all that information you will run into a kind of puzzle when modeling with the help of ngons, and you will definitely need help to get right topology. I recommend watching Youtube videos, keywords topology, hard surface, poles, edge flow. A proper topology is fundamental for perfect renders. Many issues or artefacts have their reason there, long before fiddeling in the render stats.
As a general rule, quads will do it. You can get good topology, with half the faces of tris, and edgeflow can be customised more easily. The downside is, achieving quads can require solving to quads, which isn’t always easy. The end result will be neater, though.
Personally, I will avoid ngons. Try ngons with subd and you are in for a hell of a mess.
Subdivision will still try to turn Ngons into quads though. Sometimes it can look pretty neat, there is still a minimum amount of geometry needed though. Example:
You need to take the end user into account - if the model is for your own use in Blender, then whatever renders well is good:) If you are selling it, bear in mind that renderers such as Daz studio and Poser simply cannot read N-gons correctly and make a mess of them. They also automatically apply smooth shaded, so tri’s, especially long thin ones, often don’t work too well.
To my mind it is good modelling practice to stick to quads - they work with anything!
Lots of good responses there. I don’t think the use of quads, tris or ngons matter until they cause you a problem. Either in a pipeline or your own personal projects. But I believe it takes some experience to know where they can be used effectively. I haven’t fully explored the possibilities myself. For the most part I stick with quads, simply because I feel the way they shade is slightly easier to predict.
Modeling with quads and getting comfortable with them, is a good place to start. From there you could then start mixing it up a little as your confidence grows. 