[a / b / c / d / e / f / g / gif / h / hr / k / m / o / p / r / s / t / u / v / vg / vr / w / wg] [i / ic] [r9k / s4s / vip / qa] [cm / hm / lgbt / y] [3 / aco / adv / an / asp / bant / biz / cgl / ck / co / diy / fa / fit / gd / hc / his / int / jp / lit / mlp / mu / n / news / out / po / pol / qst / sci / soc / sp / tg / toy / trv / tv / vp / wsg / wsr / x] [Settings] [Search] [Mobile] [Home]
Settings Mobile Home
/3/ - 3DCG

Thread archived.
You cannot reply anymore.

I figured this warranted it's own thread because of how essential the understanding of this subject is for novice modelers such as myself and others on this board. Experienced modelers feel free to correct me or add anything.
I break everything down like I do because I'm half retarded and need things explained to me as simple as possible

Triangulation is making an object of quads (square faces) into an object of triangles. Most artists prefer to /model/ in quads and /animate/ in triangulates (once they are finished with the sculpture; after rigging and before animating, to be exact). Triangulating is most important in the rigging process to identify any inconsistencies in your quads and correct them for optimal mesh rotating and deformation. One such inconsistency before triangulating might be a quad might invert on each other (called triangular inversion) when moving a bone in a certain way, which can be identified by a sudden change of color. This can become quite problematic later on in the animating process but can be easily resolved by triangulating, and further manually adjusting said tris if other problems persist.

Game designers choose to triangulate because game engines run in "real-time" (meaning if you press an action, the game does it on the fly. The game acts as you act), and the higher the poly-count, the greater the demand of the system to play the game. Triangulating is essential for optimizing poly count, and while triangulating on its own won’t remove excessive edges, it can be used as a guide to visualize which edges can be removed. Game engines will automatically triangulate your model in any case, so it's best to do it in the creative process before porting to make sure everything functions as intended. Movies are different in that they render in "sequence" (meaning frame by frame), so they can afford to have considerably greater polycounts.

So does all of this pretty much explain why triangulating is important? Is it accurate, and is there anything else to address?
Triangulation is also important when exporting to an external texturing program (e.g. Substance Painter). Any target engine and texturing software are likely to auto-triangulate anything that's in quads, and you can't guarantee how it's going to do that. Hence is very likely that say, Painter is gonna triangulate differently than Unity importer, which can really mess up your textures.
That shouldn't be happening under normal circumstances.
UVs are defined as vertices on a 2D plane that form quads. Split a quad in two to form two triangles, do the amount of vertices change?
didn't read but cute fox :3
This isn't something that should be that important to you as an artist, especially if you are just starting out, but I'll ass my two cents in case there are other beginners here thinking hard about this as well.

As an artist, you don't have to worry about triangulation at all. All 3D is handled in triangles internally, though they can be displayed as quads. In any 3d package, the quads are already triangulated you just don't see the edge that splits them, unless you ask the package to do so. In the industry when they ask about polycount they usually refer to triangle count because that is the true measure of the model, quads are mostly never counted to my knowledge since they are just for display.

Also polycount is not as important anymore when it comes to performance, and when it is, you have someone who budgets each model based on what platform you are deploying to and sets a target for each mesh and the artist follows that guideline as best as they can, some artists are better at using geometry better than others, they can get better shapes with less but this comes down to experience.

If you build a model out of quads and it fits your target polycount, you should not be turning them into triangles willy-nilly, that does not change the performance internally since the engine already reads it as so. If you build a mesh made of triangles and quads that is also acceptable as long as you know where to use triangles as to not interfere with rigging/animation but in most cases, the performance increase it's negligible and only really matters if you are severely restricted like in some cases it can be for mobile.

TLDR: Don't worry about triangles so much, stick to quads they are easier to work with. If you are optimizing a mesh, learn where triangles are ok to create and know that this should be the last thing you want to do, ideally you build the model with a proper polycount from the getgo.

If you have any questions or want me to clarify let me know.
OP here and I still have one other question. Why do people choose to triangulate only a portion of their quads instead of the whole mesh? Like in this post.
Also, animators don't care what the model is made out of as long as they can rig it properly or it already is, as long as it deforms well, triangles/quads don't come into the equation at all. A good rigger can work with either, though its a lot easier to rig an all quad mesh with good Poly Loop flow.

Triangles can mess up the rigging process when they are placed improperly in deforming areas of the mesh because they are harder to skin when you mix them with quads.

I don't know where you are getting this triangular inversion thing from, in all my years of making rigged models I have never heard of it or come across it.

Game designer don't decide to triangulate, the engines are built that way and they read as tris, they don't have a say in the matter.
As I was saying before, all quads are made of two triangles, internally the engine chooses how the edge that splits them is connected, normally you don't have to worry about it, but when you are dealing with such super low poly models how that inner edge that splits the quad can make a difference in how it looks, so an artist can instead choose his own way of splitting the quad to get the best looking results, because the automatic way the engine picked might not be the best for that case. Engines let you flip the orientation of the edge, or you can manually just connect them.

Does it make sense?
>If you are optimizing a mesh, learn where triangles are ok to create and know that this should be the last thing you want to do, ideally you build the model with a proper polycount from the getgo.
Elaborate on this?

And if I understand correctly, are you saying that the greater priority should be efficient triangle use rather than efficient polycount? And why aren't quads ever counted?
I think I understand, yes. That makes sense. Basically you only cut quads into tris where its needed to get the most natural looking rotation/deformation. It might be more necessary on the face/shoulders/hair than it is for the thighs, for example.
Essentially keep triangles out of deforming areas when possible, for example, if you have an arm, keep triangles out of the shoulder, put them in the bicep, since the bicep won't bend, keep them out of the elbow because it will bend, and you can have them in the forearm if you aren't doing anything fancy with the wrist like twist bones. If you have something that does not deform, meaning bend or twist, then you can have as many triangles in that area as you want, if you have something hard surface/mechanical that does not again deform, then triangles are not a problem at all.

I meant that there are more important things in game engines for performance than polycount, like draw calls. If you are building a model for something, you should be looking up how many polygons that mesh should have and sticking to it, assuming you are practicing for that or creating a portfolio piece, for example, if you are building an asset, let's say a character, and you want it to be triple-A quality, something you might find in a ps4, you should be looking up what is the average polycount for characters in ps4 triple-A games, then working with that polycount in mind, try not to just build things blindly, this will also help you if you want to have it as part of your portfolio later on.

When someone asks about polycount, they mean triangles, because they know that's what the engine uses, and that's how they budget the meshes for scenes to be properly optimized. if a game on a console can only run 1mill polys on screen at a time, they divide that 1 mill along how many assets they will have on-screen at any time, and budget something like 30-60k tris per character, this is all arbitrary and depends on many different factors like engine/platform/etc.
Also, remember that many artists will make models without ever thinking about animating them, so in that example, you brought up the person might not take into account how it would deform or how easy it would be to rig. Often people make 3D art for fun, or they are practicing a specific area of the pipeline like just the modeling/texturing. They might not know all the other areas and just be really good at a specific part of it.

That's also something eventually as an artist you might have to choose, one part of the whole process will be more fun for you than the rest, you will inadvertently be better at it and choose to do it more and will end up specializing in it. Other like the whole process and just become general 3d artists.
Another thing to add is, when the engine triangulates it will never mess with the textures, the only difference is the normals of the mesh can look a bit different, but this only happens in really extreme low poly cases, you typically don't have to worry about this.

I am beginning to understand that your use cases and your concerns mostly come out of the really really low poly cases, is this something you are really interested in?
This is awesome, you've been a great help to me and I understand it much better now. Thanks.
And yes, I am working on a project of my own that I have a clear vision for, and it aims for a low-poly style because I like the style. I believe less is more in the sense that the less particular you are about the visuals, the more it leaves to the imagination of the player. The characters I'm working on now have an average polycount of 900 and tricount of 1800.
Lots of really elaborate posts for /3/ standards in a fresh thread.
Are you guys ok?
Ikr? Loving it though. Bump
Somtimes you should use https://www.google.com/
Somtimes you should use <a href="https://www.google.com/">https://www.google.com/</a>
File: 1557507381258.jpg (26 KB, 540x540)
26 KB
File: 2 squares-distorted uv.png (80 KB, 1364x745)
80 KB
The fact that a quad can be triangulated in two ways can give unwanted results, meaning it is sometimes necessary to force one of the two triangulations.

Here, the image shows a vertically symmetric grid on the left. On the right is the vertically symmetric texture with a vertically symmetric uv map.

The mapping of the texture is not symmetric because of the automatic choice made by Blender. There is no fancy morphing algorithm to distort the texture onto a quad; the quad is just arbitrarily converted into triangles.
The mapping to triangles is equivalent to this scheme.
If you want to force the texture to be mapped with the same symmetry, you have to override the automatic choice with your own triangulation.

Delete Post: [File Only] Style:
[Disable Mobile View / Use Desktop Site]

[Enable Mobile View / Use Mobile Site]

All trademarks and copyrights on this page are owned by their respective parties. Images uploaded are the responsibility of the Poster. Comments are owned by the Poster.