I normally always work in this workflow… I’ll do everything in Quads until I’m completely done, then duplicate my mesh and convert to tris as there is no “standard” way to convert a quad into a TRI when going from different applications as our meshes auto convert prior to rendering anyhow.
My question is why are there so many people who say to always stay in quads even when rendering… as deformation will mess up… but I thought it was only applicable during rigging and actually setting up the animation.
My mesh is going to be converted to tris prior to rendering so why would it be bad to do so once I’m fully done animating with quads?
Even with baking and texturing you have to work in TRIS to remain consistent, otherwise you can take your quad mesh from application A to B and it will convert differently making your normals look off.
I guess it’s only because of edgeflow.
As long as the mesh editing is done and the deformation works with the quads there is no more need for edgeflow and tris are ok as long as you don’t need subdivide the model.
In most people’s opinion, the triangulated mesh looks a lot worse. Notice how triangulation turned a non-planar 4 pole into a non-planar 8 pole. Even with a different triangulation, we’ll still be looking at the creation of a lot of 6-poles.
Some triangulated meshes can handle a single level of subdivision, but in general, if you ever intend to do subdivision on a mesh, I wouldn’t recommend triangulating it.
The second thing it’s going to affect is the computation of normals:
Vertex normals are a function of the face normals, which are a function of the positions of the verts that make up that face. In the triangulated mesh, the central vert doesn’t take into account the same verts for calculation of its normals as in the quad mesh. In something like a cylinder, this will create the appearance of diagonal ridges on the mesh.
You can get around that by using custom normals from the quad mesh, and if you want to triangulate all your models, I’d recommend that.
It can also affect UV and normal interpolation (even for custom normals), by triangulating differently in the the presence of deformation, but that’s not necessarily something good. I feel like it’s a relatively minor issue. The lower poly, and the less care that’s been taken with topology/UV distortion, the bigger of an issue it’s going to be.
I consider manual triangulation as something that’s good for some deforming faces and not good for other deforming faces. I don’t think there’s a rule that quads are just plain bad for animation. I also don’t think anybody needs to be “No tris ever!”
I’m actually referring to converting the mesh to TRIS once everything is done. For example my mesh made in one application could look different when taking it into say UNITY if UNITY doesn’t triangulate the mesh in the same way as the initial application I rendered in.
Since there is no standard way to convert a mesh to TRIS prior to rendering in order to maintain consistency I assume you have to convert your mesh to TRIS once finished to make sure the result is correct.
I do a lot of static meshes throughout applications and if I don’t bake with a TRI version and use that version for the texture my normals will look off if the mesh goes into an application as quads, and then once converted to TRIS by the other application it doesn’t match. There is more than one way to split a quad which is why you can get a drastic difference.
When working on the mesh I wouldn’t use TRIS so any subdivision or edge flow wouldn’t be affected.
So I was wondering if converting to TRIS at the very end would affect the animation since everything gets converted to TRIS anyhow, not building or animating initially. All modifiers would be applied and subdivisions done prior to converting quads to tris.
Triangulating isn’t going to affect any vertex positions, and those are the only things that get animated. Obviously, there are going to be differences between an animated quad mesh in Blender and an animated triangle mesh in Blender-- that’s exactly what you’re trying to get rid of.
I’m not sure exactly how Unity handles triangulation. If you’re not going to do any kind of procedural tessellation (not sure about Unity support, I understand that’s something games are starting to do), then I’d say, sure. The only thing that I’d recommend is to create custom normals for the triangulated mesh from the quad mesh, rather than using calculated normals from the triangulated mesh.
Oh, it’s just with baking and texture workflows we have to work in tris and confused on how to do animation work now because I cannot use quad versions at the end, but I’m concerned they wont look as good because people told me that tris animate worse than quads… but it is confusing to me because we only render in tris so if I set up everything in quads and make tris at the end I didn’t know if it would make my deformations worse.
I know several engines that convert meshes differently which makes using a quad version not feasible.
I am not sure but if think about it i would guess only vertecies stores weights , so if you deform a quad only the four vertecies are deformed and the face is the interpolation between them.
So the four vertecies have the same deformation no matter if they are conected with 2 faces (tris) or 1 face (quad).
Only the shading could be different if the quad is not flat and i think thats the reason for the triangles in rendering because they are all time flat.
I’m afraid I’m not sure why they say that. Maybe it’s something I should investigate. I believe that carefully chosen triangles often animate better than quads, but that it’s not normally worth the trouble of carefully choosing them. I can definitely see how poorly-chosen tris could animate worse.
In any case, your SP link is totally right, it makes sense. If Unity does not triangulate the same way, I can easily see how you could end up with some artifacts in tangent-space normal maps. While I do think it’s important to use custom normals from quads, I don’t know how custom normals fit into a SP workflow.
Even outside of animation I know lots of people say “ONLY WORK IN QUADS!” so they assume when you get to the baking and texturing stage you don’t convert to tris, then once they take their mesh into Unity or another engine and apply their normal map the shading is off.
I’ve had a very hard time finding anyone outside of this thread who knows animations and topology well enough to bring light on this so thank you both for responding.
I just assumed if your quad mesh animates perfectly fine adding a tri modifier and applying it would retain the same look but I’m pretty new to animation and wanted to make sure it meshes well with my baking and texturing, and game engine workflow.
