direction about the right topology in production
by Armando Sepulveda
I don’t think there is a perfect topology and each production could require an specific topology design. Feature Animation, Film, Video Games, they all work in a total different way at the time of moving and deforming models, therefore with different topologies necessities. Even based on the engine Mental Ray, Renderman, etc, topologies might require some specific designs in order to be rendered properly. So, there is always and special situation for each model and topology design, but anyway there are some conditions to follow to do a better model.
As a modeller and understanding how a pipeline/workflow works into a production, is important to have in mind who is coming after us to work on the model, what the character (in this case) is suppose to do or even for later cleaning, editing, etc.
Sometimes a better topology makes easier and faster to change the geometry based on the requests coming from other departments or even Art Director/Director. Adding loops, removing them, adding details, etc. Sometimes even reusing the geometry to create a new character.
Probably the most important aspect in a topology is that gives a perfect deformation when the character is rigged and animated.
Here is an screen shot of a realistic rigged hand. Is NOT a modelled pose, this is the deformation coming from the rig (here is a link to a post about this rig setup by Igor Stefanovic)
Depending on the complexity of movement, flesh or sliding skin, or even the category of the CG, Film or Feature animation topologies could differ. Here you can compare two examples of them.
In both cases the flow of loops goes in the direction of the foldings, helping a better deformation
We also know that the model will be subdivided and in this process some surface artifacts might show up, because a wrong topology, triangles, wrong reductions, extraordinary vertex (more than four edges) in the wrong place, etc. So a clean mesh will result in a smother and more controllable surface
Some models could be deformed very well, but very hard to tweak, or even to be reused. So our challenge comes when we try to reach all this necessities.
It’s impossible to avoid vertex with more than four edges, where several flows meet in one point. We called this vertex “extraordinary vertex” of in nurbs “five patch corner”, and is there where all this patches are stitched.
I follow as a rule trying to keep them in a same loop.
Well if we were working in nurbs ther is no other way, but talking about polys is just because makes easier to edit, tweak, reuse or even combine with other geometries. Lets say an ear, that we can just separate and take it to another head.
Adding or removing loops won’t be difficult at all, if some department ask for that. Here is an example
So, the main difficulty is giving the right volume and interpretation at the same time we follow just a few rules, to make the geometry more effective and these are those rules.
A. Flow direction based on volume and deformation
B. Continuity between extraordinary vertex
C. Clean geometry avoiding triangles or any other different than a four side polygon
D. Patch modelling or defined areas
Here some production examples