Join Jason Baskin for an in-depth discussion in this video Facial topology guidelines and problem spots, part of Facial Rigging in Maya.
- View Offline
Before we jump right into rigging tools, it's important for us to quickly review how a three D mesh should be created to give riggers and animators the maximum level of flexibility. While every character's face is of course different, there are some standard topology guide lines that you'll want to keep in mind when creating a character who will need to speak and emote. So here we have three different characters. On the left is the Walter character which you may have seen in other lynda.com animation courses. In the middle is an older character of mine named Blake, which has been used by a lot of student animators for animation exercises.
And on the right is a new character that I'm just developing named Mike. And we'll be using this character quite a bit in later lessons. We'll be taking a closer look at the topology for all three of these characters but first let's review some preferred modeling guidelines. Ideally all modelers should strive to create their models based off of four-sided polygons, or quads. Quads have several advantages. The first is that they receive textures very well. Textures have minimal distortion when mapped onto a quad based surface.
Second, quad based models are most compatible with other software packages such as ZBrush and it's not at all uncommon for an artist to utilize multiple software packages when working on a single asset. And third, a quad based surface deforms well for animation. In this example, the edges would easily compress horizontally or vertically which makes this a very flexible surface if we want to add deformations. The challenge for organic modelers is that often times, a character's face needs to deform in several different directions.
So a single grid structure may not accommodate all the deformations that are needed for a character's face. So to deal with this challenge, modelers make use of shifts in directions of the polyflow of the surface and this is done by using polygons known as poles. You can see that unlike the quad surface we looked at previously, where all four edges were converging at a single point. For this quad, one of the points has five edges converging at a single point. And this creates a change in the direction of the flow of the grid.
Here's another pole, this one shifts direction by having three edges converge at a single point rather than four. But again you can see that the grid does shift about 45 degrees at the location of the pole. Triangles also will result in a shift in direction when combined with quad based surfaces. I should mention that in games, it's not uncommon for models to be comprised entirely of triangles. And in fact whenever an object is rendered, it is converted to triangles; sometimes as just part of the rendering process.
But we do need to pay attention to how this changes edge flow of a character when utilizing triangles in a quad based model. So I've gone ahead and made some color assignments to these mashes so that we can more clearly see what's happening. You can see that all of three models have blue edge loops radiating from the openings in the face. The two models on the left also have edge loops emanating from the mouth and this is a very common topology for dealing with a character that's speaking. The green loops indicate an eye mask edged loop, which is another common modeling practice.
And you can see that the green loop encompasses the two blue edge loops for the individual eyes into one larger edge loop. This can be handy for modifying the brow arch of the character. On the middle character there are some phases the are red, these are five or more sided polygons which generally should be avoided in all models. In this case, they didn't create any major problems because they are located in portions of the face that don't deform much, but I did go back at a later point to try to revise this model and improve the range of animation.
And the five sided polygons did interfere with my ability to make some of those changes. The yellow phases on these characters indicate areas where the polyflow has shifted direction. So these are the poles that I was speaking of earlier. So the two characters on the left are essentially using the same structure. The one on the right has a slightly different, more grid based structure around the mouth which, I found to be pretty effective for cartoonier characters. I also just quickly want to point out that whenever a character is subdivided, anything that was a five sided or three sided polygon does get converted to a quad as a subdivided piece of geometry.
But we definitely want to pay attention to how the edge flow that we create manually affects the surface of the topology prior to smoothing. Taking some time to ensure that you've considered how a model's topology might enhance or limit a characters performance can spare you a lot of headaches when you begin creating a rig for your characters.
- Understanding morph targets
- Using Soft Select, Artisan Sculpt tools, and symmetry options
- Splitting symmetrical shapes into asymmetrical pairs
- Connecting eye rotation to the GUI
- Working with eyelids, using blend shapes or fan joints
- Creating a basic UI control
- Connecting controls
- Orienting joints and creating a support structure
- Attaching joints to the support structure
- Binding joints to the final mesh
- Balancing shape fidelity with rig density and usability