This demo illustrates the steps taken to develop a character from concept sketches to finished facial rig in a simplified format. This is an actual rig I developed with the use of proprietary rigging tools in Maya utilized daily at work so I’m not going to demo those tools. However, future installments will incorporate the use of tools that I have written and that you can use to aide the building of complex facial rigs such as this, and include overviews of more advanced rigging techniques. I would also like to break the process down into chunks to really go into more detail about things you should be thinking about if you are trying to rig a face. This is an ongoing process so stay tuned, I have lot’s of useful tips to share.
The simple breakdown:
Concept Art
1) Concept painting in photoshop is the first step to try and figure out a solid direction. I had nothing to go on except it needed to be a sexy female which is difficult because what makes just a head sexy? A lot of fashion models have killer bodies that obviously encorporate a lot of the determining factors, so blocking out and finessing the important attributes of a desirable face is imperative and what make this exercise challenging.
2) Orthographic views and 3/4 of the type of head I was going for. It turns out that I didn’t make the design decision of sticking with a pinup style female until after I sketched the head turnaround drawing inspiration from another concept painting I liked. Thus, I use the head turnaround purely as a generic starting point. Getting it into 3d was more important at this stage than making a final decision on the character design. Sculpting tools make it easier to finalize the design anyway, at least for me.
Concept Sculpt and Texturing
3) It’s hard to really nail down a 2d design before you get it into 3d especially when you have a couple of different genres competing for your affections. I was fighting with making a supermodel or a pinup style face. Luckily Mudbox made that process easier. The final probably looks like a mashup of both.
Sculpting the head is really where I made my design decisions take on an afterlife from their inception, attempting to distill organic forms that looked more finalized.
4) Texture painting with ptex as an experimental process was useful in further exploration of the design. Since I didn’t have to fully commit and make UVs, and I already had good sculpting topology, the freedom to experiment was limitless.
Model Rigging Prep
5) Retopology of the low res mesh using Modo produced the best and the cleanest generic mesh result in the least amount of time. This simple mesh is also highly reusable as a starting point for other designs.
6) The final mesh to be rigged needs to have just enough loops for an optimal resulting deformation after skinning, most importantly the mouth followed by the eyes etcetera. The low res retop is not a good choice for good deformation except if you are trying to output a low poly game rig.
Since I was under the gun to get the model and rig done in a week or two I had to continue to refine the proportions at the same time as modeling good topology. The resulting workflow is a sculpting approach to modeling low res whereby slowly adding more loops, followed by a pass of smoothing and carving the subdivision preview available in Modo, and then stepping back down to the lower res geo, got the job done. Beware when trying this technique in Zbrush or other software because editing proportions and sculpting high res and then deleting the higher divisions will indirectly smooth the resulting mesh thereby erasing the detail. I suggest using this workflow only in Modo where you can hit Tab to toggle on an off subdivisions, Catmull Clark being the best (Shift-Tab). I have to say Modo is supreme for modeling not only for these cababilities, but also the superior advantage of quickly lighting and rendering a preview with occlusion and GI. The physical lighting makes the model feel more real with a sense of scale from cast shadows and a simple skin texture. Blemishes and undesirable defects become glaringly more obvious as well, making it easier to spot and fix.
Rigging with Joints
7) Rigging Starts with joints only with a control rig that either drives the joints directly via constraints or by constraining parent controls so that they drive groups of child controls. Some controls around the mouth are driven by SDK as well so a system of using multiple null buffer nodes is used to get proper tranformations and rotations of the controls or else there would be not so desirable or predictable results.
8) The weighting process is time consuming but with adequate practice and great topology like this some people can get it done in literally hours. I’m slower but that’s because I am a perfectionist when it comes to weighting.
9) The control rig is a series of stock curves constrained to the rig when the rig gets built so it’s easy to modify and save those curves for a perfectly fast result that is highly reusable.
The Osipa style face rig is preferred a lot by animators and the next demo will go over how I use multiple styles of UI on the same rig and the tools to generate said controls.
Subsequent demos will include facial poses and blends and making this rig work with both also known as a hybrid rig. The advantage of poses made with joints only are great because its less computationally expensive but blends are necessary for folding, wrinkling, and some minor correctives.
