Recently, I have been looking into the workflow for character generation. The best way to look at how to improve the process of creating and animating characters is to follow the process to find out the challenges and the limitations of standard tools. A typical pipeline for character modelling looks something like this:
The white blocks are followed in sequence, while blue blocks represent generated assets. Since I haven’t worked on a character for some time, I decided that I would avoid having to design a character, instead opting for a pre-existing subject. Rogue One’s K-2SO character seemed like a pretty good place to start. While the character does not present the challenges of organic deformation, the rigging and texturing phases would be quite interesting.
Finding imagery for K-2SO and his poses and gestures was very straightforward.
From the research I gathered, I generated a set of illustrations to be used as image planes for the 3d scene.
I also had some decent close-up reference for the head.
Producing the base model took around 4 days to complete. The symmetry of the character helped a great deal, but there were some asymmetrical details too. I had forgotten how rusty I was at modelling, and I forgot how long it takes to actually work on a model.
With the base model done, I can now look at the other areas of the character creation pipeline. There are quite a variety of tools for uv-unwrapping, surface modelling and texturing. It will be interesting experimenting to find the best new tools.