Kim Libreri looks at the current trends and projects the future paths of the vfx industry.

1. We will build a better virtual human
Recently we have begun to see a stream of movies that are using virtual humans to represent principle characters. Each movie has chosen a unique solution to the problems that animating convincing human performances provide. With The Incredibles, Pixar provided a heavily animated and stylized approach to superheroes. In complete contrast, Sony Pictures Imageworks deployed a very technological motion capture-based approach to creating the near photorealistic humans for The Polar Express.

Now that we have sub-surface scattering, measured skin and material shading models, real world sampled lights and sophisticated hair styling and rendering systems, I am absolutely convinced that we can render totally photorealistic still renditions of digital doubles that are indistinguishable from a real actor. It’s when things start to move that the problems appear. Human beings don’t move the same way that machines do.

The complex nature of the human skeleton and muscle structures mean that our movements are not smooth, linear or easily defined. This has made creating realistic animated human beings very difficult without using motion capture, and even that has a long way to go before we see perfect animated skeletal reconstructions. In addition, it is not practical to have a movie that is purely based on motion capture; there are always cases where a real human cannot achieve a desired performance and the skills of an animator are required.

What is needed is a system that allows an animator to drive a human skeletal in a very gesture/pose based way and let the computer fill in all the nuances of joint offsets and non-linearity. Animators are the souls of the digital actors and should not be burdened with the concepts of biomechanics.

We created simulated human forms for the “Burly Brawl” scenes in the second and third Matrix movies, but the process we went through was extremely complex and time-consuming. This will become much easier as animation systems evolve that are able to recreate the subtleties of human movement. By applying the data (and lessons learned) from motion capture sessions onto the tools that animators currently work with, we will have the technology to create realistic virtual human beings from scratch.

2. We will be able to create believable facial animation
We will also learn to do realistic human facial animations, but this will take a lot of work. The human face is immensely complicated and has many moving features, which are not easy for a computer to define. At the moment we are able to do motion capture of an actor’s facial expressions, but we still can’t even transfer that data to another character’s face in a simple and realistic manner. Some good first steps have been made in the area of realistic facial animation: for example, I thought that ILM did a good job on The Hulk, Imageworks deployed an interesting MoCap solution for The Polar Express and, of course, we developed our own universal capture system in the Matrix sequels for Hugo Weaving, when creating the growing army of Agent Smiths. But I expect to see dramatic progress in this area in the coming years.