The “Break it Down” Blog

Brad Agdern sent in news that a California digital imaging company, Image Metrics, (who produced the animation seen in Grand Theft Auto 4 and who’s homepage has an excellent example of what kind of capture their technology results in) has launched what might be the most accurate facial animation seen to date; possibly jumping the “Uncanny Valley” gap.

The difference between what Image Metrics is doing for example, and what typical facial mo-cap did is that instead of using little balls positioned on the actors face, and capturing the balls, Image Metrics is actually capture pixel-by-pixel the little shifts and nuances in the actor’s face and then translating that to a 3D mesh. This mean that the smallest change (eye judder, creasing around the mouth, etc.) can all be tracked and modeled.

You have to ask yourself, if the simulation of humans becomes more and more accurate, and the production of these simulation increases in complexity and production time, at what point would it just be easier to capture the actor him or herself and using some advanced compression techniques, get a high def animation of them in the game somehow?

Maybe we will skip right past the part where total 3D simulation is cool, and go right back to “just record the damn actors” Command Conquer style!

Thanks TimesOnline!