by Salahuddin Khan

Created on: December 13, 2008

The leaps and bounds by which modern computer graphics have evolved are truly amazing! Speaking as something of an old timer who got into 3D CAD software back in 1977 and spent the next 21 years in that business, leaving as the Chief Technology Officer of Computervision Corp in 1998, I had the privilege of seeing first hand every step of that journey.

Now, with the abundance of computer-based games ("video game" is too incomplete a term) that engage not just our hand-eye coordination but often our whole bodies, many of us take this whole thing in our stride and are only left to wonder about whatever might be the next jaw-dropping experience.

Both within computer-based games and more broadly, entertainment, a final frontier seems to be the use of computer graphics to represent humans and animals. Given the large amount of brainpower we dedicate to the cognitive process of understanding faces, it shouldn't surprise anyone that this is where the brain is least willing to be fooled by crude or even fairly sophisticated approximations.

Indeed, as is often the case, the science fiction writers are way ahead of the technology. An extremely entertaining treatment of the subject was given by Andrew Niccol's movie S1M0NE (New Line Cinema) starring Al Pacino and Rachel Roberts. In that movie, the beautiful S1M0NE (read SIMulation ONE as well as Simone) is so convincing that illusion is so successful at swaying the world that the truth, when revealed, is rejected and the result is a monster that the creator cannot control; more lovely and realized in a more contemporary context than Dr. Frankenstein's, but a monster nonetheless.

Enough of the film critic though, we're talking about what it takes to achieve that sort of realism. The answer is a lot! And what it takes is as challenging as it is fascinating about what it reveals about human cognition. To get underneath the technical challenges, we really have to understand how we humans approach being convinced by the rendition of a face.

The seeming paradox here is that in fact, we require both a large amount of information and very little in order to believe that we're looking at a real face. Let's discuss these requirements and challenges in more detail.

From the "large amount" viewpoint, a complete solution to animation is made up from several levels of realism. The core is obviously a 3D representation of the geometry. Nowadays, there's not much of a challenge there. It's been done and you have only to look at such classics

• 1
• 2
• 3
• 4
• 5
• 6
• Next Page >>