The new highs and lows?

Posted by Jon Peddie on March 28th 2013 | Discuss
Tags: nvidia cg computational photography gtc

New definitions needed

If you read the story on the uncanny valley in this issue (page 1), you learned a little about Jorge Jimenez’s breakthrough work. And if you’ve hung out with us for any time, you know Peddie’s First Law: In computer graphics, too much is not enough. At one of our first Siggraph luncheons, in San Antonio, Texas, in 2002, we put the question to our distinguished panel, “Are we done yet?” Not only was the answer no, it was Hell no—are you stupid? Every year since then I ask my friends and colleagues, “Do you think we’re done yet?” The answer is still no, but not quite as emphatically. TWO VARIABLES: Relative impact on disbelief from imagery and story.

One of the forefathers and all-time giants of the computer graphics industry is Dr. Jim Blinn. Jim has invented many of the foundation tools we take for granted like bump-mapping and specular reflection. And he also postulated a law: “As technology advances, the rendering time remains constant.” 

 

Our two “laws” tie together nicely and say about the same thing, except he’s a famous computer scientist and, well, I write a newsletter. But if our laws hold true, and we’re never really done yet, I’m starting to wonder if it’s getting asymptotic and we may not know if we’re done yet. However, there are plenty of challenges on the horizon, not the least of which is 4K displays, and multi-tasking dynamic load balancing between graphics operations and GPU-compute operations.

On the content creation side there’s still lots of work to be done, and truly a black hole of need for CPU and GPU horsepower. However, there are only so many DCC/M&E developers out there; not enough, with frequent enough buying cycles, to support the R&D that goes into next-gen GPUs.

What I saw at Nvidia’s GTC conference was exciting, encouraging, and exemplary. And as good as the presented images were, I could still see room for improvement. We really aren’t at the point of suspension of disbelief, not completely. However, we are—or at least I am—experiencing it more often.

There are two variables in reaching suspended disbelief: the image quality and the story quality. These are somewhat subjective variables, with imagery being a bit more quantitative, and they both have non-linear aspects.

GOOD BUT NOT GOOD ENOUGH: Aki was designed to be as realistic as possible - 12 years ago.The chart in this story illustrates how suspension of belief is more impacted by story than by imagery. Imagery has to go much further to overcome poor story. Examples are easy to find. The very pretty and universally maligned 2001 animation, Final Fantasy: The Spirits Within, really looked great, but not great enough to overcome the lame story. Viewers got bored with the story, and so they diverted their attention to the imagery, which as good as it was, was 2001, and not perfect—we weren’t done yet.

 

Aki looked great, not creepy at all, but also not quite great enough to overcome disbelief—we never really thought it was a real actor or a photograph. However, I believe if the story had been better, we would have easily slipped into it and stopped looking at the characters but rather at what they were doing and saying.

Toy Story and other Pixar-Disney hits proved that story trumps imagery. The irony of that statement is the enormous amounts of computer artistry and horsepower that went into those hits—but the characters and scenes were stylized. They were not attempts to reproduce the real world. The second the film started, viewers got the message that they were seeing animated characters, the critic within was quickly sent away, and we simply watched and enjoyed the movie.

Today we’ve learned it can’t all be done in CG. Computational photography, a semi-analog process, also has a big role to play, right now. And by the very fact of saying so, I’ve also said we’re not done yet. If the only way (today) we can create the realism we saw at GTC of Ira is to use a large apparatus like a light box, then we know we’re not done yet.
So as you and Ira can see, we have a ways to go. And, as good as Ira looks, he’s still waxy and not lifelike. No, we ain’t done yet, not by a long shot.

HOW REAL? Ira is looking at you, babe.

Discuss this editorial