Have you ever played a video game or watched a movie with almost perfectly lifelike computer graphics – save for one unrealistic movement? And it’s that one misplaced hair that throws everything off and breaks the immersion? It’s an effect called the ‘uncanny valley,’ and it drives those who work in computer graphics up the wall.
Last year, video game publisher Ubisoft, through their research and development lab La Forge, joined forces with McGill to create the new NSERC/Ubisoft Industrial Research Chair in Believable Virtual Character Experiences, with the goal of inching computer graphics closer to perfection. Filling that five-year position is associate professor of electrical and computer engineering Derek Nowrouzezahrai, and he is now tasked with finding and training personnel capable of closing the 0.01 percent gap between uncanny valley and believability.
“We can generate a single picture with no humans in it relatively accurately,” says Nowrouzezahrai, pointing out that for years, Ikea catalogue photos have been completely fake and made with computers.
“For a single shot of a human face, if an artist takes four months to go through every pore to get the dimples just right, we might get a shot that’s convincing. But a human in motion, with hair blowing in the wind? There are all kinds of subtle, physical interactions we need to nail just right.”
It might seem obsessive to a fault to put so much time and effort into incrementally improving computer graphics when so much progress has already been made, but from Superman’s CGI-removed mustache in Justice League, to Brad Pitt’s backwards-ageing face in Benjamin Button, we’ve seen countless instances where successful and unsuccessful attempts to push realistic graphics have had a major impact on the way people appreciate a piece of entertainment.
“As humans, we’re programmed to notice anything that’s off with a human face, so you need all the details right there,” says Nowrouzezahrai.
It’s the perfect time for an effort such as this one, he observes, because the field of computer graphics is now in its teenage years, achieving a new level of sophistication. In the past, computer graphics borrowed liberally from physics, linear algebra and computational statistics. But in recent years, the loop has been closed and the strides being made in computer graphics can actually inform other fields of study.
“It’s a telltale sign of the growing maturity of graphics as a field,” Nowrouzezahrai says. “Gaming is about delivering these graphics with speed, but maybe you can slow down skin modelling techniques and use them to help detect skin cancer? Every student’s project under this chair is going to be a little different. They’re going to become experts in these areas, and once you’re an expert, you have the capacity to grow in any direction you want.”
In turn, students will have access to Ubisoft La Forge’s unique resources. In the case of a big budget AAA video game title like Assassin’s Creed, development will typically take years with hundreds of people contributing, so the five-year plan for the chair isn’t considered a particularly long time in the world of gaming.
During that time, Nowrouzezahrai says Ubisoft might come to his team with a very specific problem – like a lack of realistic reflection on coffee cups in their games – in need of solving.
“This isn’t stuff that’s happening in a silo tucked away on campus,” says Nowrouzezahrai. “Our students will spend time at their studio working really closely with La Forge on these problems. Ubisoft has a wealth of expertise, experience, data and software tools that we couldn’t recreate in our lab.”
Montreal is a place where creative industries thrive, and the hope is that once students leave the program, those who are interested will be able to find compelling work creating the tools that provide emotional entertainment experiences for all.
“We’ll always need talented artists,” says Nowrouzezahrai. “They bring the life out of the algorithms. There’s a nice mix of science and art in what digital artists do. But physicists, statisticians, applied mathematicians, computer scientists and more, they all have a place to fill in solving these problems.”
