Detonating and shooting an explosion almost certainly adds colour, danger and a sense of realism to a movie's special effects. These days, however, animated technologies can offer film productions a safer, cheaper and more controllable option. A well-designed animated explosion and resulting computer-generated smoke cloud can be alarmingly convincing. This smoky visual effect has been advanced in recent years by the work of four creative computer scientists from ETH Zurich and Cornell University. In 2008, they developed and made available a specific computer algorithm and software that enables animators to recreate turbulences efficiently with lots of detail. This year, the researchers will receive the Technical Achievement Award('Tech Oscar') from The Academy of Motion Picture Arts and Sciences for their "Wavelet Turbulence" software.
On 9 February, professor of computer graphics and director of Disney Research at ETH Zurich, Markus Gross, Nils Thürey (Gross's former post-doc), Cornell professor Doug James, and Theodore Kim (James's former researcher) will walk down the red carpet at the 'Tech Oscars' awards ceremony to receive recognition for their contribution to special effects. So far, "Wavelet Turbulence" has been used in at least thirty major Hollywood productions including: Avatar, Kung-Fu Panda, Monsters vs. Aliens, Sherlock Holmes, and Battleship. Both Iron Man 3 and Man of Steel are also set to use the innovative technology to simulate the perfect blasts, and it is fast becoming the industry standard for studios.
"The basic idea was to make smoke look more realistic. Car and chimney exhausts, volcanoes, blasts and big explosions in animated scenes are all difficult to capture through conventional simulations," says Thürey. This is mainly because turbulence is irregular, dissipating and chaotic, and characterised by random swirls and eddies. Capturing how the flow of gas transitions from smooth flow to choppy turbulence is difficult to recreate via fluid simulations. Animated computer fluid simulations are also time and memory intensive especially if you want to create something with a complicated motion. "Any smoky explosion will create a huge amount of swirls with very fine detail. Even if you use a brute force simulation and a very fast machine, it can take days. But when film directors opt to use computer graphics they expect the process to be, not only fast and cheap, but highly controllable. Directors want to be able to specify the timing of an explosion, where it reaches first, as well as other parameters such as shape and magnitude. However, doing this with a traditional fluid simulation solution can be a case of trial and error," says Thürey.
However, the wavelet algorithm applies a two-step process. It enables large- and small-scale detail to be edited separately, allowing high-resolution detail to be added as a post-processing step. This basically allows animators to set up a low-resolution simulation quickly that can be approved by a film's director, and later add details without changing the overall fluid motion. "First, you carry out a small, coarse, low-res simulation that you can play around with until it gives you the motion that you would like, and you can rely on this motion being consistent with the next step where you can add more details. This was not possible with the conventional fluid simulation graphics where if you change the resolution, that typically changes the underlying physics (for example the viscosity of the liquid)," says Thürey.
The developers first elected to publish and showcase their work at ACM SIGGRAPH- the leading conference for computing graphics. Their main aim was to make the code open source, so as to make it easy for others to use and reproduce it. "People can easily incorporate into their pipelines and existing simulators. Open source definitely makes it much more reproducible," says Thürey.
Gross, who heads the Disney Research Lab in Zürich (established after the 'Wavelet' algorithm was developed and remains the only Disney laboratory outside of the US), is convinced that the award shows how important scientific publications are when it comes to propagating innovations in industrial applications. Since as far back as 1931, 'The Academy' has recognized the crucial role of science and technology in the moviemaking process. However, Gross and Thürey were certainly not expecting a 'Tech Oscar' for their work. The next research area for the Disney Research Zurich Animation Group will be improving methods for creating turbulence in fluid simulation, which Thürey predicts will offer even more of a challenge.