VFX Oscar Bakeoff 2010: Seeing is Believing
For 2012, Volker Engel and Marc Weigert described Roland Emmerich's plan "to make a disaster movie to end all disaster movies" with the help of their Uncharted Territory, Scanline VFX, Digital Domain, Double Negative, Sony Pictures Imageworks, Pixomondo, Hydraulx, Crazy Horse Effects, Evil Eye Pictures and Gradient Effects. "Since we were shooting in Vancouver, we had to create via visual effects Los Angeles, Tibet, Yellowstone, Las Vegas, Washington, Hawaii and Rome -- and then destroy them," Engel explained. "We had tsunamis, earthquakes, volcanic eruptions, ash clouds and so on. Number two is the level of detail that we needed for all of this. For example, for the earthquake, it's not enough to create a photoreal environment, but we also needed to make it shake and break. That means you can't just build parts of houses, you have to build interior walls, roof structures, chimneys of single bricks, roofs of separate shingles, furniture inside, insulation material… rig it and sim it based on all the different material parameters. Yellowstone, for instance, we had hundreds of layers of rippling and breaking ground, falling chunks, trees, smoke trails, lava, dust and fire. The water masses later in the Himalayas, had to [attack] ships and buildings, plants, air craft, mountains [with] hundreds of layers of white water, foam, spray, mist. The third point was the choreography of the visual effects. As you know, if you do physics-based simulations, if you give the computer the right parameters you can pretty much get what reality would do, but, unfortunately, that wasn't what we want because we need to tell a story. So there's a constant balance between film time, which is always faster, and perceived reality and scale that you want to achieve…"
Letteri said, "Avatar is the story about learning to see a world that we don't understand and when we started we had two overriding ideas that we wanted to go with: One is wanted to see the world through the eyes of our characters; and the other is that we wanted to point the camera anywhere in the world and say, 'Let's go over there and shoot.' And what it really came down to is that Jim wanted to erase the boundaries between visual effects and live-action filmmaking. And he had some very specific ideas how to do that and, of course, he also wanted to do it in 3-D -- he wanted it to be a very immersive experience. So, right off the bat, the idea was to build a performance capture stage where you could work out ideas and take them either in a live-action direction, if need be, or completely digital if that would be the answer as well… The character designs were based very much on the actors performing them, especially around the jaw line and lips, because we thought having a really good guide and a really good match to the dialogue would help us a lot…and what you see on Pandora is largely digital and that was dictated largely by the necessity of having flexibility and the necessity of doing 3-D… We used a number of techniques to really try to bring the characters alive… We had naked characters and really wanted to get the body muscles as correct as possible, so we tried to get the biomechanics correct. We worked out new lighting models: we used spherical harmonics and image-based lighting -- HDRI -- to try and integrate lighting over a complete world but also to see that world reflected in a close-up of the eyes of the characters…"