JL: Exactly, growing trees and things like that is still extremely difficult. On the rendering side, we did use a lot of measured data to figure out what some of these things wanted to be: matching real foliage and then adapting that to get the look of the alien foliage. That was all a step in the right direction, but a lot of that had to be done by hand, sorting through a lot of it and figuring out what it all meant, try to collate it and relate one piece to the other and just dealing with the complexity of the organic growth you get with plants. You would seed the ground with different plants; they'd have different growth rates and compete for resources (sunlight and terrain). So smaller plants initiated bigger plants and they would die off and you'd get this nice, natural growth pattern. But we didn't do that to the point where we actually grew each of the individual plants as this growing was happening. What we'd do was just pop them in like pre-built models at their specific age of development. The next step would be to actually grow the plants. Would that give us a more interesting variety? Is that even necessary in a jungle where you don't see everything? Still, it's an interesting problem and there's going to be a need for that solution somewhere along the line.

BD: And the virtual art department?

JL: I think what we want to be able to do there is have just a little tighter integration with the art department, the virtual art department and the post-production you get into a tricky area that has to be managed in between all of this because when you're creating all of the virtual art department assets, considering that you're shooting on a stage and everything is live action, you get a photographic plate, you get your shot turned over to you, you get your element, you just start working on it. If you need to cast shadows or something on that element, you take measurements, you rebuild the geometry and you match it as closely as you can. When you're dealing with a virtual art department, the assumption is those assets are the same. And so you need to know when, in fact, they are the same and when they're not because often times they're lower resolution because of the necessity of getting realtime playback. Other times, they may be changed or if there is something on the set, [it's hard] just keeping track of all that information because you go from having this live-action shooting style where you're recording camera takes and all of the information just as if you were shooting it live action, but then you immediately turn it around into shots because you have to render it for editorial purposes, and so that whole line between the two is very blurred. And it was kind of fluid the way we were doing it, experimenting with different ways that worked. But I think we'll need to talk about how we can nail it down with one way of working.

BD: What about stereoscopic improvements?

JL: For Avatar, we didn't really have too many problems with stereo. But there is one overriding technical problem if you're shooting a lot of plate work and doing any kind of reconstruction or paint out and so forth. That's still a difficult problem to do in stereo, and that's something that's on our list of R&D projects.

BD: It's been a great year for visual effects overall. What are you impressions of the other sci-fi nominees?