Avatar The Way of Water VFX Breakdown: Dive in with Wētā FX

so when reading the script for Avatar the way of water the two sort of standout things sort of lifting off the script Pages were the amount of water that we were going to have to do and then the second is the amount of Photography plate based shots from my perspective was confident that we could do high fedelity CG water we're doing all of the research for years to do it but the difficult thing or the thing that makes that even harder is that there's plate based photographic water that needs to

be merged with CG water in Stereo and it's really unforgiving water is tough because you see down into it through refraction you see off it through reflection so 10 VFX soups just over 200 compositors it's 101 lighting TDS and 92 production staff and the final disc footprint was around 182 pedabytes so we're using Foundry products through our pip I'm using Mari for uh texturing our assets and authoring our assets Katana for lighting the 3200 shots in the show and finally nuke for bringing it all

together we've been relying on nuke as a Workhorse to do everything in deep compositing and we' sort of leveraged it even further to do compositing in reflected and refracted spaces extending the Deep tool set further Mar is our primary texturing tool and we use it to texture all of our assets and allows us to work with very high resolution Maps some of our creatures in the case of Pyon large whale likee creature need to be dressed for camera you know very close up on all along its body and als

o work in the wide and Mari allows us to work at that level of detail across an entire asset one of the things that the team's really loving that's new and Mar is the node-based workflow it's non-destructive allows us to work procedurally versus the destructive layer Paradigm and we can provide artists with those procedural set that give them a great starting point on an asset the first version of our comps is actually run by the lighting TS all of the render passes that they're spitting out fro

m Katana via our render Manuka there are comps set up to catch that data and automatically deep comp it through all of our lens treatments Etc and produce first versions essentially without compositors working on the shop and once the lighting starts getting a bit more liked by everybody then the compers join the party and they start to refine it as compositors do so we knew with such a large number of shots it's uh 3200 shots that we really wanted to keep things templated we have very large and

very dense layouts and they're sort of populated a lot via instancing for example the me a Village it's 66 billion instances of 2300 unique assets down to the leaves on the trees all of the water is really large simulations you know and you got people riding elu and so we use Katana to hide that complexity we can work on those assets at a very high level and worry about the lighting not about managing the scene complexity and shuttling things in and out and saving scenes so we generally work at

a sequence level make shot level lighting adjustments in our templates we won't be diving all the way into you know a single leaf or a gourd or a piece of fruit in the village but it's all there in the layout working on of the way of water for such a long period of time it's not just a film it's like a chapter in our lives you know 5 to 7 years for some of us working with such a great team at wther everybody's you know running at the top of their game it's just been [Music] fantastic