My first major piece of work for the playset was rigging up the ATAT used for both cinematic sequences and for "training" the physically driven animation runtime solution developed internally. As I was the one and only technical artist at the studio, and since my skill set includes rigging, I was tasked with getting the model (created by Rob Griffin, a member of the Gobo art team) rigged and functional for use in engine.
Once of the biggest problems faced was constructing a rig that would function in a manner accurate to the film. The behavior of the rig had to maintain a very straight leg throughout the range of movement. In order to achieve this I had to make the leg compress slightly to avoid the knee bending. Maya also has a problem with its internal IK solver where the output is prone to "popping" as the chain approaches is maximum extension. To counter this, I implemented a soft-IK solution via expressions. This works by scaling the leg joints along their forward vector very slightly as the IK solution converges towards the maximum leg length (perfectly straight). This has the overall effect of damping the IK and works incredibly well.
In the video below, the rig on the left has the soft-IK solution disabled and you can see the knee's pop un-desirably. On the right, the soft-IK solution is active and the knees transition from slightly bent to perfectly straight with no acceleration or popping. The foot impact is intentionally quite aggressive to convey the sense of weight.
The problem is more apparent is this next video demonstrating the full range of rig control. From 50 seconds onward I demonstrate the soft-IK solution, off initially and then active. You can clearly see the difference in the behavior of the knee.
I also rigged up an ATST to be used as a simple player mountable vehicle, as well as an AI controllable mount leveraging the same PBA tech used by the ATAT but limited to collision and damage reactions. The following video shows the rig in Maya that the animators would use to create animation data to train the PBA system.
Given the lack of support in engine to handle tracked vehicles, and the desire to introduce the Iconic Sand Crawler into the game, we had to come up with a solution to realistically animate the Sand Crawler to appear to follow the terrain on which is was placed as it moved around. What I eventually built was a ribbon of geometry that conformed to the underlying level geometry, that could be manipulated and would automatically update the motion of the Sand Crawler. It turned out to be quite a complex rig, but I successfully built a solution that had the Sand Crawler follow a pre-determined path (using a spline) while at the same time conforming to the ribbon and appearing to follow the ground in a realistic way. I even managed to get the whole thing to stay relatively balanced since the player would be able to stand in and on the Sand Crawler and the underlying physics system required stability. The following 4 videos show my rig in operation within Maya. I provided the animators a "one-click" bake and export operation to keep the workflow as streamlined as possible, as well as a convenient method to switch the rig from one path and associated ribbon to another to account for a bigger change to the topography of the level. The 4th video shows animations for an early prototype for interactivity on the Sand Crawler.
Here is the Sand Crawler post-bake and export running in engine, in a very early work in progress version of Tatooine in Infinity 3
I also got some time between tasks to model a toy-version of a speeder bike. It was later finished by one of Gobo's core vehicle artists
We've also been exploring animation styles for the upcoming project, trying to find a nice blend of mechanical and "playful". This is a quick pre-vis I did for a deployable turret - focusing purely on motion.