PoseVR
After developing Vitruvian at university, I was invited by Walt Disney Animation Studios to further develop the virtual reality animation concept as a part of their Summer 2017 internship program. Consulting frequently with my mentors and WDAS animators, I developed PoseVR over the course of three months.
Why animate in VR?
Stereoscopic Display
By projecting three dimensions onto two, computer monitors kill our best indicators of depth: binocular disparity and convergence. This forces us to rely on objects' scales and interposition to plot out our z axis. While the final shot will be flattened upon rendering, giving an animator a realistic sense of depth at all times prevents the mistakes that can easily occur from relying upon one camera.
6 Degrees of Freedom
Typical mouse-and-monitor setups require animators to perform several actions to complete even simple tasks. While interfaces like control pickers and multi-manipulators have been applied to speed things up, countless wasted clicks and seconds pile up when switching between camera views, tools, and windows.
PoseVR allows users to directly manipulate individual bones and controls using their built-in kinesthetic intelligence, as if playing with an action figure. Depth is built into the cursor, and you can perform all your translation and rotation in one motion instead of (up to) six. With PoseVR you can use coarse controls to reach that pose in an instant, or constrain your axes to fine tune close frames.
Features
While many features of the classic Autodesk Maya could be translated into VR with little effort or novelty, the purpose of my internship was to produce a minimum viable animation tool, then experiment with features that animators expected could be abridged or improved in the new medium. Below is a list of the most interesting such features.
World Navigation
Because PoseVR's users are expected to be sitting comfortably, they must have a quick and intuitive way to move the world around them. Using the grip buttons on their Oculus Touch or Vive controllers, users simply grab the world to pull and turn it wherever they'd like. If they grab it with two hands, they can even scale it to work in detail. In the demonstration below, the animator makes only minor changes to Kristoff's pose, but drastically change their viewing angle to make sure their pose is right.
Forward, Inverse, and... In-Between Kinematics
Because many of W.D.A.S.' animators take different routes to hit a pose, suggestions for what happens when you grab the model were many and varied. At this level of quality rigs are immensely complex, so most users will end up using a mix of these manipulation styles for different situations.
Pure FK "Screwdriver" mode gives animators direct control of a bone's rotation, only affecting the translation and rotation of that bone's children. It's especially useful when applying the axis isolation mentioned below, and on high-detail jobs like the fingers.
Inverse Kinematics can be set up on any model within Unity using a wizard I created for the job.This is always on, and will preserve the location of the target as closely as possible unless a bone within the IK chain is directly selected for manipulation.
As per the request of W.D.A.S. animators, limbs use a pole vector control while the pelvis and chest controls create a Bezier curve that allow for unconstrained squash and stretch.
The apparent favorite mode of manipulation for this environment was somewhat novel. The animator is given control of the translation and rotation of the selected bone, at least as far as the parent bone can accommodate. This controls two bones at a time, and produces an action-figure-like mechanic that feels very intuitive in VR space:
Timeline Editor
The timeline is rendered as a tangible line upon which are placed tangible red 'coins' representing keyframes. These coins can be moved, copied and pasted, or deleted at will. Users can constrain the timeline to a specific playback range using the gray end caps: this will display both the selected range and the global timeline on separate lines.
Shot Camera
At Walt Disney Animation Studios, the final product is almost always a 2D rendering. PoseVR therefore features a movable shot camera that renders to an adjustable screen in world space. The camera can also be animated for moving shots.
Interpolation Methods
PoseVR allows users to switch between stepped playback for a stop-motion style as well as smooth bicubic interpolation playback to smooth movements and bridge between distant keyframes.
Axis Isolation
Users may hold a shortcut key to only translate using the joint- or world-aligned axis along which the most movement has taken place. Another shortcut performs the analogous behavior for rotation.
