← ArchivePaper2015
Sketch to Pose in Pixar's Presto Animation System
Ryan Stelzleni, Bret Parker, Tom Hahn, Sarah Shen, Dan McGarry, Chen Shen
Stroke-based posing tool that snaps multiple joints to a drawn silhouette, used heavily on Inside Out for character and cloth-simulation results tuning.
Abstract
This Pixar talk presents a sketch-based interface in the Presto animation system that lets animators pose multiple joints at once by snapping them to a user-drawn stroke. A multiple-joint guide depicts joints as dots connected by a centerline that is implicitly generated from the character rig and transformed to match the drawn stroke. Special rig-aware solvers detect which guides are displayed and write only into the same parameters those guides would use, so animators can freely switch between sketching and direct manipulation. Three stroke conversion schemes are provided, including flattening the model into a camera-parallel plane, moving each articulation point to touch the stroke in screen space while preserving world-space distance, and interpreting strokes within an automatically chosen XY, YZ or XZ guide plane.
How to read this
- Category
- Production talk: sketch-based posing in Pixar Presto
- Contributions
- Demonstrates a stroke-based interface in Presto that poses multiple joints at once by snapping them to an animator-drawn stroke.
- Shows a multiple-joint guide (dots on an implicitly generated centerline) plus rig-aware solvers that write only into the parameters those guides use, allowing free switching between sketching and direct manipulation.
- Presents three stroke-conversion schemes: camera-parallel flattening, screen-space touch with world-space distance preserved, and interpretation within an auto-chosen XY/YZ/XZ guide plane.
- Context
- A production application of sketch-based posing, conceptually related to learned-prior posing such as Grochow et al.'s Style-Based Inverse Kinematics but driven by rig-aware solvers, used heavily on Inside Out.Builds on: Style-Based Inverse Kinematics
- Correctness
- Studio practice rather than peer-reviewed; the approach is production-proven on a feature film, so judge it by workflow value and integration with the rig, not formal evaluation.
- Clarity
- Highly accessible; a single read conveys the interaction model, with little need for a deeper formulation pass.
- How to read it
- Read once for the interaction design and the three stroke-conversion schemes; useful as inspiration for posing UX rather than as an algorithm to reimplement.
Builds on
Built upon by
Nothing yet.
Related work
- Wig Refitting in Pixar's Inside Out 2 2024 / SIGGRAPH
- Direct manipulation of blendshapes using a sketch-based interface 2018 / Web3D
- SkinMixer: Blending 3D Animated Models 2022 / SIGGRAPH Asia
- Transposition Based Blendshape Direct Manipulation 2017 / GRAPP
Keywords
This page summarises the entry and links to its original source. The archive never hosts or redistributes the publication itself.Show it in the full archive list →