“Creating a 2D Data Format Capable of Handling Complex Transparencies” by Offertaler, Verlander, Opstal, Mahovsky, Lee, et al. …
Conference:
Type(s):
Title:
- Creating a 2D Data Format Capable of Handling Complex Transparencies
Session/Category Title:
- Painting the Wild Robot
Presenter(s)/Author(s):
- Eszter Offertaler
- Sondra Verlander
- Baptiste Van Opstal
- Jeff Mahovsky
- Ashley Lee
- Jeffrey (Jeff) Budsberg
Moderator(s):
Abstract:
The Wild Robot’s artistic style called for painterly assets to transport the audience to a lush forest never inhabited by humans juxtaposed against the technology rich world of the humans. Soft edges and smearing of assets were needed to emulate a painter’s hand in our assets which required varying amounts of transparency. Traditional flat data channels used for things like positions and normals were not compatible with assets that use non-binary transparency, creating an incompatibility in compositing. Crypto-Deep data is an extension to Cryptomatte that stores extra, mostly geometric, data in addition to the asset id and pixel coverage. It helps address compositing artifacts seen in signals generated from flat data and is a must for non-binary presence, while remaining more lightweight than a fully deep compositing workflow. We’ve developed a new suite of crypto-deep aware nodes to supersede our previous rendered data-driven tools.
