This article is also available in:
Français
On Sunday at SIGGRAPH 2023, a conferenced focused on cloth took place: Tailor Made: Techniques in Computational Cloth. Several studios and companies showcased their latest projects.
Animal Logic, showed us a tool that allow their artists to efficiently generate knitted garnments. They are able to create yarn-level models for a large number of characters.
The Smithonian and Adobe then explained how they created and deployed a capture methodology to create a digital double of Alan Shepard’s Space Suit. They told us about their capture setup, as well as about their new hybrid mesostructure reconstruction approach. Basically, they combined low-frequency data from photogrammetry and fine details/high-frequency data from a neural method. This allowed them to create a jaw-dropping digital double, that can be seen in the video below.
Adidas told us about AI Archive, a generative AI that was created in order to generate new sneakers designs. They explained how they managed to provide a good level of control over the generated results: this was a key part of the project, as boosted adoption of this new technique.
Last, but not least, Disney told us about the garnments created for their animated feature Strange World. This technical challenge stemmed from a creative need: since Avalonia is a low-tech location, costumes had to convey this fact with a handmade look and feel. The team explained how they revamped their authoring workflows for curve-based fabrics. This allowed them to handle complex woven patterns as well as finishing techniques like stitches and hems, while still providing artist-friendly tools to the team.
Overall, these presentations show that cloth is an area where lots of work, research and development is still being done, whether it’s for capturing real-life garnments or to create new ones.
