The Layered Clothing system makes clothes and accessories fit any avatar.
Roblox announced the release of its Layered Clothing system, which enables realistic clothing and accessories to fit any avatar body type. The players can create custom clothes and accessories and dress Roblox avatars in up to 6 layers of clothing.
The users can also create their own bodies, which was a great challenge for the developers as the new system needs to fit clothes to any possible body shape.
"The magic of the clothes in this system is that they can stretch to fit any platform character – from a Classic Blocky all the way to a T-Rex – and it fits nicely on top of multiple layers a character is already wearing."
Surprisingly, the system was inspired by rocket science.
"Computationally, when you have two arbitrary 3D meshes, it’s difficult to compute exactly how to fit one over the other without distorting the content or making it simply look ugly," said Sergey Makeev, the company's Senior Technical Director. "Part of the inspiration came to the team while reading several scientific papers on aerodynamic testing of rockets."
The team observed the way airflow envelopes a rocket in flight and how small adjustments to the rocket body shape affect airflow and flight efficiency. Then, they looked at automated methods that create new and improved rocket shapes in 3D CAD and applied that thinking to layering of meshes on top of each other.
Eventually, the developers considered the abstraction layer between the two meshes, "introducing a new cage layer acting as an outer boundary of the underlying body structure, then interacting with the inner cage of the shirt getting layered on top of the body."
"We found a way to wrap every vertex of one mesh around another, their dimensions governed by the behavior of the abstraction layer. This approach lets users stack items in real time and see the final layering instantly."
The system is meant to set the stage for Roblox's future technologies, like User Generated Bodies, Dynamic Heads, and Machine Learning-driven creation systems.
Check out the announcement here and don't forget to join our new Reddit page, our new Telegram channel, follow us on Instagram and Twitter, where we are sharing breakdowns, the latest news, awesome artworks, and more.