Existing AI image and video models allow for rapid iteration from text or concepts to images and videos but can only produce 2D outputs allowing for limited downstream interaction and editing. World models such as World Labs’ Marble can create complete and visually rich 3D environments in minutes from just a handful of real images, concept sketches, 360 panoramas, or a 3D layout. This session will walk through a variety of workflows combining AI image models and Marble together to rapidly build, edit, and visualize 3D worlds, as well as ways to export the resulting 3D Gaussian splat files into other tooling for further iteration.
