Introducing Meta 3D AssetGen 2.0: A new foundation model for 3D content creation

Today, we’re introducing Meta’s latest 3D foundation model, AssetGen 2.0, which was designed to create high-quality 3D assets from text and image prompts. AssetGen 2.0 consists of two models developed within Meta: one to generate 3D meshes and another to generate textures. Building on its predecessor, AssetGen 1.0, our latest innovation represents a significant leap forward in 3D generative AI research, leveraging a single-stage 3D diffusion model to deliver 3D meshes with dramatically improved detail and fidelity, while our complementary TextureGen model ensures that these assets are not only visually stunning but also production-ready with high-quality textures. Our single-stage diffusion approach ensures that the 3D assets have geometric consistency with very fine details (see outputs below). Unlike AssetGen 1.0, AssetGen 2.0 uses 3D diffusion for geometry estimation and is trained on a large corpus of 3D assets. For texture generation, it also introduces new methods for improved view consistency, texture in-painting, and increased texture resolution. We’re currently using AssetGen 2.0 internally for 3D worlds creation, and we will roll out later this year to Horizon creators.

We expect AssetGen 2.0 to help democratize the creation of 3D experiences, making it as accessible as 2D content creation and opening up new creative possibilities for artists, designers, and developers working on Meta’s Horizon and Avatar platforms. Further, in the coming months, AssetGen will serve as a foundational building block to enable auto-regressive generation of entire 3D scenes, enabling a step function change in creation for our users. This new model will be able to generate complex 3D environments by sequentially generating individual objects, structures, or elements using simple text or image prompts.

Meta is committed to pushing the boundaries of what’s possible in the field of generative AI. AssetGen 2.0 has been designed to deliver stunning visual quality, surpassing existing solutions and setting a new standard for the industry. As we continue to explore the vast potential of 3D generative AI, we’re excited to see how creators will leverage AssetGen 2.0 in the future to bring their most ambitious ideas to life and redefine 3D content creation.

Join us on this exciting journey as we unlock the full potential of 3D GenAI and empower a new generation of creators to build, innovate, and inspire. We are hiring for multiple roles including Research Scientists and Engineers who can help us build this future.