arXiv:2605.28763v1 Announce Type: new Abstract: Interactive 3D assets used in games and simulation are typically decomposed into specific semantic parts to support animation, physics, and scripted behaviors, yet most generative 3D models produce either monolithic meshes or arbitrary part decompositions that cannot be aligned with application-specific requirements. We present CubePart, a generative framework for open-vocabulary, part-controllable 3D mesh generation that exposes part structure as an explicit inference-time control signal. Given a global text prompt and a user-defined parts schem
Advances in generative AI, particularly in 3D modeling and natural language processing, are converging to enable more sophisticated and controllable content creation tools.
This development addresses a critical limitation in generative 3D models by offering application-specific control over part decomposition, which is essential for interactive simulations and game development.
Generative 3D models can now produce assets with user-defined, semantic part structures, moving beyond monolithic meshes or arbitrary decompositions.
- · Game Developers
- · Simulation Engineers
- · 3D Content Platforms
- · Metaverse Companies
- · Manual 3D Asset Retopologizers
- · Generic 3D Generative AI Lacking Part Control
Faster and more efficient creation of complex, interactive 3D assets for various digital environments.
Increased accessibility for non-expert users to generate sophisticated 3D content with precise functional attributes.
The acceleration of virtual world development and more realistic human-computer interaction within simulated environments due to highly controllable 3D objects.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.AI