π€ AI Summary
Existing 3D asset generation methods primarily focus on static geometry and struggle to incorporate the physical and functional properties required for interactive applications. To address this limitation, this work proposes PhysForge, a two-stage decoupled framework: first, a vision-language model generates a hierarchical physical blueprint encoding material, functional, and kinematic constraints; then, a physics-driven diffusion model, enhanced with a novel KineVoxel Injection mechanism, jointly synthesizes high-fidelity geometry and precise motion parameters. This study pioneers the integration of functional logic and hierarchical physical modeling into 3D generation, introducing PhysDBβthe first large-scale dataset annotated with four-level physical semantics. The resulting assets exhibit functional plausibility and are directly simulation-ready, offering a high-quality data engine for interactive virtual environments and embodied intelligence.
π Abstract
Synthesizing physics-grounded 3D assets is a critical bottleneck for interactive virtual worlds and embodied AI. Existing methods predominantly focus on static geometry, overlooking the functional properties essential for interaction. We propose that interactive asset generation must be rooted in functional logic and hierarchical physics. To bridge this gap, we introduce PhysForge, a decoupled two-stage framework supported by PhysDB, a large-scale dataset of 150,000 assets with four-tier physical annotations. First, a VLM acts as a "physical architect" to plan a "Hierarchical Physical Blueprint" defining material, functional, and kinematic constraints. Second, a physics-grounded diffusion model realizes this blueprint by synthesizing high-fidelity geometry alongside precise kinematic parameters via a novel KineVoxel Injection (KVI) mechanism. Experiments demonstrate that PhysForge produces functionally plausible, simulation-ready assets, providing a robust data engine for interactive 3D content and embodied agents.