Physics-Informed 3D Shape Generation and Simulation

The field of 3D shape generation and simulation is advancing rapidly, with a growing focus on incorporating physical properties and constraints to enhance realism. Researchers are exploring new methods to integrate physics-based guidance into generative models, allowing for more accurate and realistic shape synthesis. This trend is evident in the development of surface-based frameworks for aerodynamic simulation, which enables efficient and fine-grained interaction between surfaces and fluids. Another area of innovation is the reconstruction of physically consistent, kinematic part-level articulated objects from single RGB images, which has significant implications for embodied AI, robotics, and interactive scene understanding. Noteworthy papers include: PhysGen, which proposes a unified physics-based 3D shape generation pipeline, and Gaussian Swaying, which presents a surface-based framework for aerodynamic simulation using 3D Gaussians. SPARK is also notable for its framework for reconstructing physically consistent, kinematic part-level articulated objects.

Sources

PhysGen: Physically Grounded 3D Shape Generation for Industrial Design

Gaussian Swaying: Surface-Based Framework for Aerodynamic Simulation with 3D Gaussians

SPARK: Sim-ready Part-level Articulated Reconstruction with VLM Knowledge

Prompt2Craft: Generating Functional Craft Assemblies with LLMs

Order Matters: 3D Shape Generation from Sequential VR Sketches

ShadowDraw: From Any Object to Shadow-Drawing Compositional Art

Built with on top of