Physically Compatible 3D Object Modeling from a Single Image

We present a computational framework that transforms single images into 3D physical objects. The visual geometry of a physical object in an image is determined by three orthogonal attributes: mechanical properties, external forces, and rest-shape geometry. Existing single-view 3D reconstruction methods often overlook this underlying composition, presuming rigidity or neglecting external forces. Consequently, the reconstructed objects fail to withstand real-world physical forces, resulting in instability or undesirable deformation -- diverging from their intended designs as depicted in the image. Our optimization framework addresses this by embedding physical compatibility into the reconstruction process. We explicitly decompose the three physical attributes and link them through static equilibrium, which serves as a hard constraint, ensuring that the optimized physical shapes exhibit desired physical behaviors. Evaluations on a dataset collected from Objaverse demonstrate that our framework consistently enhances the physical realism of 3D models over existing methods. The utility of our framework extends to practical applications in dynamic simulations and 3D printing, where adherence to physical compatibility is paramount.

Authors

• Minghao Guo
• Bohan Wang
• Pingchuan Ma CompVis @ LMU Munich, Munich Center for Machine Learning
• Tianyuan Zhang SKLCCSE, Beihang University
• Crystal E. Owens
• Chuang Gan University of Massachusetts at Amherst
• Joshua B. Tenenbaum MIT Computer Science and Artificial Intelligence Laboratory
• Kaiming He
• Wojciech Matusik

Keywords

No keywords are indexed for this paper.