Towards Affordance-Aware Articulation Synthesis for Rigged Objects

  • 2025-01-21 18:59:59
  • Yu-Chu Yu, Chieh Hubert Lin, Hsin-Ying Lee, Chaoyang Wang, Yu-Chiang Frank Wang, Ming-Hsuan Yang
  • 0

Abstract

Rigged objects are commonly used in artist pipelines, as they can flexiblyadapt to different scenes and postures. However, articulating the rigs intorealistic affordance-aware postures (e.g., following the context, respectingthe physics and the personalities of the object) remains time-consuming andheavily relies on human labor from experienced artists. In this paper, wetackle the novel problem and design A3Syn. With a given context, such as theenvironment mesh and a text prompt of the desired posture, A3Syn synthesizesarticulation parameters for arbitrary and open-domain rigged objects obtainedfrom the Internet. The task is incredibly challenging due to the lack oftraining data, and we do not make any topological assumptions about theopen-domain rigs. We propose using 2D inpainting diffusion model and severalcontrol techniques to synthesize in-context affordance information. Then, wedevelop an efficient bone correspondence alignment using a combination ofdifferentiable rendering and semantic correspondence. A3Syn has stableconvergence, completes in minutes, and synthesizes plausible affordance ondifferent combinations of in-the-wild object rigs and scenes.

 

Quick Read (beta)

loading the full paper ...