Category-Agnostic Neural Object Rigging

TL;DR

This paper introduces a data-driven, category-agnostic method for representing and manipulating deformable 4D objects using a novel sparse blob-based encoding, enabling intuitive pose control without category-specific expertise.

Contribution

It proposes a new low-dimensional, data-driven representation for deformable objects that is scalable and category-agnostic, improving controllability and manipulation.

Findings

Effective manipulation of 3D object pose via blob parameters

Disentanglement of pose and instance information in 4D objects

Demonstrated generalization across multiple object categories

Abstract

The motion of deformable 4D objects lies in a low-dimensional manifold. To better capture the low dimensionality and enable better controllability, traditional methods have devised several heuristic-based methods, i.e., rigging, for manipulating dynamic objects in an intuitive fashion. However, such representations are not scalable due to the need for expert knowledge of specific categories. Instead, we study the automatic exploration of such low-dimensional structures in a purely data-driven manner. Specifically, we design a novel representation that encodes deformable 4D objects into a sparse set of spatially grounded blobs and an instance-aware feature volume to disentangle the pose and instance information of the 3D shape. With such a representation, we can manipulate the pose of 3D objects intuitively by modifying the parameters of the blobs, while preserving rich instance-specific information. We evaluate the proposed method on a variety of object categories and demonstrate the effectiveness of the proposed framework. Project page: https://guangzhaohe.com/canor