Rig Inversion by Training a Differentiable Rig Function

TL;DR

This paper introduces a novel approach to rig inversion by training a differentiable rig function using a neural network, enabling more accurate and efficient rig parameter estimation from input meshes.

Contribution

The paper presents the first method to learn a differentiable rig function with a neural network, facilitating improved rig inversion techniques.

Findings

Differentiable rig function enables gradient-based optimization.

Neural network approximation improves rig inversion accuracy.

Method demonstrates potential for real-time rig parameter estimation.

Abstract

Rig inversion is the problem of creating a method that can find the rig parameter vector that best approximates a given input mesh. In this paper we propose to solve this problem by first obtaining a differentiable rig function by training a multi layer perceptron to approximate the rig function. This differentiable rig function can then be used to train a deep learning model of rig inversion.