Modelling and Animating Hand Wrinkles

Xiaosong Yang, Jianjun ZhangLecture Notes in Computer Science 2005 · DOI · OpenAlex

Abstract

This paper presents a computationally efficient, non-physically based method for animating realistic hand wrinkles. By enforcing two key geometric constraints—volume preservation of the underlying flesh and incompressibility of the skin—the authors achieve visually convincing results suitable for real-time applications. The method utilizes Voronoi diagrams to define localized deformation regions and parametric "shape functions" to model wrinkle profiles, avoiding the high computational cost of full-scale physics simulations.

Introduction

In computer animation, wrinkles are essential for realism but difficult to simulate efficiently. A distinction is often made between fine-scale wrinkles, typically handled via static textures, and large-scale wrinkles, which are the deep, motion-dependent folds found on the palm and knuckles.

The primary challenge is balancing realism with performance. Physics-based methods like Finite Element Modeling (FEM) offer high accuracy but are computationally expensive and difficult to integrate into standard animation pipelines. Conversely, simple texture mapping fails to capture the 3D "bulging" and "dipping" geometry that occurs during joint movement. This research proposes a "middle ground": a parametric approach that remains physically plausible while running fast enough for interactive use.

Methodology

The proposed pipeline generates and animates wrinkles through several key steps:

1. Deformation Region Definition

The process begins with user-specified "wrinkle lines." To determine the area affected by these lines, the authors use Voronoi diagrams to partition the mesh and define offset curves around the lines. This creates a localized "deformation region" that can be merged into the character's base mesh, ensuring that global mesh refinement is unnecessary.

Deformation regions defined by offset curves around user-specified wrinkle lines.

2. The Shape Function and Physical Constraints

To model the cross-section of a wrinkle, a piecewise polynomial function (using cubic Hermite interpolation) is derived. This function describes the skin "dipping" at the wrinkle line and "bulging" on the sides. Instead of simulating forces, the model enforces two geometric rules:

  • Volume Preservation: The integral of the shape function is set to zero, ensuring the volume of the "bulge" equals the volume of the "dip."
  • Incompressibility: The arc length of the skin cross-section remains constant during animation.

3. Skeleton-Driven Animation

As a joint rotates, the system calculates a shrinkage/expansion factor based on the joint angle, which updates the wrinkle's width. Newton’s method is then used to numerically solve for the new wrinkle depth that satisfies the incompressibility constraint.

Results and Discussion

The core innovation of this work is the use of geometric proxies for complex physics. By maintaining constant surface area (arc length) and constant underlying volume, the system captures the most important visual cues of skin deformation without a full stress-strain simulation.

Integration of high-detail wrinkle geometry (a) into the base hand mesh to produce the final result (b).

The method is computationally efficient, requiring significantly fewer vertices than global subdivision methods. Because the depth calculation is reduced to a one-dimensional numerical search, the system achieves real-time performance. Qualitative results demonstrate successful "bunching" of skin on the palm and the stretching of wrinkles on the knuckles during finger extension.

Limitations

While effective, the method relies on manual setup, requiring an animator to draw wrinkle lines. Additionally, the animation logic assumes joints have one dominant rotational degree of freedom, which may limit its application to more complex joints like the shoulder. Future work could also address explicit collision handling for very deep skin folds.

Summary updated

BibTeX
@inproceedings{yang2005modelling,
  author = {Xiaosong Yang and Jianjun Zhang and Jian Jun Zhang},
  title = {{Modelling and Animating Hand Wrinkles}},
  booktitle = {Lecture Notes in Computer Science},
  year = {2005},
  doi = {10.1007/11428848\_25},
}
Figure