Light in Power: A General and Parameter-free Algorithm for Caustic Design

Quentin Mérigot, Jocelyn Meyron, Boris Thibert

Abstract

We present in this paper a generic and parameter-free algorithm to efficiently build a wide variety of optical components, such as mirrors or lenses, that satisfy some light energy constraints. In all of our problems, one is given a collimated or point light source and a desired illumination after reflection or refraction and the goal is to design the geometry of a mirror or lens which transports exactly the light emitted by the source onto the target. We first propose a general framework and show that eight different optical component design problems amount to solving a light energy conservation equation that involves the computation of visibility diagrams. We then show that these diagrams all have the same structure and can be obtained by intersecting a 3D Power diagram with a planar or spherical domain. This allows us to propose an efficient and fully generic algorithm capable to solve these eight optical component design problems. The support of the prescribed target illumination can be a set of directions or a set of points located at a finite distance. Our solutions satisfy design constraints such as convexity or concavity. We show the effectiveness of our algorithm on simulated and fabricated examples.

Introduction

In the field of non-imaging optics, "caustic design" refers to the inverse problem of shaping a surface—be it a mirror or a lens—to redistribute light from a source into a specific, often complex, illumination pattern on a target screen. Historically, this has been a challenging task. Existing methods frequently suffer from being overly specialized to a single optical setup (e.g., only collimated light), requiring extensive manual parameter tuning, or failing to produce surfaces that are easily manufacturable.

This paper introduces a unified, parameter-free framework that addresses these challenges. By mapping eight distinct optical design problems to a single mathematical structure based on optimal transport and 3D Power diagrams, the authors provide a robust numerical solution capable of handling high-contrast images and generating physically viable components.

Methodology: A Unified Framework

The core innovation of this work is the demonstration that eight different design problems—varying by component type (mirror vs. lens), source type (collimated vs. point), and surface geometry (convex vs. concave)—all share a common underlying structure: a generalized Monge-Ampère equation.

Optimal Transport and Power Diagrams

The authors frame light redistribution as a semi-discrete optimal transport (OT) problem. The light source is treated as a continuous measure, while the target is represented as a set of discrete points. A key technical leap is the realization that the "visibility cells"—the specific regions of the optical component that direct light to a target point—can be computed by intersecting a 3D Power diagram with a planar or spherical domain. This geometric approach allows for high efficiency by leveraging established computational geometry tools.

Figure 3: The algorithm can handle complex, multi-channel target distributions by treating light transport as a geometric mapping problem.

Numerical Solution

To solve the resulting energy conservation equation, the framework employs a damped Newton algorithm. Because the problem is formulated via OT, the Jacobian is symmetric and negative-definite, ensuring fast, quadratic convergence. Unlike previous stochastic or heuristic methods, this approach requires no manual tuning parameters. For "near-field" problems where the target is at a finite distance, an iterative scheme is used to repeatedly update target directions until convergence.

Key Results and Physical Validation

The algorithm demonstrates exceptional performance and versatility. It can solve for a 1-million-pixel target image in approximately 26 minutes for far-field targets on a standard laptop, typically reaching a solution in just 10 to 20 iterations.

Handling High Contrast and Design Flexibility

Because the method is grounded in measure theory, it naturally handles "black" regions (zero light intensity) without numerical instability; the algorithm simply assigns zero area to the corresponding cells. Furthermore, the framework allows designers to explicitly choose between convex and concave solutions, which is critical for manufacturing constraints. For example, convex mirrors are often easier to chrome-plate, while concave surfaces may be required for compact housing.

Figure 5: Comparison of simulated results (left) and physical prototypes (right). The algorithm successfully translates complex grayscale images like the "Train" into manufacturable optical components.

Physical Fabrication

The authors validated their algorithm by milling physical prototypes from PMMA (lenses) and aluminum (mirrors). After manual polishing, these components successfully projected high-fidelity images, such as the detailed "Train" photograph and the "Hikari" kanji character, proving the algorithm's real-world viability.

Conclusion and Future Work

While the current framework assumes idealized point or collimated sources, it represents a significant step forward in general-purpose caustic design. Future research may address "extended sources" (like physical LED filaments) that cause blurring, as well as integrating manufacturing constraints—such as the precision limits of CNC milling—directly into the optimization process to reduce the need for manual post-processing.

Summary updated

Abstract

We present in this paper a generic and parameter-free algorithm to efficiently build a wide variety of optical components, such as mirrors or lenses, that satisfy some light energy constraints. In all of our problems, one is given a collimated or point light source and a desired illumination after reflection or refraction and the goal is to design the geometry of a mirror or lens which transports exactly the light emitted by the source onto the target. We first propose a general framework and show that eight different optical component design problems amount to solving a light energy conservation equation that involves the computation of visibility diagrams. We then show that these diagrams all have the same structure and can be obtained by intersecting a 3D Power diagram with a planar or spherical domain. This allows us to propose an efficient and fully generic algorithm capable to solve these eight optical component design problems. The support of the prescribed target illumination can be a set of directions or a set of points located at a finite distance. Our solutions satisfy design constraints such as convexity or concavity. We show the effectiveness of our algorithm on simulated and fabricated examples.
BibTeX
@article{merigot2018light,
  author = {Quentin Mérigot and Jocelyn Meyron and Boris Thibert},
  title = {{Light in Power: A General and Parameter-free Algorithm for Caustic Design}},
  journal = {ACM Trans. Graphics},
  year = {2018},
  url = {https://openreview.net/forum?id=QvttKQrpl0},
}
Figure