Optimization of laser-induced printed colors for industrial applications

Raphaël Mermillod-Blondin, Nicolas Dalloz, Mathieu Hébert, Alain Trémeau, Rémi Emonet, Nathalie DestouchesEPJ 2025 · DOI

TL;DR

Researchers have developed a way to maximize the color range (gamut) of laser-printed images on plasmonic thin films—a technology vital for high-end security features like those on ID cards. By using Genetic Algorithms and Bayesian optimization, they can quickly find the perfect laser settings to produce vibrant, diverse colors that were previously difficult to achieve.

What Problem Does This Paper Address?

Laser-induced printing on plasmonic nanocomposite thin films is a "goldilocks" technology for security: it’s fast, non-contact, and creates high-resolution images that are nearly impossible to counterfeit. However, it has a major drawback: the color gamut is quite narrow compared to standard inkjet printing.

The technical challenge is that there is no straightforward mathematical model to predict exactly what color a specific laser setting will produce on a specific film. Currently, engineers have to rely on tedious lookup tables. Because the physical properties of the films vary from batch to batch, finding the "best" colors for a specific sample is a slow, manual process. This paper seeks a way to automate and optimize this search to get the most "pop" out of laser-printed images.

Methods

The authors implemented an in situ setup that allows them to print, measure, and adjust colors in real-time. They tested two primary iterative optimization strategies:

  1. Genetic Algorithm (GA): Inspired by biological evolution, this method "breeds" laser parameters. It selects for "fitness" based on two criteria: high color saturation and hue diversity. Over several generations, the algorithm converges on a set of parameters that cover the widest possible variety of intense colors.
  2. Bayesian Optimization: This approach uses a probabilistic model to predict what colors might be possible based on the data already collected. It specifically looks for "gaps" in the current color gamut and suggests laser parameters likely to expand the total volume of the printable color space.

Once the optimal gamut is found, they use Gamut Mapping—a technique to translate the colors of a digital original image into the best possible match within the limited range the laser can actually produce.

Datasets

The study does not use standard external datasets. Instead, it generates its own experimental data using:

  • Plasmonic nanocomposite thin films: Specifically, semi-transparent random plasmonic metasurfaces.
  • In situ measurements: Real-time color data captured during the laser processing.
  • Psychophysical study data: The authors mention using metrics derived from human perception studies to evaluate the final image quality, though the specific raw data from these studies is not detailed in this short paper.

Key Insights and Ideas

  • The Power of Multiplexing: One of the most fascinating aspects of this technology is that the colors change depending on how you look at them. A single sample can show one image in reflection and a completely different one in transmission (backlit).
  • Moving Beyond Lookup Tables: By treating color discovery as an optimization problem rather than a static mapping problem, the researchers can adapt to different material samples on the fly.
  • Saturation vs. Volume: The paper highlights a subtle but important distinction in optimization. You don't just want "bright" colors (saturation); you want a diverse range of colors (hue diversity) to ensure the final image looks natural and detailed.

Key Results

While this specific conference paper focuses on the methodology, the authors highlight several qualitative and procedural successes:

  • Automated Convergence: Both the Genetic Algorithm and Bayesian approaches successfully converged to a maximum gamut faster than random sampling.
  • Visual Fidelity: By combining optimized gamuts with custom gamut mapping, they were able to reproduce images that are "closest to the original" despite the inherent physical limitations of the medium.
  • Security Applications: The methods were successfully applied to create multiplexed images visible to the naked eye under unpolarized white light, a significant step forward for anti-counterfeiting technology.

Limitations and Future Work

The primary limitation remains the inherent physical gamut of metallic thin films; while optimization helps, laser-induced colors still cannot match the full breadth of high-end inkjet CMYK or RGB displays.

The authors suggest that future work will focus on refining the psychophysical metrics used to judge image quality. Since "quality" in a security document is often subjective (how "real" or "clear" a portrait looks to a human agent), aligning the algorithmic optimization with human visual perception is the next logical step. They also aim to further explore the "best observation mode" (reflection vs. transmission) for specific types of imagery.

Summary updated

Abstract

Laser-induced printing is an affordable, fast, and non-contact method for creating high-resolution images. Using plasmonic nanocomposite thin films, it enables the printing of color images with visual effects. However, the color gamut of these images is limited compared to inkjet printing. Therefore, it is necessary to optimize this gamut in order to print images that contain the widest range of colors and are closest to the original. There is currently no model to directly infer the color from the laser parameters used. Instead, a lookup table is required to associate these parameters with actual colors. Since colors vary depending on the type of sample used, it is essential to have a reliable method to quickly determine the laser parameters that produce the best colors. Two methods have been implemented to optimize the color gamut: a genetic algorithm approach to find colors that both increase both hue diversity and saturation, and a Bayesian approach to increase the size of the color gamut. Gamut mapping is then used to print the image, and the quality of the final printed image is assessed using metrics obtained from a psychophysical study.
BibTeX
@inproceedings{mermillodblondin2025optimization,
  author = {Raphaël Mermillod-Blondin and Nicolas Dalloz and Mathieu Hébert and Alain Trémeau and Rémi Emonet and Nathalie Destouches},
  title = {{Optimization of laser-induced printed colors for industrial applications}},
  booktitle = {EPJ},
  year = {2025},
  doi = {10.1051/epjconf/202533503033},
}
Figure