Generation of Microstructured Surfaces with Controlled Optical Properties Using Generative Adversarial Networks (M/F)

Job Description

Thesis Subject

The main objective of this PhD is to develop a predictive surface-generation methodology capable of linking a microtopography model to its optical appearance (BRDF/NDF) without systematically relying on experimental prototyping and measurement cycles. Within the framework of the project, the PhD candidate will explore the use of Generative Adversarial Networks (GANs) or other neural-network-based approaches for the synthesis of micro-textured surfaces whose optical and aesthetic properties can be precisely tuned before fabrication. AI-based methods should make it possible to generate realistic height maps from a controlled set of user-defined parameters or from synthetic data, while ensuring consistency between virtually measured appearance and real manufacturing constraints.
Research Topics
1. Design and Training of GANs for Surface Generation
• Generation of realistic height maps from synthetic or experimental data.
• Integration of virtual measurements to guide generation toward manufacturable surfaces that match a target BRDF.
2. Adaptive Mesh Densification and Refinement
• Development of progressive approaches for enriching an initially low-density mesh while preserving high-frequency details.
• Preservation of mechanical and optical properties throughout the refinement process.
3. Computational Cost Optimization
• Reduction of computational requirements through network pruning, model compression, and transfer learning.
• Use of pre-trained models to minimize the size of the required dataset.
4. Real Surface Metrology for Appearance Control
• Topography measurement using photometric stereo or confocal microscopy.
• Image processing for appearance estimation.
• Difference metrics between numerical models and experimental measurements.
The PhD project will build upon previous work carried out at XLIM, particularly in surface generation from Normal Distribution Functions (NDFs) on regular grids. Virtual measurements obtained from optical simulations will be used to guide GAN training and validate generated surfaces before physical fabrication.

Your Work Environment

This PhD project is part of the ANR APPEARANCE_ON_DEMAND project (starting in 2026), conducted in collaboration between the iLM Institute (CNRS UMR 5306) and the XLIM Institute (CNRS UMR 7252). The objective of the project is to master the reproduction of material appearance through precise control of their BRDF (Bidirectional Reflectance Distribution Function), a function that describes how a surface reflects light according to illumination and observation directions. The BRDF is closely related to the surface microtopography. A smooth surface reflects light differently from a textured one, but the relationship between microscopic shape and appearance remains complex, particularly because of multiple reflections occurring between microstructures.
The originality of this PhD project lies in an inverse approach: instead of starting from a given surface to predict its appearance, the goal is to start from a target BRDF, corresponding to a desired visual effect, and determine the microtopography capable of producing it while satisfying manufacturing constraints (embossed printing using the Mihaly technology, the industrial partner of the ANR project). By combining physical modeling, numerical optimization, and artificial intelligence, this work aims to establish a predictive framework for designing surfaces with controlled appearance without relying on extensive experimental prototyping cycles. This approach opens strong opportunities for visual customization in fields such as luxury goods, automotive engineering, and architecture.

Compensation and benefits

Compensation

2300 € gross monthly

Annual leave and RTT

44 jours

Remote Working practice and compensation

Pratique et indemnisation du TT

Transport

Prise en charge à 75% du coût et forfait mobilité durable jusqu’à 300€

About the offer

About the CNRS

The CNRS is a major player in fundamental research on a global scale. The CNRS is the only French organization active in all scientific fields. Its unique position as a multi-specialist allows it to bring together different disciplines to address the most important challenges of the contemporary world, in connection with the actors of change.

CNRS

The research professions