University of Ljubljana
Faculty of Mathematics and Physics
Jadranska ulica 19,
1000 Ljubljana
Slovenia
Room J19 502
Jožef Stefan Institute
Condensed Matter Physics F5
Jamova cesta 39,
1000 Ljubljana
Slovenia
Instead of designing optical components by hand and testing them, I use inverse design (topology/shape optimization) to let an algorithm discover the material structure that produces a desired optical response. Starting from a target, for example a transmission spectrum or a diffraction pattern, and the method iteratively optimises the refractive-index or material layout to meet it. I apply this to components such as broadband optical filters, diffraction gratings, and waveguides, often with constraints that keep the resulting designs physically realistic and manufacturable.
Computer-generated holograms encode a target image or wavefront into a physical structure that reshapes incoming light. I develop material-constrained optimization methods for designing such holograms, so that the computed solution respects what the fabrication material and process can actually produce. Compared to classical approaches like the Gerchberg–Saxton algorithm, this yields holograms that are directly manufacturable, in my case through photopatterning of liquid crystals.
Liquid crystals are soft, birefringent materials whose molecular orientation can be patterned with light and reconfigured with external fields, making them a versatile platform for tunable optics. I study how to arrange and control liquid-crystal structures to build photonic devices such as diffraction gratings, waveguides, and holographic elements. Using techniques like photoalignment, these devices can be reoriented electrically or optically, enabling low-power, reconfigurable components for light control.
Ž. Černe, P. Ropač, and M. Ravnik, Inverse design of broadband soft matter optical filters, Opt. Express 34, 16416-16429 (2026). [Link].
M. Stebrytė, E. Raffin, L. Vlieghe, P. Ropač, M. Ravnik, I. Nys, K. Neyts, and J. Beeckman, Off-axis reflective phase-only holograms with chiral liquid crystal, Opt. Mater. Express 16, 455-463 (2026) [Link].
P. Ropač, Y.-T. Hsiao, B. Berteloot, Y. Ussembayev, I. Nys, M. Ravnik, and K. Neyts, Liquid Crystal 3D Optical Waveguides Based on Photoalignment. Adv. Optical Mater., 2402174 (2025). [Link]
P. Ropač, Yu-T. Hsiao, B. Berteloot, M. Ravnik, and J. Beeckman, Material-constrained optimization of liquid crystal-based holograms, Adv. Opt. Matter., 2400972 (2024). [Link].
I. Nys, P. Ropac, B. Berteloot, M. Ravnik, K. Neyts, Highly dispersive liquid crystal diffraction gratings with continuously varying periodicity, J. Mol. Liq. 383, 122062 (2023). [Link] [PDF].
Glenn H. Brown Prize from the International Liquid Crystal Society.
Facebook Reality Labs Research Gold Award from the International Liquid Crystal Society.
SPOMSO
SPOMSO is a free and open-source Python package designed for procedural generation of complex geometry and vector fields. Its key features also include automatic differentiation, enabling sophisticated gradient-based shape optimization workflows, and strong compatibility with LLM-based tools to generate complex geometry and vector fields.
Latest release: 15.9.2026
A Method and Device for Tiling Fractal Noise - WO2025201635
The present invention belongs to the field of image data processing or generation, more precisely to the field of general-purpose image data processing. The present invention relates to a method that enables generation of complex pseudo-random patterns/noise in n-dimensional space. The user can specify and create a highly specific n-dimensional noise generator, which will output tiling fractal noise. The n- dimensional patterns are generated by recursively combining a certain number of tilings of random values of the same type, for example Quadrilateral, Hexagonal, Voronoi, outputs of other noise generators, but with different translation, rotation or length scales. The algorithm also allows for the generation of geometric patterns without randomness. In this case, rotation and translation is not random, i.e. tiling is always performed with the same orientation, but with different scale, and wherein individual tiles have a pre-defined value, which is not randomly generated.