AI4Optics
Automate optical design with agentic AI and differentiable optimization.
AI4Optics is a family of open-source tools that bring agentic AI and differentiable optimization to optical system design — from lens design to full camera-in-the-loop image formation. Pick a project below to get started.
-
DeepLens
Differentiable optical lens simulator for end-to-end camera system design — geometric, diffractive, and hybrid lens models with PSF rendering and automated design.
-
End2endImaging
End-to-end differentiable imaging framework — optics, sensor, ISP, and reconstruction networks as one optimizable graph for camera-in-the-loop system design.
-
DiffTMM
Differentiable transfer matrix method in PyTorch for multi-layer thin-film optics — isotropic and anisotropic solvers, GPU batching, and gradient-based inverse design of optical coatings.
More projects coming
Additional AI4Optics projects will appear here as their documentation lands. Each gets its own tab in the top navigation.