optiland/optiland

Comprehensive optical design, optimization, and analysis in Python, including GPU-accelerated and differentiable ray tracing via PyTorch.

What it solves

Optiland provides an open-source platform for designing, simulating, and optimizing optical systems. It bridges the gap between classical lens design (refractive and reflective systems) and modern computational optics, allowing engineers to model everything from simple singlets to complex freeform assemblies with professional-grade precision.

How it works

The platform uses a Python-based API to construct optical systems by defining surfaces, materials, and apertures. It features a dual-engine backend: NumPy is used for fast CPU-based calculations, while PyTorch is integrated to enable GPU acceleration and automatic differentiation (autograd). This allows users to optimize optical designs using either traditional merit functions or differentiable modeling where gradients are calculated automatically to refine the system.

Who it’s for

It is designed for optical engineers, researchers in computational optics, and students who need a flexible, scriptable alternative to proprietary optical design software for prototyping and production systems.

Highlights

  • Differentiable Core: Seamlessly switch between NumPy and PyTorch for hybrid physics-ML workflows.
  • Comprehensive Analysis: Generates spot diagrams, wavefront error maps, PSF/MTF plots, and Zernike decompositions.
  • Advanced Ray Tracing: Supports sequential, non-sequential (pre-release), and multi-sequence tracing for ghost and stray-light analysis.
  • Interoperability: Imports and exports files from industry-standard software like Zemax, CODE V, and OSLO.
  • Professional Tooling: Includes a GlassExpert module for material selection and Monte Carlo simulations for tolerancing analysis.

Related

  • Project
    vccimaging/DeepLens

    A differentiable optical lens simulator for computational imaging and optical design that allows lenses to be optimized as learnable parameters within a PyTorch training loop.

  • Project
    facebookresearch/pytorch3d

    A PyTorch-based library for 3D Computer Vision research providing differentiable rendering and efficient tools for manipulating 3D meshes and point clouds.

  • Project
    pytorch/pytorch

    PyTorch is an open-source machine learning framework that provides GPU-accelerated tensor computation and a flexible tape-based autograd system for building deep neural networks.

  • Project
    tomas-gajarsky/facetorch

    facetorch is a Python library that bundles open‑source face‑detection and analysis models (embedding, verification, expression, etc.) into a single, configurable API. Models are exported with `torch.export` for portable inference, and the toolkit handles lazy downloading, caching, Hydra‑based configuration, and safe input handling. It offers both a Python package and ready‑to‑run Docker images (CPU and GPU).

  • Project
    Lightning-AI/pytorch-lightning

    A deep learning framework that abstracts away the engineering boilerplate of PyTorch, enabling researchers to pretrain and finetune AI models at scale across various hardware.