Examples

Every example below has a runnable script in the repository root and a dedicated page with a code walkthrough and the figures it produces.

Getting Started

Example	Description
GeoLens	Load a refractive lens, run optical analysis, render an image.
DefocusLens	Thin-lens defocus / depth-of-field & bokeh.
DiffractiveLens	Pure wave-optics phase plate with ASM propagation.
HybridLens	Refractive lens combined with a DOE.

Lens Design

Example	Description
GeoLens Design	Curriculum RMS-spot optimization of a refractive lens.
DiffractiveLens Design	Optimize a Pixel2D DOE to focus, via a Strehl (peak) loss.
HybridLens Design	End-to-end refractive–diffractive design (ray–wave model).

Advanced

Example	Description
Automated Lens Design (RMS)	Ab-initio lens design from target specs.
PSF Network	Neural surrogate that predicts the spatially-varying PSF.
4f System	Fourier-plane diffractive filtering in a 4f relay.
Pupil Field & Wavefront	Exit-pupil wavefront by coherent ray tracing.
Multi-order Diffraction	All grating diffraction orders in one ray–wave PSF.
Diffractive Surfaces	Three paper-based DOE parameterizations.

---

Recipes

A couple of general rendering recipes that apply across the lens models.

Image simulation methods

GeoLens.render() supports three flat-scene rendering modes:

img_ray = lens.render(img, depth=-20000.0, method="ray_tracing", spp=32)
img_patch = lens.render(img, depth=-20000.0, method="psf_patch", patch_center=(0.0, 0.0))
img_map = lens.render(img, depth=-20000.0, method="psf_map", psf_grid=(10, 10), psf_ks=64)

# psf_spp controls samples per PSF in psf_map mode (default: 8192).
img_map_fast = lens.render(img, depth=-20000.0, method="psf_map",
                           psf_grid=(10, 10), psf_ks=64, psf_spp=1024)

For RGBD scenes, Lens.render_rgbd() blends reference depth layers:

# depth_map is positive depth in millimeters, shape [B, 1, H, W].
img_rgbd = lens.render_rgbd(img, depth_map, method="psf_map",
                            psf_grid=(8, 8), num_layers=16, interp_mode="disparity")

Distortion warp

GeoLens can compute an inverse distortion grid and apply the lens distortion:

inv_grid = lens.calc_inv_distortion_map(num_grid=(32, 24), depth=-20000.0)
img_distorted = lens.warp(img, depth=-20000.0, num_grid=(32, 24))

This is used internally before GeoLens.render(..., method="psf_map").