LESS

3D Radiative Transfer Model

Version 3.0.0rc1

LESS is a GPU- and CPU-accelerated 3D radiative transfer model for quantitative remote sensing. It combines a Python interface with NVIDIA OptiX, Vulkan Ray Query, and Intel Embree native backends for optical, thermal, solar-induced fluorescence (SIF), vegetation physiology, energy-balance, and LiDAR simulations.

LESS 3.x builds on the original LESS model. The original LESS 2.x implementation remains available at lessrt.org.

Highlights

• Remote-sensing-oriented Python API — describe a scene using terrain, objects, optical properties, illumination, sensors, and products.
• Three native backends — OptiX for NVIDIA GPUs, Vulkan Ray Query for supported GPUs, and Embree for CPU execution.
• Persistent scenes — geometry is prepared automatically on the first simulation and reused when only illumination, spectra, or properties change.
• Mesh and turbid vegetation — simulate explicit geometry, volume canopies, or mixed scenes.
• Imaging and biophysical products — optical and thermal imagery, BRF, radiation fields, irradiance maps, SIF, photosynthesis, energy balance, and LiDAR.
• Sensor effects — pushbroom geometry, spectral response, PSF/MTF, detector noise, ADC quantisation, and geolocation.
• Interactive inspection — use scene.show() to explore scenes and products in a browser-based viewer.

Installation

pip install less3d==3.0.0rc1

Prebuilt wheels are provided for Windows and Linux with Python 3.10–3.13. A compiler and CUDA toolkit are not required. Compatible graphics drivers are still required for GPU backends.

Linux 3.0.0rc1 note: the wheel uses a unified native module and requires libcuda.so.1 to be available when less is imported, including when Embree or Vulkan is selected. Install a compatible NVIDIA driver before using this release on Linux.

Check whether a feature can run with a particular backend on the current computer:

import less

less.can_use("optical_imaging", backend="optix")
less.can_use("irradiance_map", backend="embree")

After creating a scene, scene.can_use("feature") performs the same check using that scene's selected backend.

Available in 3.0.0rc1

The following workflows are available with OptiX, Vulkan, and Embree when the selected backend meets its hardware and driver requirements:

Coupled atmosphere-to-sensor propagation, passive and active microwave simulation, and water optics are not available in 3.0.0rc1. Requesting one of these workflows raises a clear availability error instead of returning an approximate result.

Repetitive scenes require flat, statistically homogeneous terrain and cannot be combined with terrain_following=True. Moving a TurbidBoundary instance requires scene.rebuild(); ordinary property and illumination changes do not.

Quick start

import less

# Create a 10 m × 10 m bare-soil scene.
scene = less.Scene()
scene.size = 10.0
scene.terrain = less.Terrain(
    property=less.Lambertian(reflectance=0.3)
)
scene.illumination = less.Illumination(
    source=less.Sun(zenith=30, azimuth=150),
    atmosphere=less.NoAtmosphere(),
)

# Simulate a nadir RGB image.
sensor = less.OpticalImager(
    less.Orthographic(image_size=256),
    bands=[650, 550, 450],
    quality=128,
)
image = scene.simulate(sensor)
image.save("output.png")

Change the illumination and simulate again; the prepared geometry is reused automatically:

scene.illumination = less.Illumination(
    source=less.Sun(zenith=60, azimuth=90),
    atmosphere=less.NoAtmosphere(),
)
image2 = scene.simulate(sensor)

Forest example

import less
import numpy as np

scene = less.Scene()
scene.size = 30.0
scene.terrain = less.Terrain(
    property=less.Lambertian(reflectance=0.12)
)
scene.illumination = less.Illumination(
    source=less.Sun(zenith=35, azimuth=225),
    atmosphere=less.NoAtmosphere(),
)

birch = less.Object("birch", mesh="birch.obj")
birch.set_property(
    "leaves",
    less.Prospect(cab=40, car=8, cw=0.012, cm=0.009, N=1.5),
)

xy = np.random.default_rng(42).random((50, 2)) * 30
scene.add(birch, positions=np.c_[xy, np.zeros(50)])

image = scene.simulate(less.OpticalImager(
    less.Orthographic(image_size=512),
    bands=[550, 670, 800],
    quality=256,
))
image.save("forest.tif")

GeoTIFF output preserves the physical values of the simulated product.

Examples and viewer

import less

less.examples.list()
less.examples.run("autumn_forest")
less.examples.run("maize_field")

scene.show()

Validation

LESS is cross-validated against LESS 2.x on reference scenes:

Documentation

Start with the public tutorials:

• 中文：安装
• 中文：核心概念与场景生命周期
• 中文：API 参考
• English: Installation
• English: Concepts and scene lifecycle
• English: API reference

Citation

If you use LESS in your research, please cite:

Qi, J., Xie, D., Yin, T., Yan, G., Gastellu-Etchegorry, J.-P., Li, L., Zhang, W., Mu, X., and Norford, L. K. (2019). LESS: LargE-Scale remote sensing data and image Simulation framework over heterogeneous 3D scenes. Remote Sensing of Environment, 221, 695–706.

License

LESS is distributed under the MIT License.