An optical engineer trained in sequential lens design who is asked to design an illumination system for the first time will likely open the tool they know best and start tracing rays. Sequential ray tracing tools such as OSLO, Zemax OpticStudio and CODE V are precise, well-supported, and deeply familiar to anyone trained in imaging optics. They are also the wrong tools for illumination design, not because they trace rays incorrectly, but because they are built to answer imaging questions that do not arise in illumination work, and they lack the physical models required to answer the questions that illumination engineers actually need to answer.
Using a sequential lens design software for an illumination problem produces results that look like ray traces but do not predict what the engineer needs to know: how much flux reaches the target, how uniformly it is distributed, how much escapes the system without contributing to the output, and how the intensity distribution compares against a photometric specification. Getting to correct answers for an illumination system requires a different class of software built around non-sequential ray tracing, Monte Carlo flux accumulation, and BSDF scatter modeling. TracePro is built for this purpose.
What Sequential Ray Tracing Does, and What It Does Not Do
Sequential ray tracing requires the engineer to define a surface sequence: rays interact with surface 1, then surface 2, then surface 3, in order. Every ray in the model follows this sequence. This structure is correct and efficient for an imaging lens, where the design intent is for rays to pass through each element once in the prescribed order and arrive at the image plane with minimal wavefront error.
In an illumination system, there is no prescribed surface sequence. A ray from an LED die may reflect off the interior of a TIR lens, undergo total internal reflection at the outer lens surface, exit the lens, and then reflect off a diffuser before reaching the target. Another ray from the same source may refract directly through the lens and miss the target entirely. A third may back-reflect off the target surface and re-enter the lens before contributing to the output distribution. The fraction of flux that follows each path, and whether those fractions sum to the required uniformity and efficiency at the target, are the quantities the illumination engineer needs.
Sequential ray tracing cannot answer these questions because it constrains every ray to the defined surface sequence. Rays that would physically travel from the LED to the reflector to the diffuser cannot take that path in a sequential model without the engineer pre-defining it as one of the allowed sequences. Complex illumination geometries have dozens of relevant ray paths, and building them all into a sequential model manually is impractical and error-prone. The constraint that makes sequential ray tracing efficient for lens design makes it unsuitable for illumination design.
Source Models: Point Sources vs. Physical Emitters
Imaging lens design uses a point source or a collimated plane wave as the test source for image quality evaluation. The location of the source in the object plane defines the field angle, and the goal is to measure how well the lens maps that point source to a focused image point at the image plane. The physical extent of the source is not relevant to the image quality calculation; it affects resolution but is handled separately as a scene convolution.
Illumination design requires an accurate model of the physical source. A 1 mm x 1 mm white LED die has a spatial extent that determines the etendue of the source, which in turn sets the minimum solid angle over which any downstream optic can distribute the source flux. A point source approximation for the same LED underestimates the etendue by the ratio of the actual source area to the assumed point area, which for a compact TIR lens with a 20 mm aperture at 5 mm working distance means the source extends over 11 degrees of the acceptance cone. Rays from the die edges that are outside the paraxial approximation carry significant flux and interact with the TIR lens surface geometry in ways that the paraxial model does not capture.
TracePro models LED sources from rayfiles, from extended emitting surface definitions with user-specified spatial and angular distributions, or from a combination of multiple source elements representing die, phosphor, and package optic contributions. The source model determines whether the irradiance map at the target plane is a physical prediction or an artifact of the point source approximation. For a compact illumination optic where the source size is not negligible compared to the collection aperture, the difference between a point source and an extended source model can change the predicted irradiance uniformity by 10% to 30%.
Metrics: Image Quality vs. Photometric Performance
Imaging lens design optimizes and reports in the language of image quality: RMS wavefront error, Strehl ratio, modulation transfer function at specified spatial frequencies, and distortion as a percentage of field height. These metrics describe how well the optical system preserves the spatial and phase information in the incoming wavefront. They have no direct meaning for an illumination system, where the goal is not to preserve wavefront information but to redistribute flux.
Illumination design requires a different set of metrics. Total flux at the target, in lumens or watts, describes whether the system delivers enough photons to meet the illuminance or irradiance requirement. Spatial uniformity of the irradiance distribution, typically expressed as the ratio of minimum to maximum irradiance across the target zone or as the coefficient of variation, describes whether the target is evenly illuminated. Intensity distribution in candelas describes how the luminaire directs flux into the surrounding space. System efficiency, as the fraction of source flux that reaches the intended target, describes the energy cost of the optical design.
TracePro computes all of these metrics from the same Monte Carlo ray trace. Irradiance maps at the target plane are generated with spatial resolution matched to the uniformity specification. Total flux at the detector is integrated from the irradiance map. Far-field intensity maps are generated from the angular distribution of rays exiting the system. System efficiency is computed as the ratio of flux at the target to the total flux emitted by the source. None of these results is available from a sequential ray trace that optimizes wavefront error.
BSDF and Scatter: Defining the Output Distribution
In imaging lens design, scatter from surface roughness is a degradation mechanism. It reduces contrast by adding a diffuse background to the image, and it is characterized by a TIS value that feeds into a ghost and stray light analysis. In illumination design, scatter is not a degradation mechanism. It is the primary means by which the optical system achieves the required output distribution.
A diffuser that converts a narrow LED output cone to a wide uniform output distribution works entirely through scatter. Its BSDF, which describes the angular distribution of scattered flux as a function of incident angle, is the defining property of the element. A reflector that distributes LED output across a roadway or a work plane works through the combined effect of specular reflection from its curved geometry and diffuse scatter from its surface texture. The scatter contribution controls the uniformity of the output at wide collection angles where the specular reflection falls below the target irradiance level.
Sequential lens design tools model surfaces as either perfectly transmissive, perfectly reflective, or lossy with an efficiency factor. They do not model the angular distribution of scattered flux from a physical surface. A diffuser in a sequential model is represented as an element with a transmission efficiency, not as an element with a BSDF that controls where the transmitted flux goes. This means a sequential tool can tell you how much flux passes through the diffuser but not where it lands on the target, which is the only question that matters for illumination uniformity.
TracePro assigns BSDF data to every surface in the model: optical elements, diffusers, reflectors, and mechanical structure. The BSDF determines where each scattered ray goes after interacting with the surface, and the Monte Carlo accumulation of millions of scatter interactions at the target plane produces the irradiance distribution that the hardware will deliver.
Monte Carlo Is the Correct Method, Not an Approximation
Imaging lens design uses deterministic ray tracing. A ray at a given entrance angle and field position follows a fully determined path through the optical system, and the wavefront map at the exit pupil is computed from the optical path lengths of a small number of deterministic rays. With 100 to 1,000 rays, a sequential lens design tool produces an accurate wavefront map because the wavefront is a slowly varying, smooth function of pupil position.
Illumination analysis requires Monte Carlo ray tracing because the flux distribution at the target is the statistical result of a large number of ray interactions, each of which involves sampling a probability distribution at a scatter surface or a diffuse reflector. The irradiance at a given point on the target receives contributions from rays that entered the system at different angles, bounced off different reflector facets, and transmitted through different regions of the diffuser. No deterministic ray can represent this statistical average; only the accumulation of enough random ray samples converges to the correct result.
For illumination uniformity predictions at the 1% level, a Monte Carlo simulation in TracePro requires approximately 10^7 to 10^8 rays per source. This ray count is several orders of magnitude larger than what a sequential lens design tool uses for its wavefront analysis, but it is the minimum needed for the irradiance noise floor to fall below the uniformity specification. TracePro's ray tracing engine is optimized for this workload, with multi-threaded execution that distributes the Monte Carlo sampling across all available CPU cores.
The Cost of Using the Wrong Tool
An engineer who runs an illumination design in a sequential ray tracing tool using a point source and deterministic ray tracing will obtain an irradiance distribution at the target plane. That distribution will have some spatial pattern that corresponds to the geometry of the reflector or lens. It will not be an accurate prediction of what the hardware will produce.
The specific errors this approach introduces are predictable. The irradiance at the target center will be overestimated because the point source approximation ignores etendue, and the extended source fills the collection optic less efficiently than a point source does. The irradiance uniformity will be optimistic because the Monte Carlo averaging of millions of scatter interactions, which smooths the distribution, is absent when only a few hundred deterministic rays are used. The efficiency will be wrong because rays that miss the target due to the extended source footprint, scatter losses, and wide-angle reflections are not tracked. The intensity distribution will miss contributions from rays that take non-sequential paths.
Correcting these errors after a prototype is built costs hardware iterations. Correcting them before prototype commitment by running the design in TracePro from the start costs nothing beyond the time to build the correct model. For an illumination system where uniformity, efficiency, and intensity distribution are all specified, TracePro is the tool that produces a simulation result the hardware will actually confirm.
When Sequential Tools and TracePro Are Used Together
The distinction between illumination design and lens design does not mean the two disciplines never interact. Many optical systems include both an imaging component and an illumination component in the same assembly: a machine vision camera that requires a co-axial illumination system, a microscope that combines an imaging objective with a fluorescence excitation illuminator, or a structured-light scanner that uses a patterned illumination projector alongside an imaging receiver.
For these systems, TracePro provides a path that accepts lens prescriptions from sequential design tools. A lens assembly designed and optimized in OSLO, Zemax or CODE V is exported as a prescription file and imported into TracePro, where it becomes the imaging component of the full system model. The illumination component is modeled natively in TracePro with the correct source model and BSDF assignments. The combined model predicts the interaction between the illumination and imaging subsystems, including stray light from the illumination path that enters the imaging channel and ghost images from the illumination source reflected off the imaging optics. This workflow uses each tool for what it does correctly: sequential lens design software for optimizing wavefront quality and tolerancing the imaging train, and TracePro for predicting flux distribution, uniformity, efficiency, and stray light coupling in the combined system.
Contact Lambda Research to request a TracePro demonstration for illumination or non-imaging optical system design, or to discuss how TracePro integrates with your existing sequential lens design workflow.