Ray tracing before you cut metal

The expensive way to find a stray light path is to build the instrument. The cheap way is to trace it.

What the model needs

Import the mechanical CAD, not just the optics. Stray light almost never comes from the lenses; it comes from a bright flange, a bare screw head, or a housing wall at exactly the wrong angle. A model containing only the optical train will tell you the design is clean and be wrong.

Then give every surface a real optical property. Specular, diffuse and scatter behaviour drive the result, and this is where a trace is either useful or misleading. Measured or manufacturer-supplied data is worth the effort of obtaining; a plausible guess is not.

Illumination versus imaging

Illumination design asks where the light goes and how much arrives, and the answers are illuminance maps and candela plots against a photometric specification. That is a different question from image quality, and it is the question a raytracer of this kind answers well.

Using the optimiser sensibly

An optimiser will move a reflector geometry toward your target far faster than hand iteration, and it has no opinion about whether the result can be moulded, machined or assembled. Constrain it with manufacturing limits from the start.

When it pays

The value is concentrated immediately before a tooling commitment. Finding a stray light path two weeks before the tool is cut saves a prototype cycle; finding it afterwards costs one.