In this context, signal-to-noise means maximizing the feature of interest against a background that looks nothing like it. Angle, intensity, pulse length, wavelength, and polarization do more of that work than any filter. Dome light softens specular highlights. Low-angle or dark-field light picks up edges and surface breaks. Diffuse axial light reveals shoulders inside holes. A backlight makes a silhouette for counting or shape. A ring light changes what it emphasizes as its height changes. Bar lights from both ends suit cylinders.
The useful lab is one where learners see those effects on simple shapes and a reflective syringe, rather than memorizing a catalog. Windows, stack lights, and room lights are noise. Overpower them or block them, and shorten exposure so vibration does not paint the feature across extra pixels.