Color Cast

A cast is easiest to measure on blank glass, which should be colorless. Convert its average color to CIE L*a*b*, where L* is lightness, a* runs green to red and b* runs blue to yellow, and take the distance from neutral: ΔE = √(a*² + b*²). The space is built so that equal distances look roughly equally different.

Worked: blank glass reading R 225, G 224, B 222 has a* ≈ 0.0 and b* ≈ 1.1, a ΔE of 1.1, effectively neutral. Glass at 232, 220, 188, which looks only slightly warm, has b* ≈ 17 and ΔE ≈ 17. Uncorrected halogen glass at 230, 180, 105 reaches ΔE ≈ 46. A ΔE of about 2–3 is near the smallest difference a trained eye notices.

The usual causes, roughly in order of frequency: a white balanceLoading... set under different conditions, such as another objective, filter or day; a halogen lamp running at a different voltage from the one it was balanced at (see color temperatureLoading...); auto white balanceLoading... left on; a filter left in or taken out of the light path after balancing.

Two less obvious ones. Room light can enter through the eyepieces and reach the camera on some microscopes; close the eyepiece shutter or cover the eyepieces while acquiring. And LED illuminators drift slowly with age, so a balance that was right in spring can show a cast by autumn.

To the eye a mild cast is cosmetic. To color deconvolutionLoading... it is extra optical densityLoading... in one or two channels at every pixel. A yellow cast, which is a blue deficit, adds most to the blue channel, which is where DABLoading... absorbs most strongly, so it is read as a faint, uniform layer of DAB across the image.

The effect is largest exactly where it is hardest to see: in weakly stained cells near a positivity threshold, and in comparisons between slides imaged on different days. Check blank glass ΔE at the start of each session, and treat anything above about 3 as a reason to re-balance before acquiring.