Compare white light, simultaneous RGB, and three-shot capture

Before you start

All three examples use a Bayer color camera and include later processing. Compare them on the same screen at the same magnification. They show differences in appearance, not a controlled accuracy test; brighter or more saturated does not necessarily mean more accurate.

Inspect and compare

1. What changes with three exposures

   With simultaneous RGB, illuminant contributions add in linear signals. Sequential capture takes R under red only, G under green only, and B under blue only, excluding the other lights from each role record. It removes neither dye side absorption nor the Bayer CFA. Results may improve if the input becomes easier to calibrate reliably. However, an ideal invertible one-shot mixture can recover the same three band values; three shots do not guarantee more information or better accuracy. Read the conditions on the principles page when interpreting these appearance differences.

   Reading the comparison: distinguish the two changes

   State	What changes	What to inspect
   Bayer + white, one capture	Broadband illumination, separated by the CFA	Establish snow, neutral mountains, and grass as the baseline
   Bayer + simultaneous RGB, one capture	Illumination uses selected bands; the CFA still receives mixed light	Compare hue relationships with the white-light result
   Bayer + sequential RGB, three captures	Additionally separates the bands in time	Compare neutral casts and color distinctions with simultaneous RGB

2. Inspect the snow, grass, and black cow

   Switch first between white and simultaneous RGB, then between simultaneous and sequential RGB, keeping the two changes distinct. Inspect casts in snow and mountains, shadow detail in the animal, and distinctions between grass, earth, and sky. Snow reflects scene illumination and need not be colorless white; a darker animal does not necessarily mean better shadows. Select an image to enlarge. Original processing is retained with only web sRGB conversion and resizing, without added saturation or contrast.

   

   Bayer + white, one captureBayer + simultaneous RGB, one captureBayer + sequential R/G/B, three captures

3. Use Bayer mode for this example

   Three captures describe the acquisition method, not a change of sensor. This example belongs to Bayer Trichrome and requires a CCM matched to that input and capture setup. The rgb_7/8/9 files on the capture lesson are a separate Monochrome standard sample, not the sources of this black-cow comparison. Do not mix their sensor settings or CCM profiles.

4. Off-diagonal values alone do not measure color accuracy

   Off-diagonal CCM terms describe cross-channel correction; they are not automatically contamination. Channel scale/order, coordinate basis, and target affect the matrix and its ratios, so one percentage is not a cross-system accuracy ranking. Assess matching charts, fit residuals, independent references, clipping, and photographs together.