Continuing my exploration of the effects on color accuracy and image noise in various cameras, I ran my simulation on a group of cameras with lighting from D22 through D90.
Here’s what I got for color accuracy, using sensitivity metamerism index (SMI) as the metric:
With the exception of the Sigma DP1 Merrill, which is all over the place, and the XYZ and optimal Gaussian responses, which are unaffected by the temperature of the light source, there is a tendency for color accuracy to decrease as the color temperature drops below D40. As before, the optimum compromise matrices are computed for each light source; if that were not the case, the color would be worse at other than that where the training occurred.
What about noise?
As usual, the DP1 Merrill is an outlier. Other than that, the H3D seems to be the only camera where noise is a strong function of color temperature of the lighting.
Leave a Reply