Demonstration of the problem with Foveon sensors' problems shifting photons between the green and blue channels, in arbitrary blotches.

The green square contains an area of the green channel of a shadow area (poorly lit smooth red subject), and the blue square contains the same 300% crop, but from the blue channel, and inverted (made negative). There is a "zero sum" game going on here with blue and green channel noise. This is the reason for the low-frequency chromatic noise in Foveon images. Added together, the blue and green channels make an excellent, low-noise B&W image. The red channel, by itself, also. The Foveon therefore makes excellent cyan-software-filtered and red-software-filtered B&W images, but any imbalance between blue and green brings on noise, and in interpolating color images, low-frequency chromatic noise is fairly strong.