Yes, this is normal for many color cameras. A color CMOS sensor typically uses a Bayer color filter array, so the camera does not directly measure fully separate RGB images. It measures filtered raw pixels, then builds RGB output through demosaicing and color processing.

Two things make the channels interact:

1. Spectral overlap: Red, green, and blue filters are not perfectly isolated. Each channel responds somewhat to neighboring wavelengths.
2. Demosaicing / color conversion: The final RGB values are computed from neighboring filtered pixels, so boosting one raw color can influence the reconstructed values of the others.

The fact that the effect stops once green saturates strongly suggests the interaction is happening in the camera’s raw-to-RGB processing, not because the channels are truly independent analog paths.

If you need more independent measurements, the best approach is to work with the most raw sensor data available and avoid in-camera color processing where possible. A monochrome camera with controlled illumination or narrowband filters may also be better if channel separation is critical.