Hey Harry,
different weights are given the RGB channels from the same image some times very different weights
The different weights are computed from automatic noise estimates for each channel, as you're using the
automatic noise evaluation weighting method. With this method each channel is treated as an independent image. The computed weights will provide the highest possible SNR improvement in each channel of the final integrated image.
doesn't this unbalance the image
The resulting RGB image can be slightly different if you use a weighting method where equal weights are assigned to all channels: either the "don't care" method, or the "exposure time" method, for example. The differences can be larger if your images have been acquired under varying conditions: different exposure times, ambient temperatures, or even different light pollution and atmospheric conditions. However, automatic noise estimation will always yield the best result in terms of SNR increment.
It's up to you, but I bet you'll prefer less noise and more signal, even if this means altering the inherent color balance of your image set. And we have very good color calibration tools!
