Robert,

Thanks for the work, very interesting.

I just have a really difficult time believing that blue, though. That's just an amazing color if true!

What worries me is the "correcting the autoexposure" bit. How exactly does that work, and how can you be confident the formula applies to a given set of images?

What I wonder is whether the blue images had longer exposures than your software is assuming. That is, if you were to make the blue images an f-stop or two darker and then do the combination, what do you get? I think that would be a better test than just turning down the saturation.

If you make the blue images a lot darker that will tend to make the RAT area more blackish, which is easier to swallow. Even black-purple would probably give you a wider range of mineral possibilities than shocking cobalt!

The question is whether darkening the blue screws up the color in the non-RAT areas (to become too red for "True Mars[tm]"). If you can darken the blue images and the rest of the scene stays a credible color, then I'd be inclined to believe that the "autoexposure correction" is failing in this case and is making the blue component brighter than it really should be.

Thanks also to jms for the rundown on mineral possibilities!