Doug,

I'll grant that the autoexposure problem is a serious one in terms of trying to construct color images on one's own, and that most of the blue-green stuff seen in amateur color images is probably not real.

In the case of Mazatzal, though, there does seem to be something weird with the color of the "rind".

It's clear even without the exposure data that the un-brushed rock (which is presumably the usual dust color -- rust-brown) and the coating have different spectra. The contrast in both the red images and blue images is very large, and looks to be larger than for previous RATted areas.

For the blue images, if we are looking at much longer exposures than we think, that's a way to explain the RAT area being grey instead of blue. But then, the rest of the rock is so incredibly dark in blue that the rock would have to be red, not rust-brown.

In other words, if you look at the RGB components for other color images that JPL has assembled, and then look at the Mazatzal images, you would think, "However their black box works, this time it looks like it will spit out something bluish".

Now, if the processing does NOT just include linear brightness differences due to different shutter speeds but also means that things like the gamma can vary from one filter to the next on the same target, then it gets easier to explain (and much harder to reverse engineer).