When browsing the latest press releases at JPL site, I found this gem. What stood out was this little blurb:

"This is the panoramic camera team's best current attempt at generating a true-color view of what this scene would look like if viewed by a human on Mars. It was generated from a mathematical combination of six calibrated, left-eye panoramic camera images acquired around 1:50 p.m. local solar time on Opportunity's sol 322 (Dec. 19, 2004) using filters ranging in wavelengths from 430 to 750 nanometers."





http://www.jpl.nasa.gov/missions/mer/images.cfm?id=1539
http://www.jpl.nasa.gov/missions/mer/images.cfm?id=1538

What's interesting is that the same thing is noted for both of the images, but they appear slightly different.

(apologies if this was already posted here somewhere)

The second of the two you posted, as well as the image below, are 6 filter color and would be the more accurate of the group :

"

The rest are L2 L5 L7 and don't have as much information about the exact wavelengths responsible for each objects brightness, causing the discrepency. I think the yellow tint in the L257s comes from the curve fitting between filters on an object, like the sand, which is bright in L2 (infrared) and dark in L5 (green). When you see the same object in L2/3/4/5, you see that the brightness drops off in the red/orange wavelengths fairly sharply, whereas a curve fit might think there is more yellow than there really is. I don't envy them having to try to produce completely consistant images with only 3 filters, as this sort of problem isn't easily solved unless you try to fit the curves of the 3 filter images to the known spectrum of similar objects seen in full series.

Well heck, look here, about halfway down, the Bonneville panorama.
"The image is in approximate true color, based on a scaling of data from the red, green and blue (750 nanometers, 530 nanometers, and 480 nanometers) filters."
That is, L2, L5, and L6.

Now look at this one, also taken with L2, L5, and L6.

As I understand it, L2 is closer to the infrared portion than the L4 filter, which is technically "red". I think L2 is used because it shows better detail, or something like that - geology related stuff. They might do that to save space in memory - just take the L2, L5, and L6, and not bother with the L4 unless they have time and space for it.
Is weird though how one is called "false color" and the other "approximate true color". Maybe they just toyed around with the "approximate" one a bit more? (Also love the totally uniform sky on the first one.)