It looks pretty good except for the slightly purple cast to the blue reference.  I suspect that the L3 filter is leaking some IR and that is causing the problem.  It may be that the filter has changed somewhat during its 7 months in space, perhaps due to exposure to vacuum or temperatures that it was not meant for.

   Or, there could be some other cause.  I will go into that later on.

   The original images are here at the NASA web site.

   I noticed the pale grayish cast to the soil right away, and the fact that even though the colors are less "fruity", the image is flat or lacks something.

   Then I realized that it was the anodized aluminum base plate.  I have the exact same materials in my lab here and I know what color they should be.  Gold anodized aluminum has a characteristic and very recognizable look to it.  This was washed out and did not look right at all.

   Therefore, the soil also must be wrong in color.  The reality is somewhere between the last image and this one.

   Also, since the image has known intensity white and gray circles, I could perform a standard color densitometry test and determine what the starting values should be to arrive at the proper "pure white" figures.  This image is the result of a few simple calculations and comparisons to known hardware that I own.

   And sure enough, this is a very realistic rendition and the color has body and matches the expected white and anodized colors.  So except for the violet cast to the blues, it is a good image.

   But instead, we see a horrible mixture of blue-green and pale rock.  What went wrong?

   We used the exact same filters, and the images were taken only minutes from the calibration images.  The sky surely did not cloud over, and the lighting conditions should be exactly the same.  But something is radically different.  What could it be?

   Original Sol 027 images are located here at the NASA website.

   While this image is more realistic looking, the problem is simple- it has no bright reference object in the image field, and therefore, the camera calibrated itself now to the much dimmer ground, and the colors that make the image are at very different percentages from those colors of the reference fixture.

   This means that 100:80:60 is wrong for dark fields, and when you have no white reference in the field, the camera will set each according to the overall scene but not based on a bright white reference object.

   This means that our carefully calibrated figures will no longer work, even though everything else is apparently the same.  So how can we know exactly what we should see?

   This image is made to match as closely as possible the color values that are shown with the calibration reference and are the proper brightness and chroma if the calibration reference is in the image.

   But our organic eyes adjust automatically to varying lighting conditions, and just like going outside after being in a house for a while, the true brightness levels are incredibly different from what we expect.

   As a result, we tend to adjust the images until we can see the most detail in a manner that is comfortable to us.  And that is one reason why the images look very different- people make them look best.

   This very clearly explains why the ground images from the previous pages are so dark- when your eyes adjust to the brightness of the sky, they lose the ability to see the details of dark objects well.  Based on NASA's own calibration reference, this is the true color of the ground and rocks in this picture.

   This is backed up by the incredibly dark interior of Endurance Crater in visible light, as well as the calibration image shown above.

Compared to the sky and the color calibration reference, the surface of Meridiani Planum is pretty dark.  This is very close to the way it would look to human eyes, if a white object was placed in the field of view.  Otherwise, our irises would expand and make the scene much lighter.

   This is because the greater brightness of the sky sets the upper limit for the brightness in all frames of color data.

   By adjusting the contrast and brightness levels to match the average rock and soil densitometry, this process has produced the truest image that is possible using the L2, L5, and L6 filter data from Opportunity on Sol 001.

   Since these values are attuned to the calibration from NASA's own color references, it is therefore the true and natural color of Mars' sky as it would appear to a human observer on the surface of the planet.

   But take note of the fact that this image still has a bit too much red in the ground, due to the use of L2, which is (as stated before) infrared.

   The brightness and contrast on these images was set to +65% and the saturation was fine without adjustments.

   The dust is getting thick and we can expect degraded performance from the solar panels due to this accumulation.  It is slowly blocking out more and more sunlight and making the panels unable to collect the full amount of power they are capable of.

   Just out of curiosity, I decided to test the IR leakage idea and I used the L4 filter (601 nm, really an orange instead of red).  See the next image.

   So although the orange is not a very good red, the images it makes are very clean and have an excellent purity and color balance.  And, it works for us because the "red" pigment in the human eye is most sensitive to light very close to 601 nm anyway.

   There is no visible degradation of the red, yellow, or green in this image, and on top of that, the blue is perfect.  Dusty, sure, but the chroma is exactly as we would expect.

   Once again, I used the RGB ratio of 100:80:60 for this brightly lit white reference object.  Other darker objects will have different white balance levels and will require something more like the rule of thumb, which is 100:60:30 for RGB.

   The original images are here at the NASA web site.