An Examination Of NASA's Color Methods
How to assemble a color image, and how to undo the errors in the NASA data
Now that I have chosen three images, I will show the steps to creating a good color image.
| The same three images seen before have now been converted to color frames. These are thumbnails of the images. Now we can start the process of assembling them into a color picture. |  |
 |
 |
| The red data is really an
infrared frame. However, all the frequencies we have are shortwave
IR, green, and blue. For many images, the difference will be small
and human eyesight will not have any problems with the results.
So, using the L2 image, I have turned the image into a red frame. This is done by setting the level controls for each "channel" of the image. In a monochrome (or "black and white") image, there are identical amounts of red, green, and blue light present. This is what produces "clean" white light. I have removed the green and blue spectrum from the picture and the result is a pure red frame of image data. This is the first step on reconstructing the true color representation of a picture. |
 |
|
| This is the L5
filter image. Since L5 is the 535 nm filter, that means that this is
green light only.
In exactly the same manner as I created the red image above, I then adjusted the L5 image to have no red or blue light in it. The resulting image frame is pure green and represents the "slice" of image data that the rover camera would have captured. Using this image, we can add it to the data for the final image and get the green data into the picture. |
 |
|
| The last image is the blue
data. I performed the same operation on this frame, which
incidentally is the 483 nm image data. This is a good color of blue
light to use in making the final color picture.
Notice how dark the ground appears here? Normally, there is much less blue light in Martian soil and rocks because they are mostly red or orange. It makes sense that the ground, being red or orange, would have less blue light. If you do this process of color separation on a picture of the grass at home, the red and blue images will be dark where the grass is, but the green image will be much brighter. This is perfectly normal, as green grass will have more green light reflected off it. Now we can assemble the three frames into a color image. |
 |
If we take these three colored images and overlay them, we can construct a full color picture that should look like Mars. Of course, nothing is ever as simple as that.
Sidebar: looking at the color filter data from Spirit, from Sol 001 to Sol 040, NASA posted the filter widths instead of the actual wavelengths. This is a major mistake, as without a reference chart, it is impossible for anyone to construct color images or even begin to interpret the data properly. Starting in Sol 041, some are filter width frequencies and some are actual wavelengths, but the frequencies and widths are mixed up for many sols to come.
In Sol 044, the right and left filter frequencies are actually reversed, so that the left filter frequencies are shown as being infrared band, and the right filters are shown as visible light. These numbers have been transposed and the results are that you cannot trust ANY filter data except for the L or R numbers. You have to know them or look them up yourself.
It took NASA until Sol 078 to start posting the correct left and right filter frequencies, meaning that in 78 sols (which is about 80 days) they had not figured out the errors. Maybe their secretary was having a bad three months.
Opportunity also had its filter frequencies botched up until at least Sol 059. I have not yet checked each and every one, but they had right and left filter frequencies swapped as well as other problems. Any new information, however, that ends up being posted in old sols has the proper data attached to it.
Personally, I think that they should assign somebody to correct this. A good typist with a list could correct the annotations in less than a week. It's a sad statement that this sort of public reference information is not usable to the individual who knows something about photography, unless they happen to have the list of filter frequencies and take the time to cross-reference it.