Colour images from varying illumination Options

[Fran Ontanaya]

I thought it could be amusing to try this:



It's a composite of rear hazcams from sols 1256, 1264 and 1268 with enhanced, unaltered colors. Oppy didn't move during those days, and took images of the storm at roughly the same time each sol. Since tau was changing, the color of the ambient light must have been different in each one --apparently enough to distinguish the blueberries and some bright rover pieces, and perhaps the sky and the dust too.

[ugordan]

I don't follow you. The hazcams are panchromatic imagers and any color you see here can only be due to:
1) contrast stretch differences
2) moving shadows

You simply cannot judge anything about the color of ambient light based on monochrome images from 3 sols.

[djellison]

I would echo UG's comment. Any inference of 'colour' from Hazcam (or indeed Navcam) is pure fiction.

[ngunn]

Hang on a minute - I think this does make sense. If you illuminate something with three different colours of ambient light surely you can make some kind of pseudo-colour image. Isn't that how the Phoenix scoop camera works?

[hortonheardawho]

Very clever Fran.

It will be interesting to use the Radiometrically Calibrated images to do this exercise when they are available.

[Juramike]

I agree with Nigel. Depending on sun angle, the illuminating light would be filtered. High up in the Earth sky, it's white (well, yellow-white + the blue scattered illumination from the sky). Lower down, the light would be yellowish as more blue light gets scattered away. And then right at sunset, the light is that golden yellow.

So I suppose you could try to recombine the images knowing the illumination wavelength.

Heck, you might even be able to dig out spectral information. (Not much, since you are only varying illumination source wavelength, not detector wavelength.) At the best you might be able to say: "That object must be reddish since it stays brighter when the sun is lower in the sky, that object must be bluish since it gets much darker when the sun is lower in the sky."

Didn't they use this kind of analysis when looking at Huygens descent images?

-Mike

[Nirgal]

I thought it could be amusing to try this:


It's a composite of rear hazcams from sols 1256, 1264 and 1268

Interesting Idea: this way the three different tau ambient light conditions could possibly act like three different "filters", thus producing albeit not a true color image (of course) but still could potentially reveal interesting information as a false color image.

However, i do suspect that the different "tau-filters" would just reflect differences in overall brightness, thus unfortunately not revealing much additional "spectral structure".

To test this hypothesis I did a quick principal component analysis (PCA)



on the composite image's color space which shows that the additional "filtes" do indeed not reveal any significant additioanl "intrinsic" dimensions to the data other than the 1-dimensional brighness information.

[Juramike]

To test this hypothesis I did a quick principal component analysis (PCA) on the composite image's color space which shows that the additional "filters" do indeed not reveal any significant additional "intrinsic" dimensions to the data other than the 1-dimensional brightness information.

Wow! Nice work!

-Mike

[djellison]

If you illuminate something with three different colours of ambient light

Yup - you do. Unless someone has installed street lighting......

[helvick]

The effect might be subtle but I can see how differing tau levels would change the distribution of flux intensity at various frequencies. Certainly as Tau increases the overall color cast would tend towards "reddish" due to the increase in flux from light reflected from atmospheric dust vs the low tau scenario where the overwhelming majority of incident flux is unimpeded direct solar insolation.

The change from low tau (0.5) to high tau (2.0) changes the ambient light from a 90:10 ratio of direct unimpeded solar flux( "clear" or to be more accurate flux that has a frequency distribution that comes from our own dear Sun): reflected diffuse flux ("reddish" or whatever color reflected solar flux from martian atmospheric dust tends to be) to something closer to a 50:50 ratio.

So if you take a panchromatic image under the 90:10 illumination regime and compare it to a panchromatic image taken under the 50:50 regime then you will see different details.

Edited to add: I see Nirgal's point but I can't comment on the limits it places in our ability to see real variation between images taken under varying flux regimes as I haven't done the numbers on this one. That said I'm happy to accept that the effect is may be too subtle to detect given the 8bit contrast scale we're working with on these images but I think it should be plausible even if it is undetectable in the images in question.

[Fran Ontanaya]

Real or not, it can be gorgeous. Spirit landing raw composite:



I've got a couple of pictures more, but this is already off-topic.

Yes, it is very off-topic - Moderator

[jamescanvin]

Yes, it is very off-topic - Moderator

Ture, but this is an interesting idea, and that one of the lander is lovely whatever it really shows, great work Fran.

Split to a new thread.

[4th rock from th...]

Interesting indeed. It really seems to work to some degree, somewhat like an image made with just an orange and a cyan filter. Perhaps with some calibration and comparison with RGB images the color effect can be calibrated and/or confirmed.

[Fran Ontanaya]

The lander one is mostly or completely a lighting artifact.

The following composites are 600 days apart and both in the solar panel shadows, so they receive scattered light only.
In both, one cable has apparently a distinct hue. Saturation is stretched, though, and the jpeg artifacts are awful.



I googled around but I only found this picture with cables being manipulated in the back side (large version):

http://mediaarchive.ksc.nasa.gov/detail.cfm?mediaid=18868

[djellison]

enhanced, unaltered

Isn't that something of an oxymoron?