Uncompressed Opportunity Images

On Wed, 11 Feb 2004 17:50:57 GMT, Joe Knapp wrote:


"Mike Simmons" wrote
regarding your claim of NASA's
"witholding of the raw images"

That is the FIRST time you have quoted me accurately.

It is the third of fourth time I've quoted it so your statement above is
false.

The only negative
comment I have made about you personally is that you lack basic
expertise on
USENET, while swaggering around complainign about others lack of
expertise.

This is also untrue, but you appear to feel that words from your keyboard
are all statements of objective fact while similar words from others' are
"rude" or otherwise tainted by emotions. The reactions to you would seem
to indicate otherwise.

Please quote a statement from me in which I complained about someone
else's expertise. I don't believe I ever made any such comment. I asked
for evidence to support your claims. My statements might have been taken
as a complaint about making unfounded statements and accusations and I
have no problem with that. But lacking evidence for an accusation is not
the same as lacking expertise.

Considering the purposes of this science-related newsgroup I don't think
that requesting supporting evidence for claims is unreasonable. I doubt
that many here would agree that it constitutes "swaggering". I don't.

Considering the number of people who have had similar reactions to your
statements, a "reasonable" person might consider whether he was doing
something to generate the negative reactions. And a wise person
definitely would consider it. IMHO, of course.

Mike Simmons

"Greg Crinklaw" wrote
True. But your comment smelled like you were accusing them of
withholding data for a sinister reason.

Well, let me just say that there are certain popular debates on this ng
(witness by the number of replies that say a post by Dr. Min gets) that
certain people like to respond to with stereotyped replies, whether they fit
or not. Call it a skeptical vanguard. That can lead to a certain browbeating
over any criticism, pigeon-holing, and over-deference to whatever is
perceived as authority. That is kind of stifling. The body of knowledge
about Mars is still not great, and again, without off-base ideas it would
have no history at all. So the people with nutty theories are not really the
problem. I see no reason to hammer the Hoaglands or whomever--24/7, anyway.

I saw some of your earlier color reproductions. They looked good. How
are those coming?

Interesting composites can be made, but it's so hard to get totally
satisfactory results. Seems like the goal ought to be an objective algorithm
to feed all the filtered images of a scene into which unambiguously spits
out the best color match, no human judgment or ad hoc fudge factors
involved. That is impossible to do just by grabbing three images, RGB, as I
was doing. There's too much, human judgment--"Well that's too blue, let's
crank down the blue," etc. Too many assumptions, bound to create a
conservative result. Slightly better would be to take all five filters in
the visible spectrum (roughly blue, cyan, green, orange, red) and combine
them all to get the RGB values. I have no software to do that. But even then
the balancing process would still be somewhat heuristic.

The color calibration target doesn't really help that much, at least on the
first level. Since the spectrum of the Mars skyglow and direct sun at any
given time are not really known, who's to say what colors the color chips
should be calibrated to? To use an exreme example, if the light was pure
red, then all color chips should be varying intensities of red. An effort to
make sure the blue chip is blue, red is red and so on is forgetting the goal
to render the colors as they appear, not as they somehow intrinsically are.
Not to mention that the chips have a complex response to light at different
wavelengths. For example, the blue chip is almost as reflective as the red
chip at the L7 ("red") wavelength.

However, I wonder if the sundial doesn't after all provide a feedback loop
for a mechanical algorithm to compute the best fit color. That is, the
mirrors give a spectrum of the sky at the current balance parameters. The
gnomon shadow also allows a double-check of skyglow spectrum, as well as a
means to estimate the direct light spectrum. So the algorithm could assume a
reasonable lighting situation to begin with, compute the expected apparent
colors of the chips based on that illumination, then using photos from all
five visible filters calibrate the balance parameters to get the photo as
close as possible to those chip colors, and then read out the new derived
sky and sun spectra. Feed those back to refine the initial estimate of the
illumination spectra. Repeat.That process might converge on the best
solution automatically. End result would be the spectra of the sky and the
direct sun at the scene.

But that is a lot of work. Also, it might be that the sundial is after all a
bit too crude to attempt this kind of data mining.

I've tried my hand at a few using Photoshop with
pretty good results. I was able to mostly reproduce the calibration
colors by reducing the blue channel to 70-75% and the green channel to
80-90% (relative to the red). But I still get a green that's a bit too
bright, even if I cut down the green channel more. The opportunity
images are easy (less color) but whenever I do a Spirit image I seem to
get a wide variety of garish colors. Whether they are correct or not,
it seems clear to me that the Spirit site is much more colorful.

I noticed the same thing, which is why I soured on a simple solution
manipulating three color channels in Photoshop. To get rid of the greens and
blues you basically have to just nuke those channels. Then everything's too
red. There seems to be no linear solution for three colors. Which leads me
to believe that the colors on Mars are a lot weirder than shown in color
photos to date.

Joe