Is The Mars Rover Cam Life-Blind?, page 3

Originally posted by ArchAngel
It is not impossible for there to be life, and it was one thing that the cam is blind to.

What else can the cam not see?

Continuing to repeat something that is false, does not make it more true. You've been presented with an opportunity to educate yourself on this subject, yet you seem to be going to great pains to AVOID actually understanding what's going on. Why is that?

Your use of the term "blind" is grossly inappropriate in this case, and your citations about the comparisons of "RGB" images to "true color images" is quite simply a misinterpretation of what we are trying to deal with here.

We are dealing with the subset of colors that can be displayed in a quantized fashion. Since we are creating images for display with computers, it is patently obvious that there are colors that your computer can never display.

Recognizing that fact does not mean there is some vast conspiracy, nor does it mean that computers are an inappropriate means for examining color data. There is such a thing as reasonable approximation of photonic data using other means (like computer displays).

Once you recognize that what NASA is trying to build is data for display on COMPUTER SCREENS, then it should become clear that, out of the INFINITE number of color mixes that could strike the human eye, we are really only interested in the 16 million or so points in the COMPUTER RGB DISPLAY SET.

We are looking for mappings that, when displayed via a computer screen, will stimulate human eyes in such a way that the generate meaningful signals for our brains.

To do that, computers need to generate only THREE POINT FREQUENCIES out of the infinite number of naturally-occurring frequencies present in light that hits human eyes in the everyday world.

Stop and think about that for a minute.

Computers can display 16 million different colors using only three distinct frequencies. By your logic, if I created a set of filters that blocked out everything but a narrow bracket around those three frequencies, and took Pancam photos of a laptop screen, I still couldn't see "color", because look at all of the frequencies I couldn't see, and no overlapping bands to boot!

We are looking at a reasonable compromise between what a human eye can see, and what a computer screen can display. Of the two, the computer screen is the one that is far more limited in capabilities.

The functionality of the Pancam was not built around trying to mimic human eyes. Instead, it was built around trying to gather light REFLECTANCE data from objects using a variety of bands. Each of those bands provides information of different kinds, and some of those bands can be combined to provide RGB signals FOR COMPUTER SCREENS which will carry enough data to be "close enough" to what a human would see... and that makes them quite useful.

There are an infinite number of points in the color-mix space that can be seen by human eyes. Computers can display only a tiny fraction of that (16 million is a hell of a lot less than infinity...) Yet I don't hear anyone crying about the fact that computers are somehow grossly inappropriate tools for trying to do analysis of "color" images in order to extract meaning from them.

If you take the opportunity to educate yourself on hyperspectral imaging, you'll see that it is not that uncommon to have a variety of fairly narrow-bandpass filter images for CCD systems being used to pull LOTS of good data together, despite the fact that they don't necessarily measure every point in color space, and in almost every case fail to overlap their bands.

Chlorophyll is a good example. I've already explained in quite some step-by-step detail how a composite using the L3-L5-L6 channels for (computer) RGB would show a highly convincing green signal in a Pancam shot of plant life.

You act as if my explanation was never even offered.

The term "RGB" means different things in different disciplines. In the context of this conversation, you'll need to remember that the ONLY context which is appropriate for the purposes of this discussion is computer screen RGB, not human-vision RGB.

As Kano outlined, human vision has three overlapping response curves. The quote that you keep trotting out in various places refers to human vision response curves, not color display using computer monitors.

Computer monitors display color solely via the additive application of three different, distinct frequencies, at varying degrees of intensity. That's it. Your computer monitor has no "overlapping curves" either.

Which conspiracy caused that?

You could take a multispectral shot of a decent LCD display showing a rich color scene, but when you decompose it, you'll see three incredibly narrow spikes.

Since the Pancam is trying to generate data that a computer screen will use to create images, the question is definitely NOT "which frequencies can't the Pancam see", because at least one filter can see EVERY frequency in the human range at some level of brightness... as I already pointed out to you, but you apparently keep ignoring.

The more appropriate question is whether the Pancam filters can be used in appropriate combinations to create COMPUTER IMAGES that are reasonably useful.

And they most certainly can. Will they be identical to what your eyes would see? Of course not... but your computer couldn't display that anyway. So instead, we are left with trying to come up with reasonable points in the data space that "come close enough".

Of course, the other advantage is that this gives the scientists even more useful combinations than some theoretical "true color" method, since they can combine a wide variety of combinations beyond just the typical 3-axis data used by computer monitors.

We can look at deductive comparisons of the different frequency outputs, and determine useful things about given sets of pixels that go far beyond what a human eye could ever determine. It's a nice side effect that we can produce reasonable facsimiles of human-friendly color schemes along the way.

Understanding how to use the data means knowing when other similar-sounding concepts are simply INAPPROPRIATE, and this is typically when they are wrongly applied to a different discipline.

As you said elsewhere: "I learned about this before digital, and am a novice with Photshop."

Maybe that should act as an indicator to you that you might not understand everything that's involved here... and that your lack of understanding should not serve as support for (or constitute evidence of) some conspiracy on the part of the Pancam designers.

Until you understand why chlorophyll looks green to human eyes, and how a computer goes about representing that, and how the Pancam can produce a computer image that indicates that... you simply won't understand whether the tool is reasonable or not.

An infinite number of mismatched-context quotes will not change that in the slightest.

Learning, however, can change it. If you have questions (instead of off-base proclamations) that will help further your understanding, please feel free to ask them.

Without that, I think you're simply basing your position on a combination of ignorance and confusion.