PDA

View Full Version : "True Colour" Mars Images from MER team



Jack Higgins
2004-Jan-21, 11:24 PM
AHH!!! :evil:

Firstly- NASA, JPL- you're doing a great job with the Spirit at the moment. Hopefully Opportunity will do just as well when it lands. But there's one thing that you're doing that REALLY ticks me off... Those "true colour" images you're releasing.

Today's, for example:
http://marsrovers.jpl.nasa.gov/gallery/press/spirit/20040121a/Lander_Pan_Sol16-A18R1_br2.jpg
Really nice image... But if you look at the lower right petal of the empty lander, you'll see the NASA meatball logo, and the JPL one.

Now, correct me if i'm wrong, but the NASA meatball is BLUE. Not RED. Why oh why do you keep boosting the red channel in all the released images- that so obviously show them not to be true colour?!! I've even seen images with the calibration target in view, where the colours are well off.

I would put together R G and B channels myself in Photoshop, but I haven't been able to get it working after many attempts. I don't believe any of the Hoagland-style conspiracy coverup of Mars' "real" colour- I just find it REALLY irritating!


...

On the other hand, all colour surface images of Mars look fantastic, but I prefer "sandy brown" to "deep red".

Somebody help, please... :roll:

Sparks
2004-Jan-21, 11:40 PM
The calibration target hasn't been "off", it's just that the pigments in the target are bright at several specific wavelengths. Combine that with the fact that the CCD cameras are grayscale, not colour, and the filters used, and you get odd-looking composite shots.

And that photo you posted was shot with infrared instead of red for the R channel.
If I'm remembering correctly from the press release...

George
2004-Jan-21, 11:50 PM
I would put together R G and B channels myself in Photoshop, but I haven't been able to get it working after many attempts. I don't believe any of the Hoagland-style conspiracy coverup of Mars' "real" colour- I just find it REALLY irritating!

I am surprised also as, by now, I expected at least some near true color rendering to be available. Very frustrating to have such great resolution knowing it is not what you would really see.

The horizon view is interestingly false also, or, did we find Martian high-rise apartments on the horizon? :)


On the other hand, all colour surface images of Mars look fantastic, but I prefer "sandy brown" to "deep red".


Yep. They all still deserve a standing ovation.

I think that would be "maroon" so Aggies (possibly Alabamans) may be in charge of the color rendering. :)

Mokele Mbembe
2004-Jan-21, 11:57 PM
Check this out for an in-depth explanation of the color changes and how you can change them to "real color".

http://www.atsnn.com/story/30048.html

Phobos
2004-Jan-22, 01:37 AM
Typical RGB values for recording and display are Red-600nm, Green-530nm and Blue-480nm. As we can see these coincide with the L4, L5 and L6 filters on the PanCam. The difference is, in this panorama image, and in most images taken by the Rover, the L2 is used for the Red-Channel instead of the L4.

This is the section of the document that explains the problem. The key point they are saying is that if they are releasing images to the public they should be using the L2 filter rather than the L4 filter if they want as close a match as possible to what would be seen with the human eye.

By using the L2 filter they are producing the colour alteration effects that are seen on the blue coloured objects.

However, I would say that it's about time they released some images that use the L4 filter.

Phobos

rwald
2004-Jan-22, 01:59 AM
Well, aside from some specific pigments that are very brignt in the IR spectrum but not in the red spectrum (blue seems to be a prime example of this), the pictures are basically what your eye would see. Actually, if they released pictures with the L4 filter, the images would appear less red, not more. This is all discussed in much greater detail at the site that Mokele Mbembe referenced.

Oh, and for those who have read that site: I've discovered that all the 3D landscapes that the JPL has released (like this one (http://marsrovers.jpl.nasa.gov/gallery/press/spirit/20040112a/airbag_drag_mark_3d-A10R1_br.jpg) or this one (http://marsrovers.jpl.nasa.gov/gallery/press/spirit/20040119a/Viz_Adirondack-A16R1_br.jpg)) do not compensate for the Spirit rover's auto-normalization of the color channels (called "hard limiting" in the article). That's why those images have different colors from the main body of images.

George
2004-Jan-22, 03:49 AM
Check this out for an in-depth explanation of the color changes and how you can change them to "real color".

http://www.atsnn.com/story/30048.html


Thanks, thanks, thanks. Looks like you found the best answer. =D>


Dr. Bell [ of Cornell U.] also touched on this issue.

quote:
--------------------------------------------------------------------------------
Ultimately you are right that people will need the calibrated data to do the color balancing correctly. We are working on doing that and will eventually get all those images out to the public using the NASA/JPL "Planetary Image Atlas" web site. It will take several months or more to get the work done, however. In the meantime, we thought it would be best to get *something* out there, and so that's why we opted to get the raw data out fast, even though it's still raw. The team has taken some criticism for this within the planetary science community because not many past missions have adopted such an open-data policy.

snabald
2004-Jan-22, 04:55 AM
I noticed that NASA seems to be very fond of the L2 filter myself.
I come across some stuff in the Celestia forums about how you could combine the raw images on NASA/JPL's site to make color images but I always noticed the problems in the blue channels, then I came across this page: http://www.keithlaney.com/spirit_color_images_calibration.htm which not only explained why the colors were coming out wrong, but also how to make your own "closer to true color" images using the data provided.

It seems the L2 filter has more scientific us than the L4 (which when combined with the L5 and L6 give very nice pictures).

also here are some color corrected images made by the fellow who made the above explaination... http://www.keithlaney.com/spirit_color_images.htm

rwald
2004-Jan-22, 05:12 AM
Those images all posess the color problem I hinted at earlier regarding Spirit's normalization of the color channels. I'll go into some detail here about it, to explain why Mars really isn't that blue.

As you've probably gathered by now, the Pancam is basically a black-and-white CCD with a series of filters. To create color images, filters of the appropriate colors (red, green, and blue in an ideal world; near-infrared, green, and blue in most of the JPL images) are used to create a series of black-and-white images which are later combined. One important thing to note here is that there isn't necessarily an equal amount of near-infrared, green, and blue in any given image. However, before Spirit sends back the pictures, it streaches each one so that it spans the full brightness spectrum. That is, for each black-and-white image it has, it redefines the brightest point as #FFFFFF (pure white) and the darkest point as #000000 (pure black). This way, the absolute maximum of information is contained in the image. Unfortunately, this causes problems when one attempts to recombine the images later. What is defined as #FFFFFF in the blue-filter black-and-white image may not be as bright as the #FFFFFF in the near-infrared-filter black-and-white image. If one just blithly recombines all of the black-and-white images, the colors come out wrong; in particular, the rocks and sky look too blue. Without knowing the absolute brightness of any of the channels, we can't actually tell the proper mix of near-infrared, green, and blue to approximate human vision. So any sites complaining about blue skies or rocks is misinformed.

Again, I strongly recommend reading this article (http://www.atsnn.com/story/30048.html) for more information; it is far more cogent than I am, if a bit longer.

Sparks
2004-Jan-22, 09:03 AM
Check this out for an in-depth explanation of the color changes and how you can change them to "real color".
http://www.atsnn.com/story/30048.html
Thanks, thanks, thanks. Looks like you found the best answer. =D>


Ooooh!
My thunder stolen! :D

TrAI
2004-Jan-22, 12:50 PM
Hmmm I see you already have gotten the primary answer, but I thought to mention that these are press release images, and as far as I can remember, none of the texts accompanying the pictures claim they are true color... These pictures are not meant for scientific use you know, its more like eye candy...

Sparks
2004-Jan-22, 01:08 PM
But it's just the one piece of eye candy I want, and for $550 million, it doesn't seem so much to ask for a single true-as-you-can-get-it colour 360 take of the area :D

rwald
2004-Jan-22, 01:10 PM
Actually, many of the press-release ones do claim to be true-color. They achieved that through mixing the L2, L4, and L5 filters and knowing enough to account for Spirit's equalization of the different color channels. Unfortunately, we don't know enough to do the equalization properly, so we can't verify most images. There is an exception, though: Any image which has a region which is supposed to be white will not have the equalization problem. So if you want to verify the colors in one of those, be my guest. Otherwise, we'll just have to trust that NASA adjusted the images correctly.

TrAI
2004-Jan-22, 03:42 PM
Actually, many of the press-release ones do claim to be true-color. They achieved that through mixing the L2, L4, and L5 filters and knowing enough to account for Spirit's equalization of the different color channels. Unfortunately, we don't know enough to do the equalization properly, so we can't verify most images. There is an exception, though: Any image which has a region which is supposed to be white will not have the equalization problem. So if you want to verify the colors in one of those, be my guest. Otherwise, we'll just have to trust that NASA adjusted the images correctly.

I have looked at the texts over on the mars rover site, and some do say "approximate true color"(I had not noticed that), and some says just color... Hmmm.....

Doesn't using white areas for color correction only work when you know the light illuminating it has a uniform distribution of color components, or accurate rendering of color is not very important?... I would think that if the light is colored, adjusting the levels of the different components so that one gets white would skew all colors.

rwald
2004-Jan-22, 06:37 PM
The color of the light is somewhat important, but the biggest problem in most of the images is that "white" in the near-infrared is actually brighter than "white" in the green channel. By having a true-white object in the image (like the reflection of the sun on the low-gain antenna), we know that there should be a spot of "true" white in all of the color channels, so we know that the near-infrared's "white" is the same brightness as the green's "white."

In other words, yes, it is a problem, but it's nowhere near the biggest problem.

George
2004-Jan-22, 07:03 PM
Doesn't using white areas for color correction only work when you know the light illuminating it has a uniform distribution of color components, or accurate rendering of color is not very important?... I would think that if the light is colored, adjusting the levels of the different components so that one gets white would skew all colors.

Great point. Suppose the atmospheric dust augments reddish or pinkish light. A "white" object would not look so white.

Although unlikely, suppose the Sun's true color is actually "peachy pink" or blueish. SADly - it's not well known. :cry: :)

The martian atmosphere may not flatten the Sun's irradiance in the visible spectrum as our atmosphere does. Earths' pre-atmosphere blue seems to be about 2x the intensity over post atmosphere blue.

A monchromatic white ( good grief, an impossible oxymoron :) ) light on Spirit used at night on the color tabs for calibration might work nice..... ok, ok. A calibrated white light (known spectrum). Anyway, you know what I mean.

SciFi Chick
2004-Jan-22, 08:13 PM
Speaking of true color, has CNN lost it? I mean, the capital never looks like this. :lol: I'm only posting this due to the red coloring in the picture.

http://www.cnn.com/2004/ALLPOLITICS/01/22/spending.bill.ap/index.html

George
2004-Jan-22, 08:36 PM
Speaking of true color, has CNN lost it? I mean, the capital never looks like this. :lol: I'm only posting this due to the red coloring in the picture.

Hmmm. Maybe the sun is peachy pink! :)

Your comment reminds me....

During the State of the Union address, I surfed all the channels to get a comparison of the color imaging since they all used the same camera. It was surprising the big difference I found in color. Most notably in tint as skin color changed dramatically from channel to channel.

Part of my frustration is the wonderful and very close-to-true-color pictures we get from $200 digital cameras. If it weren't for these things, the reported problems with true color imaging would sound less like excuses. Of course, they still have my respect and trust, as well as, admiration for what they do give us and a little time is warranted.

Dancar
2004-Jan-22, 09:10 PM
What's the big deal of the exact color balance of the Martian pictures?

And what exactly does "true color" mean anyway? Have you ever noticed photographs taken in florencent lighting that had a green tinge? That's because film is designed to match the color balance of sunlight filtered through earth's atmosphere, and florencent light IS green tinged. We don't notice it much because the human vision system can rebalance color to some degree. Have you ever put on yellow ski-googles and noticed the colors going from yellowish to normal in about a minute? And when you take them off, everything is blueish at first? Same thing!

So even Kodak color film used on Mars may not reproduce the same color balance that a naked eye would see, because the Martian atmosphere would filter sunlight diffrently. And what's the point of JPL jumping through hopes to make the colors appear the same way they would to a naked human eye at the surface of Mars when a human on Mars not wearing a space suit & helmet would die very quickly?

So the only issues really for coloring Mars pictures are aesthetics and making detail and texture easy to descerne. I think the JPL techs are doing a great job at that.

Dancar

Rift
2004-Jan-22, 10:27 PM
What's the big deal of the exact color balance of the Martian pictures?

Beats the daylights out of me, I'm still trying to figure out the woowoo's angle here. One site I read said NASA wanted to make Mars look less life-like then it really was, lol.

The really funny thing is to make it look like a human eye would see it NASA really would have to alter the color of the images. So far the only thing i can see they are griping about is that the blue swath on the sundial isn't blue... Who the heck is looking at the sundial??? Who cares???

It's NOT a human eye up there.

rwald
2004-Jan-22, 10:33 PM
...Not to mention that the sundial complaint is the easiest one to counter.

Sigma_Orionis
2004-Jan-22, 11:05 PM
What's the big deal of the exact color balance of the Martian pictures?

And what exactly does "true color" mean anyway? Have you ever noticed photographs taken in florencent lighting that had a green tinge? That's because film is designed to match the color balance of sunlight filtered through earth's atmosphere, and florencent light IS green tinged. We don't notice it much because the human vision system can rebalance color to some degree. Have you ever put on yellow ski-googles and noticed the colors going from yellowish to normal in about a minute? And when you take them off, everything is blueish at first? Same thing!

So even Kodak color film used on Mars may not reproduce the same color balance that a naked eye would see, because the Martian atmosphere would filter sunlight diffrently. And what's the point of JPL jumping through hopes to make the colors appear the same way they would to a naked human eye at the surface of Mars when a human on Mars not wearing a space suit & helmet would die very quickly?

So the only issues really for coloring Mars pictures are aesthetics and making detail and texture easy to descerne. I think the JPL techs are doing a great job at that.

Dancar

Thank you, you have explained in three paragraphs something that has puzzled me since the rover pictures came in, while I understand that a camera sees things different than the human eye I couldn't understand in the middle of the talk about color filters and the like the how the pictures were taken and why all the talk about the color of them

George
2004-Jan-22, 11:32 PM
What's the big deal of the exact color balance of the Martian pictures?

PR for the most part. The key is to get people to mentally grasp the magnitude of what they are seeing. Viewers are looking through the eyes of Spirit finding themselves on the surface of another planet. NASA/JPL has accomplished something truly fantastic! Billions of people will likely see these images.

PR is critical to funding. It would be nice not to have to say that what we are witnessing is really not how it really is. If we knew they were very close, there would be little to talk about.

Look again at the sky in the original post. Is that a blue sky which should exist due to Rayleigh Scattering? But then there is Mie Scattering to consider (and others) due to dust particles. Also, there is the sky color due to the dust color itself. There may be other varaibles.

According to Dr. Bell, we will be getting color corrected imaging at a later date due to their obvious work load. I am content with that answer.

In the mean time we can have fun guessing the real color. Hmmm

JonClarke
2004-Jan-26, 05:10 AM
The colour of Mars is being discussed at

at http://uplink.space.com/showflat.php?Cat=&Board=sciastro&Number=647498&pag e=0&view=collapsed&sb=5&o=0&fpart=7&vc=1#Post64750 8

There is the usual NASA conspiracy stuff there, and great excitment over blue pacthes in Mars Express images.

Jon