Pages: [1]   Go Down

Author Topic: Accurate colours in planetary astrophotography  (Read 839 times)

Tulloch

  • Newbie
  • *
  • Offline Offline
  • Posts: 7
Accurate colours in planetary astrophotography
« on: November 23, 2020, 01:28:49 am »

Hi all, this is my first post here, so my apologies if this is not the right forum, or even the correct website, just hoping someone understands what I'm trying to do.  :)

A couple of years ago I got interested in astrophotography (AP) of the planets using my Canon 700D, and was told to use the "daylight" setting for white balance, after all the planets are lit by the sun. After using this camera for a while, I upgraded to a dedicated AP camera, the ASI224MC, a high sensitivity colour camera capable of very high frame rates. After using this camera for a while, I naively asked, "What white balance settings should I be using for this camera?". After not obtaining a clear answer from the AP set, I set off on a quest to find the answer.

After much reading and testing, I eventually was able to use ArgyllCMS/CoCa on an image of a Macbeth colour chart captured in approximately D50 lighting conditions to produce a colour profile icm file and colour correction matrix, which appeared to do a pretty good job (on the colour chart, anyway ;D), but I'm unsure as to whether the profile is working as it should be on the planets (see example below)
https://www.cloudynights.com/topic/735310-color-calibration/?p=10615644

The capture details were as follows
https://www.cloudynights.com/topic/697940-planetary-imaging-and-colour-accuracy/?p=10046396

The issue is that the ASI224MC camera captures the light in linear format (ie with no gamma correction applied), which is saved as an 8 bit avi/ser video file, the "best" frames of which are then stacked to produce a final image in 16 bit linear format. The capture software is able to control the gain of the camera to achieve set exposure levels and a gross red/blue correction factor. It just so happened that in the image of the Macbeth chart I captured using this camera, the (r,g,b) values of middle grey came out as (48,49,48) in linear space, or (128,126, 125) in sRGB space after applying the icm profile.

FYI, the colour matrix from CoCa/Argyll came out as:
 0.89509583   0.26640320  -0.16140747
-0.75019836   1.71350098   0.03669739
 0.03889465  -0.06849670   1.02960205

My real question is how can I apply this profile correctly to an image of a planet? I believe that the colour matrix I measured is only valid when the exposure value of the colour checker chart image is the same as the target image, but I have no way of taking an image of a grey card standard at night to get the correct exposure. Or, am I mistaken, is this even necessary?

Interested in any thoughts, or pointers on a way forward.

Thanks,

Andrew
« Last Edit: November 23, 2020, 01:37:51 am by Tulloch »
Logged

Lessbones

  • Full Member
  • ***
  • Offline Offline
  • Posts: 122
Re: Accurate colours in planetary astrophotography
« Reply #1 on: November 23, 2020, 02:16:40 pm »

This is an interesting experiment for sure, and while I have absolutely no experience whatsoever with astrophotography, I guess my first question would still be: what do you intend to do with these images?

If you're displaying on a screen or in print form (not sure how else you could possibly display... anything...) then you're also going to have to contend with a lot of different issues of spectral response and relative contrast that come from these display methods...  so is the goal to make these look pleasing to the eye?  Or is it to measure the exact spectral response?  In the case of the latter, I don't believe you can even display these in a way that satisfies, since you can't truly replicate the sun's rays, so you'd really only be able to present a series of mathematical values.

If you're trying to display them as they "look" in the sky, then I would just adjust your a* and b* values to taste with color temp and tint and be done with it-- but it all comes down to what your endgame is IMO
Logged

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 19893
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Accurate colours in planetary astrophotography
« Reply #2 on: November 23, 2020, 02:27:48 pm »

True Accurate color is often not pleasing color.
Let me see if I can succinctly provide guidance between color accuracy and pleasing color.

Pleasing color has one goal: make color that is pleasing. Usually to the image creator. It has nothing really to do with color accuracy both in its goal and in how it is reported.

Pleasing color IS subjective. We cannot measure nor report how pleasing the color is to anyone.

Color Accuracy is not ambiguous, the results are based on measurements and analysis whereby the accuracy of two colors, the color photographed at the scene, and some resulting digital color number, output somewhere, are compared.  Then an accuracy metric (deltaE) can be provided. Stating the accuracy between two solid colors is a dE 2000 of 4 is just as precise as saying my backyard wall when measured, is 6 feet tall.

Accurate color is more often not pleasing to look at. Scene referred color is often very ugly and why it is output referred. Examples and explanation here:

http://www.color.org/ICC_white_paper_20_Digital_photography_color_management_basics.pdf

Pleasing color is simply arrived when someone states what they are seeing is pleasing.

Accurate color requires measurements and analysis and a numeric report of the accuracy.

I can measure the color of an object at the scene someone is photographing. I will get a spot measurement of one color. That one color may take up 000.1% of the resulting image or 100% of the resulting image. I can compare that measured number or many more, and the number or many numbers produced on the computer or resulting print and produce a color accuracy measurement. For each color, for an average of all the colors. And if that one color is say 1% of the image that tells us 1% about the color accuracy. We rarely see such images in photography.

We define all kinds of accuracy by measuring things: distance, height, weight, output of an illuminant, heat, and color. We don’t (we should not) subjectively report any of the above, we should measure them. We have instruments to do this. There is no need to assume or guess! There are reasons we use thermometers, rulers, and Spectrophotometers instead of assuming a measurement. Then stating incorrectly it is accurate by assumption.

If you pay to have a wall built in your back yard, and you ask for it to be 6 feet tall, you can buy a $3 wooden ruler at Home Depot and measure that wall. If you find it is 6 feet 1/2 inch tall, you know the accuracy in how tall it was built. You might use a far more expensive laser measuring device and find it is 6 and 5/1000th of a foot but if your goal is a 6 foot wall not costing tens of thousands of dollars for the added accuracy, the first wall, and measuring device are fine for the task. If the wooden ruler shows the wall is 5 feet, you need to decide about having a conversation with the builder! 5 feet or 5 feet 1/10000 of an inch, something is wrong.

The third category is a tough one: Matching Color. It’s very, very hard to do in many cases. It takes a bit of pleasing color and maybe accurate color into this mix.

You shoot a Macbeth 24 patch target. The goal is a print from an Epson that matches it. The colors might be accurate (color in equals color out) but probably not. Some color patches may match, and some patches may not. Because of a slew of issues: the illuminant and its spectrum, the illuminant and it’s interaction with OABs in that paper. The white of the paper. The surround in which both the target and print are viewed. The observer (maybe he/she has cataracts, is red/blue color-blind), the issues of both observer metameric failure and other metameric failure in the process from capture to print. Digital cameras do not ‘see’ and record color like humans do. Color Matching may require hugely inaccurate colors to produce a visual match!

There are photographers who take pictures and fewer that make pictures but everyone in these parts fancies themselves photographers. Some say they want “accurate color“  without understanding what they want. Or what they mean. Some understand they wish for pleasing color for them and hopefully their audience. Some buy into the marketing hype that someone is selling something and promising their product will produce “accurate color“ . And photographers who don’t understand this think they do want and need and should pay for “accurate color“ . Neither they nor the people selling this stuff can define what they mean by accurate color. So now *some* photographers here can see what a silly notion buying into that concept is. And why. And if they really do want accurate color, that they need measuring devices and software to report accurate color at the scene of capture and then thereafter.

I’ll submit again as I have in the past, and which some here admittedly forgotten: Most photographers simply desire and should attempt to produce pleasing color (or non color) rendering. That’s been the norm for the nearly 200 years of phtography.
Logged
Author “Color Management for Photographers" & "Photoshop CC Color Management" on pluralsight.com

Lessbones

  • Full Member
  • ***
  • Offline Offline
  • Posts: 122
Re: Accurate colours in planetary astrophotography
« Reply #3 on: November 23, 2020, 02:32:16 pm »

^^^  a much, much, much better version of what I was attempting to say
Logged

Tulloch

  • Newbie
  • *
  • Offline Offline
  • Posts: 7
Re: Accurate colours in planetary astrophotography
« Reply #4 on: November 23, 2020, 03:42:48 pm »

Hi there, thank you both for your replies here. This Quest for Colour Accuracy in my planetary photography has been a long one, if you have a few hours you can read about the steps I've taken on my quest ;). I've slogged my way through the process for colour calibration of my monitor, using Registax's auto-balance feature, using G2V stars as colour calibration targets, measuring the colour coordinates of an actual (purchased) Macbeth colour chart, finding the "best" values for the AS224MC camera for correct white balance settings, measuring the effect of planet elevation on colour cast and using measured spectral data as measures of "colour truth" for calibrating image data. My second last attempt is shown below here with the steps I've taken to use the colour profile I created in Gimp shown here (along with a few others interested in this topic) shown here.
https://www.cloudynights.com/topic/735310-color-calibration/

While I understand the difference between "pleasing" colours and "actual" colours, my aim here is to somehow show images of the planets on a colour calibrated screen in all their real (hideous) colour glory. I believe it should be possible that with the right colour correction matrix set up for these cameras to display the corrected images. The differences between scientific accuracy that you can obtain using a spectrophotometer on the planets (which can and has been done) and making pretty pictures (which is what NASA does) is significant. All I would like to be able to do is create a white balanced image with this camera and show the planets as they actually are, pretty or not.

I also understand that not everyone will perceive the images the same way, due to their own quirks in visual range, red/blue colour blindness, incorrectly setting up their own monitor, whatever. That bit doesn't really interest me, what I am interested in is trying to measure the "true" colour and display it accurately on a calibrated monitor. Just because I may perceive it to be more yellow than you because the sensitivity of the cones in my eye are slightly different to yours, doesn't matter to me. We will both see it as "accurate", even though we may perceive it differently.

Given a reflectance spectrum of a planet measured over the visible range of wavelengths, then it is possible to convert that measurement into an "average" RGB value on a calibrated monitor using the CIE 1931 XYZ tristimulus values (which is what I've already done). If I can take an image of the planet that matches these values, then I should have achieved some sort of colour accuracy.

I guess what I'm asking is, how do I apply my own white balance settings to a camera that has no concept of white balance, gamma or sRGB and simply outputs the raw sensor data in linear format? In other words, I want to do what Canon/Nikon/Sony/etc do inside their cameras to convert the raw, linear sensor data into an sRGB gamma corrected image that can be interpreted by something like Photoshop/Gimp with an embedded colour profile. What makes it even more difficult is that I have no 18% grey card target available in space to match my exposure level to the one I took of the Macbeth colour chart.

Any advice appreciated.

Andrew
« Last Edit: November 23, 2020, 08:12:48 pm by Tulloch »
Logged

Alan Klein

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 15850
    • Flicker photos
Re: Accurate colours in planetary astrophotography
« Reply #5 on: November 23, 2020, 08:19:45 pm »

What does my Epson V850 scanner do with the target color slide I scanned in as a basis for color matching when I scan in my landscape film shots?

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 19893
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Accurate colours in planetary astrophotography
« Reply #6 on: November 23, 2020, 08:35:59 pm »

What does my Epson V850 scanner do with the target color slide I scanned in as a basis for color matching when I scan in my landscape film shots?
It makes a scan of the target....
You want to make a scanner profile?
Logged
Author “Color Management for Photographers" & "Photoshop CC Color Management" on pluralsight.com

Alan Klein

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 15850
    • Flicker photos
Re: Accurate colours in planetary astrophotography
« Reply #7 on: November 23, 2020, 08:39:36 pm »

It makes a scan of the target....
You want to make a scanner profile?
I have no idea.  The scanner allows automatic adjustments, flat scans or you can select this color slide as a reference target.  So what is happening if I select it while scanning let's say Velvia 50?  Is it trying to match the color target slide's color or what?

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 19893
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Accurate colours in planetary astrophotography
« Reply #8 on: November 23, 2020, 08:46:30 pm »

I have no idea. 
Indeed.
Quote
So what is happening if I select it while scanning let's say Velvia 50?
You're getting a scan of Velvia 50.
You have a manual with the software?
Logged
Author “Color Management for Photographers" & "Photoshop CC Color Management" on pluralsight.com
Pages: [1]   Go Up