Old 18-11-2014, 08:45 AM
irwjager's Avatar
irwjager (Ivo)
Registered User

irwjager is offline
Join Date: Apr 2010
Location: Melbourne
Posts: 512
Originally Posted by Shiraz View Post
my basic problem in understanding it is this: if colour is processed separately from brightness, then surely dynamic range must be affected - I can't see how you can simultaneously have full dynamic range and deep saturation. For example, if a part of an image is saturated red, that means that the blue and green emitters in the screen are turned off. Ignoring what the eye does, this means that fully saturated colours can never produce more than 1/3 the brightness of white regions (where all 3 emitters are on). Once you get above 1/3 max scene brightness, you cannot display deeply saturated colours and the brightest parts of an image can really only be white if you want to use the full dynamic range of a display. Have I got this right and if so, what does it mean for colour in astro images? I think that it means that the brightest parts of an image (eg star cores) should not be coloured, but am I missing something here?
It's a very good question indeed! And the very short answer is that you're right.
Colour is a very hairy subject and we're getting really down into the nitty gritty of colour spaces here.

When we look at how the human eye works, it's actually even *worse* than you suggest; instead of an nice even 1/3, 1/3. 1/3 contribution to luminance, a fully saturated blue (and blue only) pixel peaks at around 12% of *perceived* maximum brightness. Red is about 30% and green (to which the eye is most sensitive) peaks at the remaining 58% (interesting side note; this sensitivity to the green channel is also why there are 2 pixels allocated to green in your DSLR's bayer matrix, it is also the reason why red is more heavily compressed in JPEGS, etc.).

Firstly, due to this discrepancy in sensitivity to the 3 different channels, we need to start working in a different colour space (CIELab is the space of choice) which can account for the psychovisual peculiarities of humans. E.g. it's a colour space that offers perceptual uniformity where an equal measure of change in colour value should cause an equal measure in change of visual importance (the latter roughly being a different way to say 'luminance'). In short, working in this space, allows us to define perceived color as seperate from luminance. This obviously allows us to do all sorts of cool stuff like process color and luminance separately. The cool (and simultaneously problematic) thing is that the L*a*b space exceeds the gamut of the RGB space; e.g. it can represent more colours than an typical RGB screen can render in response to setting R, G and B for a pixel. Not only that, it allows whiter-than-white and blacker-than-black values as well. Converting back to R, G, B obviously will have to involve mapping some of the values that R, G, B cannot represent to alternative R, G, B values that fall within the range that the RGB space *can* represent. This will indeed cause some R, G or B values to max out ('saturate' as you say), *however* fortunately nature and our fallible CCDs come to the rescue!

It so happens that saturation in the individual R G B channels and luminance often correlate; chances are that when a lot of signal has been picked up in the red channel, a fair bit of signal will also start to seep into the green and blue channels (purity is extremely rare in nature!). This will turn a pure, saturated red into a lighter, less saturated red. And where the wells of our CCDs saturate (e.g. star cores), some or all channels obviously max out resulting respectively in a colour tending to white, or in a colour that is completely neutral/white - that is if colour calibration was done properly and saturating channels are handled correctly (I'm looking at you DSS...).

Hope this helps (and doesn't confuse things more )
Reply With Quote
Old 18-11-2014, 07:02 PM
Shiraz's Avatar
Shiraz (Ray)
Registered User

Shiraz is offline
Join Date: Apr 2010
Location: ardrossan south australia
Posts: 4,823
Originally Posted by irwjager View Post
Hope this helps (and doesn't confuse things more )
Thanks very much Ivo, your explanation certainly did make the complexity of the issue a lot clearer. Looks like I will now need to do a lot more reading though
Reply With Quote


Thread Tools
Rate This Thread
Rate This Thread:

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump

All times are GMT +10. The time is now 03:26 AM.

Powered by vBulletin Version 3.8.7 | Copyright ©2000 - 2021, Jelsoft Enterprises Ltd.
Limpet Controller
NexDome Observatories
Astronomy and Electronics Centre