Human eye gamut .icc profile

[davidedric]

Hi,

I have the ColorThink program which allows the display of various colour spaces, printer, monitor, photo gamuts. I use it mostly for interest. It would be nice to have the human eye gamut expressed as a .icc profile for comparison purposes. Does anyone know the whereabouts of such a file?

Dave

[pnodrog]

http://en.wikipedia.org/wiki/CIE_1931_color_space

Enjoy.... Not sure if it will lead you to a .ICC profile.

P.S. I just read that .icc profiles should be available from the manufacturer so I suppose you ask your parents (I am sure that's a great help)

[Inkanyezi]

It's an interesting question indeed, but there won't be a straight answer. It is philosophically related to the EttR issue, that is so difficult to understand and yet so simple. It is a bit more fundamental though, as it forces an idea about "true colour"; something we can never achieve, considering colour in itself is not a physical property, but a figment of our imagination. Colour is our personal three-stimulus perception of light. It's an evasive property.

The "human eye" isn't standardised for colour perception, and there is a great variance. most people can see a wider gamut than any of the ICC gamuts, and the most extreme gamuts may be found in young children and in people operated for cataract, where a clear lens has replaced their opaque one. Whether you'd regard the particular vibrance as colour or not is more a philosophical question. Colour perception is a brain function, and the eye is an extension of the brain.

So I second pnodrog's suggestion, fully aware of the irony. In the future we might have sufficient knowledge to quantify colour perception, but it would not help us to visualise someone else's perception.

[davidedric]

Thanks gentlemen.

I wasn't looking for anything too rigorous, but something that would work as a ball park, and something that I could view in three dimensions rather than the usual too. In fact, I have discovered that it is there in the set of profiles bundled with ColorThink, I just didn't recognise its name.

Has my colour perception widened since my cataract operations of a few years ago? How could I now know without being able to do a before and after?

I suspect this is another of those topics where the more you learn the more you realise you don't know.

Dave

[Inkanyezi]

Your visual perception probably has changed, although you might not recognise it as colour. We are still tri-stimulus mammals, although there is also a fourth blue-sensitive receptor in the eye. That receptor has direct connection to the pituitary gland and is not associated with vision. It has a role in melatonin release.

So the colour gamut still depends on the tristimuli, although the stimuli after operation also contains UV, which can be percieved as vibrance or as a particularly bright white, in those white objects that reflect much UV. Also other colours reflecting much UV are affected. Those that have operated only one eye may compare the sensations of the operated versus the immaculate one. Also children, before the lens of the eye is yellowing, can percieve UV. But getting used to it, one will not notice what the difference might be.

There is one way to find out though, and that is placing an UV filter in front of the eye. If you perceive a difference with and without it, that is the difference when UV is present. It is best seen when there is a substantial amount of UV in the environment, particularly if other frequencies are subdued.

[xpatUSA]

Hi,

I have the ColorThink program which allows the display of various colour spaces, printer, monitor, photo gamuts. I use it mostly for interest.

It would be nice to have the human eye gamut expressed as a .icc profile for comparison purposes. Does anyone know the whereabouts of such a file?

Dave

I have it too (the 'cheap' version) but I also have a feeling that the program uses a tristimulus model somewhere it it's calcs - whereas the CEI 1931 diagram is clearly not triangular. So, IMHO, the best you could do is find the nearest color space, perhaps Adobe 'Wide' (not RGB 1998), and look at it.

BTW, I didn't know that it could display .icc profiles as gamuts.

Also, there's been some correspondence (now old) over on Luminous Landscape about the proper image size to give the most correct gamut view.

And, also, a MacBeth card shot in my version (in the Lab 3D view) doesn't show the color patches as the blobs that one might expect. Instead, it shows lines radiating out from center and upward, which implies that said patches have variable saturation and lightness. I've been meaning to look at that for a while.

[edit] Just looked at Koren's artificial MacBeth image where the patch RGB values are perfect and they still show as lines, not blobs. Makes no sense, unless I'm missing something.[/edit]

Pardon the negative comments, the program is indeed most useful for identifying gamut clipping, for example in flower shots.

[revi]

Indeed, the response of the eye is not a standard tristimulus response, as the receptors don't correspond to 3 separate colour bands (e.g. the red-sensitive receptors are also sensitive in the blue, which is allows the eye to recognise spectral violet as different from blue).

And, what do you mean exactly by an ICC profile? Normally that term indicates a device-specific set of values to correct deviations from ideal colour reproduction. As the average eye is used to define what 'ideal colour reproduction' means, such a profile for the average human eye would be the identity profile (i.e. what comes out is exactly what goes in).

If you mean to get a colour space corresponding to the human eye, isn't that the Lab space?


@Teb: colour patched showing up as lines, and not as blobs:
you would expect points in the ideal case (like the articifial MacBeth image) : 1 colour <==> 1 point in the diagram.
What is probably displayed, is the actual value, the expected value and the line connecting those two. One would expect
a small symbol at each end of such a line to identify which is which in that case, though.

Note that the colour (or spectral composition) of the illuminating light also plays a role in where the patch colour ends up in the diagram, so even if you have a perfect card and perfect profiles, the actual values can be different from the expected values.

[ajohnw]

I believe you can get what you are after by viewing the LAB colour space as this reckoned to be the gamut of the human eye. It's shown all over the place including here

http://www.peachpit.com/articles/art...30296&seqNum=2

There are several thing to note. In order to cover it with a mix of 3 colours some or all of them depending on which gamut you look at need to be outside of the range of our vision - so called imaginary colours. That sounds great but taking the blue end for instance it would be in true UV and usable levels would damage our eyes. We can actually see near UV but fortunately there isn't much of it about. The levels are low. Things get complicated further by our eyes response to light levels eg when dark colour vision tends to disappear. One interesting aspect of that is viewing star colours with even a small telescope - they concentrate the light so the star colours become visible.

There is a more meaningful gamut than LAB etc - Pointers. He sort of disregards the colours that our eyes are not really that sensitive to so make little difference and also makes assumptions about reflectance but even then this colour space can not be adequately covered with 3 colour mixes. There is a lot about that here

http://www.tftcentral.co.uk/articles...ters_gamut.htm

This page compares this gamut with a number of others. It's interesting to look at the wavelengths that are used in them and see the actual eyes sensitivity to them here which has both linear and log scales.

http://www.giangrandi.ch/optics/eye/eye.shtml

People tend to disregard this aspect without realising what in means when say mixing say 470nm with some other colours to produce some shade. As out eyes don't respond all that much to wavelength like this rather a lot of it is needed.

John
-

[chauncey]

Whenever I have questions like this, I consult my go-to guy...http://digitaldog.net/
Aside from that, I do question the value of such knowledge considering the vast differences in our vision.

[davidedric]

This was the kind of thing I was playing with. It's a bit hard to see without being able to rotate it, but the outer wire frame is (I think) the human gamut, and the inner solid shape is sRGB. Sorry, I am a geek

Dave

[xpatUSA]

If you mean to get a colour space corresponding to the human eye, isn't that the Lab space?

I think not, if I understand the question correctly. The human eye response fits within and is not the same as the L*a*b* model , just as it fits within the XYZ, or L*uv or any other model.

@Teb: colour patched showing up as lines, and not as blobs:
you would expect points in the ideal case (like the articifial MacBeth image) : 1 colour <==> 1 point in the diagram.

It seems to me that you have not used ColorThink. In ColorThink you choose between square, cubic or circular 'blobs' to represent points in the 2D or 3D gamut views.

What is probably displayed, is the actual value, the expected value and the line connecting those two. One would expect a small symbol at each end of such a line to identify which is which in that case, though.

No. In the view that I use, delta-E (if that is what was meant) is not shown.

[edit]Well, the plot thickens! I just made a pure monochromatic image, every pixel was L=85, a=-102, b=83, and saved as sRGB. Opened it in ColorThink, selected the "point" representation to be a small cube. In both 3D and 2D Lab gamut views, it showed as . . . . . a single cube (in other words, not a line of them). Re-opened Koren's Macbeth image and also Lindbloom's for good measure - and the lines were still shown.

This is just for info, not a problem to be solved. Just one of those mysteries . . . [/edit]

[ajohnw]

I think not, if I understand the question correctly. The human eye response fits within and is not the same as the L*a*b* model , just as it fits within the XYZ, or L*uv or any other model.
. . . [/edit]

Even the wiki had problems with LAB but I have been lead to believe that the usual gamut plot shown is supposed to be the extent of human vision - not that it's possible to show this pictorially colour wise on any PC etc screen as they can not cover the same scope.

Where the confusion comes in is LAB style adjustments. These are reckoned to model how our eyes respond to colour and light variations. They can have contrast, lightness/brightness and chomaticity/saturation sliders just like the normal adjustments of this type but the effects are noticeably different. It's also possible to shift hue,saturation and brightness of specific colours in some packages that use this model. I have seen one example where a blue flower was turned to white leaving the rest of the image as it was. There is also a more recent model that uses the data in the exif to adjust shots according to the lighting conditions when the shot was taken.

https://en.wikipedia.org/wiki/Lab_color_space

https://en.wikipedia.org/wiki/CIECAM02

John
-

[xpatUSA]

Even the wiki had problems with LAB but I have been lead to believe that the usual gamut plot shown is supposed to be the extent of human vision - not that it's possible to show this pictorially colour wise on any PC etc screen as they can not cover the same scope.

Hi John,

Here's a human vision gamut plotted by Hoffmann in 2D CIE Lab space:



The outer, lighter area is the projection of the CIE standard human vision gamut onto the chrominance axes a* and b*. As can be seen, it is within the a*b* area bounds but also goes outside of them. This diagram alone should be of great interest to OP Dave and should be compared to his one posted earlier in this thread. The axes above are scaled +/-250, although +127/-128 is more common, e.g. in RawTherapee.

The scaling in ColorThink is unknown because their axes are unmarked. One thing I have noticed, though, is that their Lab 2D view clips at it's boundaries, unlike Hoffmann's view above. So - users beware!

The above diagram is from a Paper by Gernot Hoffmann which is the very best that I have re: Lab. I have put it up on my site and I strongly recommend it for further reading:

http://kronometric.org/phot/xfer/CIE%20LAB.pdf
.