# How many shades of gray are there?

There are 256, counting black and white.

2 Likes

Only if you limit yourself to 8 bits. Give each pixel 32 bits and suddenly you have room for 4,294,967,295 shades!

Good luck telling gray #3,216,444,330 from gray #3,616,321,001 however.

1 Like

When oh when did Crayola add âManatee?â That is the best!

1 Like

Grays are actually more useful than colors in understanding a scene. Scramble the colors of a typical image while holding luminosities constant and the image is still very readable. Scramble the values while holding hues constant and you wonât have much chance of understanding the image.

Itâs actually possible to give an objective answer, measuring the smallest difference the human eye can detect, populating points throughout a 3-space mapped to RGB valuesâŚ damn, it turns out there were fifty!

Itâs possible to give an answer for one person, but everybodyâs eyes are different. Sometimes by quite a lot. Color is one of those things that gets really fuzzy really fastâyour eye and brain pull a lot of tricks with it.

Iâm surprised she actually wrote the Indigo chapter in this book. Everybody knows Indigo was just a color someone made up to make ROY G BIV work.

XKCD did a very informal survey awhile back about color that is very interesting: http://blog.xkcd.com/2010/05/03/color-survey-results/

1 Like

#380282

Actually, with 32 bits of gray youâd typically have a floating-point number, resulting in an infinite number of possible shades in the colorspace, but with decreasing accuracy as the values got very large or very small.

This is useful for representing the entire range of light values in a scene, which then get mapped down to a black -> white tonal range when you want to show it to somebody (the same idea as setting the exposure on a camera, but in post).

1 Like

Itâll cost you; but you can buy displays that allege to actually usefully handle 10-bit and very occasionally 12-bit greyscale to a degree of accuracy that makes that more than mere big-numbers-marketing.

Mostly radiology toys, because they are pricey, and few other people are willing to pay nearly that much for very high resolution displays that donât support color but do support absurd numbers of greys.

1 Like

The easter egg is that those crayons have a crimson core under a thin layer of Manatee-color, so you feel like a terrible person for rubbing their skin off onto your paper.

3 Likes

1 Like

Props for the Delany reference, Rob! For an analogue dinosaur like me, raised in the ancient truths of the sacred H&D curve as revealed by its prophets Adams and Archer, this is an easy one: âGrayâ is a continuous tone that spans 8 doublings of exposure.

There are several possible answers to the number of shades of grey. Boundegarâs reply is not far out when judging shades of grey against a white surround. I would have put the number slightly higher, put probably less than 100. The white surround is necessary because our retina is not fully black so we have scatter and flare within the eye, which limits our ability to see small contrasts in deep shadows.

If you take out the white point, then the eye can see contrasts over something like 9 orders of magnitude range of contrast. If you looked at film X-rays, you would probably only see 100 shades of grey at once, but once you had masked off the highlights, you could see details in the shadows. This means if you have good eyes you can see something like 400 shades of grey, but only 100 of these would be visible at any one time. There isnât a huge variation in healthy human eyes, but it is hard to stick an exact number of the just perceptible difference.

If you are printing with halftones, you can see smaller luminance changes at the white and black extremes, because a 0.1% black dot is visible as a dot, and so is not the same as plain paper. However, if the paper albedo changed by 0.1% you would not notice it.

Newton probably used blue (actually âblewâ in his manuscripts) for an azure sky-blue, which is probably a blue-green in modern terms, in which case his âindigoâ is closer to what we would today call âblueâ. This is hard to prove, but Thomas Youngâs early papers gave measured wavelengths, which suggest that in 1800, âblueâ meant a more azure colour, but has shifted by about 1813 to a more modern blue value.

There are lots of risks when assigning words to colours, apart from the Sapir-Whorf hypothesis. I associate âvioletâ with the very short wavelength, or the colour of the African violet, but there are many people who take âvioletâ to mean the colour of the European violet, which is much more pink.

3 Likes

Itâs kind of important to state what you mean by âhow many shades of grayââŚ

Number of shades of gray that a healthy human eye can see at one time? Iâd say itâs around 150-200.

If you consider photometry as it is used in the photo industry, measured transmissive density can vary from 0.00 (no density whatsoever, nothing blocking the sensor), to roughly 4.00 to 4.50 density. Beyond that number, itâs essentially opaque. Looking through filter of a 4.5 density, a 6.00 density, and a 15.00 density filter side-by-side, you wouldnât be able to detect the edges where the density changed.
For a filter with, say, a 3.90 density and a 4.30 filter, youâd be able to just barely see the the line where the two filter met.

So, armed with this, we can generously set the upper density (at which no further increase in density can be seen) at 4.50 density. The scale is logarithmic, with a doubling every time the value changes by 0.30.

The smallest detectable change in density that Iâve seen trained people detect is about 0.03 (letâs call it 0.025). So 4.50 divided by 0.025 gives you 180. Even if you set the upper density limit detectable by our eyes at 6.00, you still only get to about 240. (Yeah, 8 bits per colour is plentyâŚ)

Even given very generous parameters, the human eye can only detect, at the very maximum possible, 180 (more like 150) shades of gray of transmitted light (kind of equivalent to looking at a light source, maybe even your monitor if the room is dark).

For reflected light (yâknow, when youâre looking at a print, or paint) the maximum range of density is from 0.05 to about 1.50, which yields a possible 60 or so shades of gray.

I work in a field where knowing and understanding this kind of stuff is important, nay, critical.

1 Like

The only list that matters says 101 shades of gray.

Gray isnât just a range of spots on a 1-dimensional spectrum between black and white. Human color vision is a really strange and twisty process, and pigments are more complex than light is. For instance, if you want to match a given sample of gray paint, you may need not only grays, but blues, reds, yellows, purples, magentas, possibly some greens and browns. And itâs going to be different depending on whether youâre looking in daylight, incandescent, fluorescents of various color temperatures, LEDs of various prices, etc., and that doesnât even count the effects of shininess; flat vs. eggshell vs. semi-gloss makes a huge difference, as does brand of paint.

One of my rooms is now something painted something approximately Portland Gray, anotherâs off-white in a direction that weâd originally wanted to be slightly grayish purplish (we ended up with blue and red pigments but no purples.)

The number of just noticeable differences along the grayscale between black and white is about 450 according to several studies.

http://www.visualexpert.com/FAQ/Part2/cfaqPart2.html#p2.5

Bonus info: there are between 1 and 7 million (estimated, obviously, since you couldnât do that many comparisons) distinguishable colors.

3 Likes