Jump to content
Science Forums

Recommended Posts

Posted

Color rendering and perception is smearily complicated

What dazzles me about additive color devices (color TVs, computer screens, etc.) is that underlying the apparent simplicity of the standard color values (eg: 24-bit RGB values like 000000 black, FFFFFF white, 6E8E23 olive drab, A0522D brown, etc) is the complexity of the spectra of the primary colors (eg: FF0000 red, 00FF00 green, 0000FF blue) rendered by actual display devices.

 

Here’re the spectra of the 3 primary colors emitted by a typical old-fashioned phosphor screen CRT:

(its attribution page)

 

While the blue and green spectra are fairly fat-banded, they at least have distinct peaks. The red spectrum is so wildly spiky it’s hard to believe we perceive it as the “pure” primary color we do.

 

More complicated still, different display devices don’t render colors the same. Simply looking at a several smartphones side-by-side displaying the same image, or watching a store showroom full of TVs, you realize that even the seemingly uncomplicated rendering of white varies between products, and even between different units of the same product.

 

On top of this complexity human color perception is tremendously complicated. We don’t, as one might wishfully assume, simply have 3 different kinds of photoreceptor cells that respond to something close to the center frequencies of the primary colors. Individual cells may respond identically to different frequencies of light, or differently to the same frequency at different times, so perception involves neuronal averaging still only vaguely understood.

 

Conscious perception clearly has a role in color perception, as even when two color-emitting devices render a know object with very different spectra, we easily believe they are identical. For example, if context suggests a color is white (say, a piece of paper) or red (say, a strawberry), we readily accept that as its color, even if it is fairly badly mis-rendered.

 

From emission to conscious perception, color is, for lack of a better work, smeary.

Posted

 

...snip...

 

]While the blue and green spectra are fairly fat-banded, they at least have distinct peaks. The red spectrum is so wildly spiky it’s hard to believe we perceive it as the “pure” primary color we do.

 

It helps to keep in mind that in the biological environment we seldom encounter monochromatic stimuli. Our visual system evolved in a way that any stimulation across the absorption spectra of the photopigments will produce some afferent affect which varies in proportion to the amount of energy absorbed. Discrimination only occurs in the comparative outputs of at least two different photopigments of the three normally present.

 

Aside from some very minor variations, the vast majority of human visual pigments are very, very close. In most cases of color defect, the defect arises when one of the pigments is missing. The defect I mentioned earlier in the thread, Anomalous Trichromatism, does actually appear to result from slightly different absorption spectra in either the red-catching or green-catching pigments. The competing explanations were either different absorption spectra or different amounts of "normal" pigment packed into Cones. Essentially I did the critical experiment that supported the changed pigment hypothesis and pretty well eliminated the amount of pigment hypothesis. Heady stuff for a young grad student. :D

 

More complicated still, different display devices don’t render colors the same. Simply looking at a several smartphones side-by-side displaying the same image, or watching a store showroom full of TVs, you realize that even the seemingly uncomplicated rendering of white varies between products, and even between different units of the same product.

 

My guess is that the emitters are not emitting exactly the same monochrome light. The Trichromatic theory of Color Vision only requires any 3 different, and fairly widely spaced, monochromes.

 

On top of this complexity human color perception is tremendously complicated. We don’t, as one might wishfully assume, simply have 3 different kinds of photoreceptor cells that respond to something close to the center frequencies of the primary colors. Individual cells may respond identically to different frequencies of light, or differently to the same frequency at different times, so perception involves neuronal averaging still only vaguely understood.

 

Sadly I haven't kept up with the literature since I retired in 2001.

 

Conscious perception clearly has a role in color perception, as even when two color-emitting devices render a know object with very different spectra, we easily believe they are identical. For example, if context suggests a color is white (say, a piece of paper) or red (say, a strawberry), we readily accept that as its color, even if it is fairly badly mis-rendered.

 

There are quite a few perceptual phenomena that are dependent on things like color-contrast, brightness contrasts, etc. That's the psychological side of PsychoPhysics.

 

From emission to conscious perception, color is, for lack of a better work, smeary.

 

 

:thumbs_up

Posted

No doubt the phisiology is complicated, that's why it isn't easy to engineer and fine tune things so that a simple triplet of numbers is sufficient to obtain the corresponding stimulus. The goal is, however, only for the proportion between the three numbers to give the right result for any given patch of a same colour, as exactly as possible. All the matters of context, comparisons and so on aren't something that the designer of screens or films can (nor should have to) allow for.

 

green is between yellow and blue as you go around the colors of the rainbow

 

FF0000 red

FFFF00 yellow

00FF00 green

00FFFF green-blue

0000FF blue

FF00FF purple

 

http://en.wikipedia.org/wiki/Web_colors#HTML_color_names

If it "goes around" (and as far as our visual system goes, it does) then you might as well say any one of them is between any other two. You could even say yellow is between blue and mauve. :shrug:

 

But human and animal photic perception is based on another sort of "engineering", different sensors, different sensor distributions, different signal "multiplexing" (or coding), different kinds of analyses, and a function of both long-past and immediate-past data streams.
Given that the one sort of engineering aims at following the other, at least regarding each little patch of view, given also that RGB is a notation independent of hardware details that just refers to the three primary colours, there's no mistake using it to discuss our perception.

 

As far as we care, violet merges into purple. Saying this doesn'e mean I believe that increasing the frequency incident on the retina toward ulraviolet would begin to stimulate the red receptors, nor that decreasing it toward infra-red would begin to stumulate the blue receptors. It just means that if we stimulate those two kinds, in varying proportions, the perceived sensation varies between red and blue with continuity.

Posted

No doubt the phisiology is complicated, that's why it isn't easy to engineer and fine tune things so that a simple triplet of numbers is sufficient to obtain the corresponding stimulus. The goal is, however, only for the proportion between the three numbers to give the right result for any given patch of a same colour, as exactly as possible. All the matters of context, comparisons and so on aren't something that the designer of screens or films can (nor should have to) allow for.

 

If it "goes around" (and as far as our visual system goes, it does) then you might as well say any one of them is between any other two. You could even say yellow is between blue and mauve. :shrug:

 

Given that the one sort of engineering aims at following the other, at least regarding each little patch of view, given also that RGB is a notation independent of hardware details that just refers to the three primary colours, there's no mistake using it to discuss our perception.

 

Minor quibble. There really aren't THE three primary colors, rather three primary colors in humans based on unique photo-pigment genetics. Helmholtz' Trichromatic theory only requires three suitably selected monochromatic stimuli. The behavioral end-point in terms of color matching to a sample might yield different proportions of the generic R, G, B but for the subjects doing the matching they would state a perfect match achieved.

 

As far as we care, violet merges into purple. Saying this doesn't mean I believe that increasing the frequency incident on the retina toward ultraviolet would begin to stimulate the red receptors, nor that decreasing it toward infra-red would begin to stumulate the blue receptors. It just means that if we stimulate those two kinds, in varying proportions, the perceived sensation varies between red and blue with continuity.

 

Yes, it becomes a binary comparison, i.e., the only stimulation is from the long end and the short end of the portion of the EM Spectrum, and the mid-wavelength receptor is not stimulated. Light outside the range of absorption of the visual pigment passes through the pigment with no effect.

 

I'm finding it interesting how our two "languages" say pretty much the same thing on the stimulus side. B) but sometimes show a little drift on the neurobiological, and psychological subjective side. :)

Posted
Minor quibble. There really aren't THE three primary colors, rather three primary colors in humans based on unique photo-pigment genetics.
Not even a quibble! :)

 

I was, of course, implicitly talking in terms of human (our) perception and I doubt anyone has ever designed photographic film, video screens or other contraptions optimized for a dog or a bird's vision. You never know, though; it might be a great success with elderly ladies. No use for cats... :D

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
×
×
  • Create New...