Howcome we arent seeing more CMYK sensors?

[Csae]

Especially on the higher end cameras, i would of expected this to be alot more common since CMYK is more standard in industrial printers ?

Screw the 4k iso100 f10 nighttime shots, wouldn't CMYK sensors enable us to capture more colors and gradiants ?

[Margus]

Can you explain your question, I'm confused :confused: I just found only one definition for the term 'CMYK sensor'.

The CMYK sensor is able to recognize whether an object was printed with three or four colors. CMYK stands for the colors of cyan, magenta, yellow and black, which are commonly used in color printing. Three-color printing manages without black.

Are there any cameras with special sensors what record the pictures in CMYK colors instead of RGB? If there are then I would be glad for a link for more info. Never heard before.

[devin]

I thought CMYK was just for print, and RGB was for light. This is the case since you can't mix any combo of R G or B inks to get black. You can, however use CMYK inks on white paper to get almost any color.

On another note, in optics, black is the absence of light. A typical color CCD measures the amount of light hitting it and what wavelength it is. What type of sensor would measure CM and Y, and also Black?

[Beowulff]

Especially on the higher end cameras, I would have expected this to be a lot more common.....

I'm confused too!

Are you talking about printers, which use CMYK, rather than digital camera sensors which utilise RGB? Also, as you say 'a lot more common' this implies that some digital cameras do in fact utilise a CMYK sensor. If so, could you please supply a link?

Thanks :)

[Csae]

Its a bit tricky to explain, but there is no reason other then price and complexity that a CMY sensor could not be made.

RGB is easier and has been around more often, but most printers (especially higher up) are CMYK due to the range of tints allowed, i dont know about you guys but i take photos to print them.

I've been trying to found out more stuff about RGB/CMYK but so far i've only gathered two things :

1) RGB Can get more deep colors at its extremes, (reds,greens,blues) but is limited to three values, 255,255,255

2) CMYK Has alot more tints and gradiants but has problems creating deep reds,greens or blues.

Here is a basic example of what i mean :

http://download.cell.com/images/journalimages/0006-3495/PIIS0006349505731520.gr2.lrg.jpg Im not sure how accurate this graph is, but its about the gist of it. What the graph doesn't show is that the gradiants are alot more varied in cmyk then in rgb.


Unless i misunderstood something, RGB sensors, just measure the amount of red, green,blue that hit by the different wavelenghts, im sure a cmy sensor would be made much in the same manner.

This might not be ideal for consumer point&shoots because things will look different and not oversaturated, however for any professional this would integrate perfectly into a cmyk workflow. After some googling i had found a couple cameras which use CMY or CMYK rather then RGB, but i can't find the page anymore, i think there were some Kodaks.

I think the main problem is not recording the image in CMY, but rather displaying it, as pretty much all displays (monitors,tvs,etc) were normalized with RGB.

[cdifoto]

i dont know about you guys but i take photos to print them.
I take pictures to sell 'em. :D

[Csae]

sellout.


+10characters for minimum posting lenght.

[Robert Besen]

For the record, the ccd only measures light intensity. The color information is obtained by putting filters, (usually R,G, and B) in front of the ccd pixels.

[Margus]

I'd recommend some additional reading to understand the color reproduction process.
http://en.wikipedia.org/wiki/Color_printing (overview)
http://en.wikipedia.org/wiki/Subtractive_color (cmyk explained)
http://en.wikipedia.org/wiki/Additive_color (rgb explained)
(many thanks, wikipedia :) )

As far as I can conclude, it seams that all of the sensors can be only of the type RGB. Because the sensors are measuring the light emitted from the subjects. It would be possible to include additional processing in the image sensor to convert the captured RGB picture into CMYK coded picture, but it's still post processing as the picture can be captured in RGB only. CMYK is applicable for printed colors only.

Cheers,

[mattdm]

There are some things that could be done along these lines. Kodak announced a sensor design a while ago which has grayscale pixels next to the RGB ones — sort of like the K in CMYK, or like the design of the human eye.

One could also make photosites tuned to purple or different shades of green, to help the sensor be more accurate at picking up the subtle differences in those colors. See http://en.wikipedia.org/wiki/Tetrachromacy for examples of this in nature. Of course, it doesn't make much sense to capture colors outside what humans can distinguish, but the idea is that the current technology isn't great at even matching that, so rather than trying to get the individual RGB photosites to be perfect, instead make a wider range and do it in software.

[Beowulff]

This is my understanding too...


.....As far as I can conclude, it seems that all of the sensors can be only of the type RGB. Because the sensors are measuring the light emitted from the subjects. It would be possible to include additional processing in the image sensor to convert the captured RGB picture into CMYK coded picture, but it's still post processing as the picture can be captured in RGB only. CMYK is applicable for printed colors only.

CMOS and/or CCD digital camera sensors don't actually "collect" or "measure" the "intensity" of any particular "coloured" light (per se) at all. All each photosite does is to record all the available light (of ALL frequencies) falling upon its microlens. This light is electromagnetic radiation, and as such doesn't have any intrinsic "colour". The sensor would then produce a grey-scale image. The "colour" of light is endowed by the Bayer array of RGB filters placed over the photosites, demosaicing, firmware algorithms, and our eyes of course.

My question is, therefore, how would the CMYK sensor work in Csae's hypothetical digital camera?

RGB is an additive colour process (for electronic devices), whereas CMYK is a subtractive process (for printing devices). My understanding is that there's no such animal as a CMYK sensor, but according to Csae, there is.

What's the story? :confused:

[Csae]

If you remove the K for black, the comparison is alot easier.

Annoyingly enough its very hard to find pages relating to this exact topic, since the moment you type in CMYK in google, you end up with god knows how many pages of people complaining about their printing and asking the difference about rgb and cmyk.

I suppose the real question is not whether it exists, but if it was a different sensor or if it was just the same file saved differently. If you just measure light amount through a filter, wouldn't it be possible to switch the filters from RGB to CMY or the oposites of CMY?

I also found this : http://www.dpreview.com/news/0809/08091002leaf_afiII_aptus_10_7_6.asp

Medium format digital, with option to save in 8bit rgb, or 8bit cmyk among other things.

I also have no idea what you mean by cmyk animals.. or how it can be a subtractive system, if its printing... color has to be added and mixed to get anything, i dont think printers suck colors out of the paper to get a specific color.

CMYK and RGB both have less colors then we can distinguish normally. But nowadays most of us dont have near 20/20 vision, and i'd hate to see us even try our hand at distinguishing colors and gradiants correctly, but theoretically, im sure the normal/healthy human eye can process alot more colors then are currently available for display on RGB monitors and or CMYK printers.


That being said, cameras do filter out the colors, so that system would actually be subtractive. A certain amount of light stops at each of the filters RGB and moves on to the rest.. so its both subtractive and additive... ok this last part is starting to confuse me.

[fresco]

The problem is that printers *do* suck color out of the incident light. They produce different hues and shades by subtracting different proportions of each color from the light that they reflect to our eyes. The cyan ink absorbs (subtracts) red wavelengths out of the incident light, and reflects green and blue unchanged. The magenta ink absorbs green from the incident light, and reflects red and blue unchanged. The yellow ink absorbs blue from the incident light, and reflect red and green unchanged. That is why printing is called a subtractive process, and these are called the primary pigments.

In theory, by mixing just these three primary pigments, CMY, you could produce any hue and shade that you want. In practice, however, black is used also, to reproduce deeper colors, because the other pigments are not perfectly pure, and also not perfect in absorbing their complementary color, and they would produce a muddy dark gray when all used together, instead of a true black.

We perceive different hues and shades of color, according to the proportion of the three primary colors, red, green, and blue, reaching our eyes. In order to measure the hue and shade of an object, we need to measure the amount of each of the primary colors that it is reflecting, or radiating. We can produce almost any hue or shade by adding together light sources of these three primary colors, in varying proportions. If we want to produce the color blue, we would add together 100% blue light, 0% red, and 0% green light. If we want to produce the color magenta, we add together 50% red light, 50% blue, and 0% green light.

However this is not possible if our light sources were CMY. There is no combination of cyan, magenta, and yellow *light* that can produce the color blue, because each is the combination of two primary colors. That is why they are called the primary pigments. They only work by subtracting their complementary color from the ambient light.

With that said, any RGB sensor could already be considered as a CMY sensor, if you look at it from a slightly different perspective. We can transform any RGB representation of a scene into a corresponding CMY representation by a simple mathematical formula:

C = 1-R
M = 1-G
Y = 1-B

So in that sense, they are equivalent, but each representation is useful for its own purpose. For example, you could consider that the red elements of the sensor are simultaneously telling you how much red light is falling at that location, and also telling you how little cyan pigment should be used to print at that location. For example, 100% red means 0% cyan. The same is true for the green sensor telling you how much green light, and how little magenta pigment, etc. It really is just a matter of perspective, in that sense.

Just my 2 cents, fwiw.

[Beowulff]

I eventually managed to track down a camera that didn't use an RGB array.

The 2001 Nikon Coolpix 995 used a CYMG array utilising CYAN, YELLOW, MAGENTA, and GREEN filters in equal numbers. Interesting to note that way back when, they didn't use twice the numbers of "G" filters as is the case now. I guess also there'd be no such thing as a "K" filter component as we already effectively have that with an RGB array — a 'no light' scenario.

Cheers :)

[TheObiJuan]

Good find Beowulff!
Innovation and risks are the only reason we progress.

[Csae]

So.. i wonder how well that worked?

I cant imagine it competing off the bat with rgb, but im sure it had some potential ?

[Rooz]

its a coolpix. it worked like every other nikon compact...
shithouse.

[Csae]

From a dpreview it seemed to have done okay.

There didn't seem to be alot of difference in how the sensor handled things though.

[AlexMonro]

Fresco, thanks for the clear and informative description of the difference between additive and subtractive colour. I sort of knew about this, and wanted to mention it in this thread, but you've put it into words far more clearly than I could.

[mattdm]

its a coolpix. it worked like every other nikon compact...
shithouse.

Wrong. The 9xx series was their high-end line, with an innovative body design, great features, and the best image quality of any digital camera of the time in its approx-$1000 price range.

That said, it's funny to think that I paid more for my 950 than I have for any dSLR body. If I knew then what I know now, I probably would have gotten a film SLR and a good negative scanner instead.

It's also interesting to note how far technically dpreview has come. They seem entirely baffled by purple fringing.

[Beowulff]

Oh dear... foot-in-mouth disease strikes again!


its a Coolpix. It worked like every other Nikon compact...
shithouse.

According to Pocket-lint, the new Coolpix S620 overall gives good quality images.

They say that "the picture results are very impressive. Colours are reproduced well with a good contrast between colours with little if no bleeding. We especially like how the camera performed in some of our test photos that featured strong contrasting colours..."




— With my apologies for being somewhat off-topic. It's simply that I'm sick of the dSLR snobbery that seems to be insidiously infiltrating most photography forums of late.

I can only assume that the dSLR fanboys are running scared in the face of ever-increasing high image quality from the upper-end point 'n' shooters?

Cheers :)

[mattdm]

— With my apologies for being somewhat off-topic. It's simply that I'm sick of the dSLR snobbery that seems to be insidiously infiltrating most photography forums of late.

Maybe, although it may simply be that ever since the Coolpix 9xx series, Nikon's point & shoot cameras — regardless of how good their dSLRs are — have been average-to-subpar with very few exceptions. (Not that it was a constructive comment at all!)