I seems that all cameras take photos in RGB format and printers take photos in CMYK format. If the photo is not in CMYK, then the printer converts it to that format with some loss of quality. Can the loss in quality be noticeably bad ? Is it possible to set my camera, say Canon T3i or Nikon D3100 to take photos in CMYK mode or something similar ? Or is it better to convert the image using some software ?
Thanks.
Inherently, no. The RGB model is natural for recording light, and the CMYK model is natural for printing (where reflected light is subtracted). But see Are RGB numeric values equal to CMYK percentages? — the loss in conversation isn't inherently because RGB to CMYK is inherently lossy, but because the actual color spaces of the devices used are different, and converting in camera wouldn't do anything to help with that.
Some early digital cameras experimented with CMY filters (often CMYG). But these weren't actually producing CMYK output; it still went to RGB. The idea was basically that CMY primaries allow lighter, more transparent filters and therefore more sensitivity in low-light - but in practice, the conversion involves subtracting channels that don't share information, raising the noise and canceling out any advantage. Plus, there were apparently problems getting the filter dyes to align nicely. So, those have pretty much gone by the wayside.
So, the vast majority of sensors are RGB (the array is usually 1 red, 1 blue, and 2 greens or Foveon), so right off the bat you're working with handicap with respect to CMYK. However, that's not really the reason why this doesn't work...
CMYK is a subtractive color model, it works by subtracting from white which is the presence of all colors. RGB is additive, it works by adding the colors to make white. If you consider the sensor a bit, it basically works by gathering light into photoreceptors and that's inherently additive process, it adds some amounts of lights from the red, green, and blue parts of the spectrum to produce the final image.
Doing a little bit more digging (thanks @Gabe for the nudge)... If you drop K (key, which is black), then you could do this. Nikon and Kodak, at least, had CMY(GY) cameras with green or yellow as the fourth color. I haven't found a great explanation for the G, but the premise of two green in the Bayer model is true, then a green or yellow additional makes sense since we're more sensitive to it.
So, I guess the question would be... why did GRBG (or variants) win the race as it were? According to Olympus Micro:
The downside of CMY filters is a more complex color correction matrix required to convert CMY data collected from the sensor into RGB values that are necessary in order to print or display images on a computer monitor.
They have a pretty nice article on this: Introduction to CMOS Image Sensors and the explanation makes sense. If the vast majority of photographs taken are going to be displayed on a monitor, then GRGB sensors make it easier.
Cameras are RGB because of the color filters over each of the pixels (photosites) on the sensor. There are a couple of rare cameras with a filter that instead samples cyan, magenta and yellow (not sure about luminance (black)), however this then would require a conversion to RGB to work with most screens and even a lot of photo printers which are RGB based.
That said, conversion from RGB to CMYK doesn't really have significant problems. There are some parts of the color space that can't overlap particularly well, but it isn't a major detractor.
Also, not all printing is CMYK. In fact, CMYK is primarily used for document and mass publication printing. C-type photo printers (which directly expose photo paper to print a photo) generally use either RGB lasers or RGB LEDs to expose the paper in true RGB.
Similarly, ink jet photo printers use neither CMYK or RGB, but rather a combination of numerous different colors (my printer uses 12 different colors) to achieve the desired final color.
Thus, it doesn't particularly matter what system the camera works in and since monitors and TVs work in the RGB color space and it requires less complexity to process, it makes sense to have cameras work in the RGB color space and then convert color spaces only when necessary.
You have to make a distinction between mixing lights and mixing ink. When you take a picture, whatever camera you have (Digital, Analog, iPad, iPhone...) you are capturing lights.
And according to physics laws, in the lights, the white is the mix of all other colors. The rainbow is a good example of lights diffraction. It's simply a separation of the white light to the different colors. And if you analyse the rainbow, you can notice that all the colors can be obtained using a gradient of Red, Green and Blue with different intensities. So when you capture a white photo and display it in your monitor, your are mixing the Red, The Green and the Blue "sub pixels" of your monitor at their maximum intensities...
However when we talk about printing, it's another story. The printing support is already white (the paper) so logically, whenever the printer finds a white space in the picture it should print nothing on this area. With this example we start differentiating between the RGB mode and CMYK mode.
Now let's continue further. If you take oil painting and you mix Red, Green and Blue. Will you get white ? No ! It's gonna be probably something close to brown. Not even black. So to have "all" the color possibilities you need to mix Cyan (which is close to blue), Magenta (Close to Red), Yellow and Black.
As for the Cyan and the Magenta: these are simply equivalents to the RGB Mode. The black is there to darken the black spaces in a picture but there is no green like the RGB mode. Why? The green it self is a combination of the Blue and the Yellow in the physical world. Try mixing oil painting and you will see. So this is why in printing, you need CMYK colors which are the most basic physical color substances.
Now the conversion can be done using software such as Photoshop.
Hope it clarifies.
This might be taken as rude, but I have to say it - All the answers you got here should be disregarded. B. Sagdiyev probably does not follow answers to this page anymore, but many will Google it like I did. Some are just wrong information, while some stray of the question.
sRGB color space/gamut was made in the 90s in need for color space adjusted for CRT monitors. CMYK is printing standard. Most common color space you will find in files on the Internet are sRGB. Some are CMYK or Adobe RGB.
Adobe RGB has ALL the colors of RGB and CMYK and even colors not in either of these color gamuts. Digital cameras use this color space and 12-bit or 14-bit precision when writing RAW files and sRGB or CMYK with 8-bit precision for JPEG output. (Web images use 8-bit and so do our monitors and that is the great tragedy).
Most people confuse sRGB and RGB. RGB is a type of capturing and displaying color. Anything that uses only red, green and blue is RGB, including the first color televisions. sRGB and Adobe RGB are color spaces written as R, G and B values. They are definitions of which color to keep, which to disregard. It makes thing easier, files smaller and technology cheaper by ignoring color that don't get used on screen, print or whatever the standard is for.
Digital SLRs and other pro-orientated cameras capture not just wider color space than CMYK, but even more colors than any monitor with 3-digit price can show. Same goes for any TV in the WORLD. There are some very expensive monitors than show the entire Adobe RGB color space, maybe even almost all of ProPhoto colors.
You just need to be careful to reduce color space to sRGB or CMYK BEFORE exporting from Lightroom or Aperture since 8-bit color precision is not enough for Adobe RGB or ProPhoto.
