I am very ignorant on these subjects. I mean, I know SRGB is better for the web, so does that mean Adobe RGB is if you want to print? If I'm capturing in RAW, can you set the colorspace later?? What should I print, a TIFF or PSD?

You don't have to answer these questions, links to web tutorials or even BETTER, is there a book out there that is regarded as awesome for teaching this stuff?

thanks.

The classic reference is Bruce Fraser's "Real World Color Management" (http://www.amazon.com/World-Color-Management-Bruce-Fraser/dp/0201773406)

Kaa

A quick rundown:

sRGB has a lower number of colours in its gamut (i.e. its range) than Adobe RGB. The thing is, it tends to be the standard for web browsers, hence it being "better" for web. A web browser will display an Adobe RGB image just fine, but they tend to look a little washed out or drab because the browser just ignores any colours that aren't in the sRGB gamut.

You can change an image's colour space as many times as you'd like (although try to keep it to a minimum as you can lose quality). When working on an image in PS that you'd like to have in a different colour space (e.g. to display on the web) be sure to use "convert to profile" which maps the current colour space onto the new one (Adobe RGB -> sRGB, for example). You likely won't see any change, but the difference is apparent when using a browser like IE or Firefox (Safari supports embedded profiles, so it displays Adobe RGB properly).

As for printing, that depends. If you don't print at home ask the staff at the place you have prints made, they'll tell you what's best for their setup. This goes for file types as well, but in my experience most places use either TIFF or JPEG.

DPI was in your title, so a quick bit on that too. Changing an image's ppi (pixels per inch, DPI=dots per inch) will not affect how you see it on screen because your monitor has a set resolution. My screen, for example, displays 110ppi, so when I look at an image I'm seeing it at 110ppi - only changing the image's dimensions in pixels will change how it displays. Changing an image's resolution only affects how it will print: a 100x100 pixel image will print at 1 inch by 1 inch if printed at 100ppi, but only 0.5in by 0.5in if printed at 200ppi.

Ok, that's enough typing for one night. Hope it helps.

Give the workflow guidelines link a the top of this page a read. Its a great rundown of the basics.

kylegehmlich wrote:
sRGB has a lower number of colours in its gamut (i.e. its range) than Adobe RGB.

That is incorrect. An 8-bit sRGB image will have exactly the same number of colors as an 8-bit AdobeRGB image. The colors will be different, in particular AdobeRGB will be able to represent colors that sRGB cannot. But the number of colors will be the same.

kylegehmlich wrote:
The thing is, it tends to be the standard for web browsers, hence it being "better" for web. A web browser will display an Adobe RGB image just fine, but they tend to look a little washed out or drab because the browser just ignores any colours that aren't in the sRGB gamut.

That's also incorrect. A web browser doesn't "ignore" colors outside of the sRGB gamut. What's happening is that the the pixel's RGB numbers represent different colors in sRGB and in AdobeRGB. If you try to display an AdobeRGB image in a non-color-managed system which assumes sRGB, all pixels will display (numbers are numbers), but the colors will be off.

There are also color-managed browsers which will display an AdobeRGB image correctly.

Kaa

So if all browsers just supported color management we wouldn't even need to think about it? How does it relate to printing? Is one better than the other if you are printing (printing on my school's Epson 9800)?

Future Man wrote:
So if all browsers just supported color management we wouldn't even need to think about it?

No, you still would because -- as with most everything in real life -- it's a matter of trade-offs and different color spaces make different trade-offs.

The most important trade-off is that narrow color spaces (e.g. sRGB) have a limited range of colors they can represent, but within their range they can represent colors more finely. Wide color spaces (e.g ProPhoto) can represent a lot more colors, but the trade-off is that the "distances" between the colors they can represent are bigger and so the color granularity is more coarse.

Go read the Fraser's book.

Future Man wrote:
How does it relate to printing? Is one better than the other if you are printing (printing on my school's Epson 9800)?

Printers have their own gamut. Ideally you want to use the color space that matches the printer's gamut -- which never happens, so you have to make -- guess what? -- trade-offs :-) Besides, you might be limited by your printer's software (drivers) with respect to which color spaces it will understand.

Kaa

Thanks for the corrections KaaX, but now I'm a little unclear about Adobe RGB vs sRGB. I'd always thought that Adobe RGB had a wider gamut...

The way I learned it, Adobe RGB has all the colours of sRGB in its gamut, but sRGB doesn't have all the colours of Adobe RGB. You're saying that's not the case, they just have different colours?

kylegehmlich wrote:
Thanks for the corrections KaaX, but now I'm a little unclear about Adobe RGB vs sRGB. I'd always thought that Adobe RGB had a wider gamut...

The way I learned it, Adobe RGB has all the colours of sRGB in its gamut, but sRGB doesn't have all the colours of Adobe RGB. You're saying that's not the case, they just have different colours?

No, you're basically right, but you've been tripped by terminology :-)

Colors are continuous. There is an infinite number of colors between red and yellow, for example. However our current digital systems can only represent colors as separate, discontinuous "dots" in the color space.

For example, the color RGB(130, 140, 0) is some kind of yellow. The color RGB(131,140, 0) is a different yellow. But you can't represent a color in-between them -- RGB(130.5, 140, 0). Such a color exists in real life, it's just that you can't describe it using, say, 8-bit RGB triplets.

Thus, there are two different concepts.

One is gamut which is the part of the color universe that a particular color space can describe. Gamut is like an area or volume -- it's not countable, but you can speak of smaller gamut or larger gamut. In particular, AdobeRGB gamut is larger than sRGB gamut. AdobeRGB can indeed represent colors that sRGB cannot.

The other concept is number of colors which refers to how many different colors can you represent within your system. This is completely separate from gamut. Because color space is continuous, you can represent many, many colors in a small section of the color universe -- or you can represent a few colors in a large section.

The number of colors is determined, basically, by how many bits you have. An 8-bit RGB triplet can represent 256 x 256 x 256 = 16,777,216 different colors, from pure black (0, 0, 0) to pure white (255, 255, 255). sRGB picks these 16m colors from a smaller region of the color universe, so its colors are "denser", but do not "reach" as far. AdobeRGB picks these 16m colors from a larger region, so its colors are more "sparse", but "reach" further. However the *number of representable colors* in sRGB and in AdobeRGB is the same.

Kaa

Go to your book store and pick up a book on digital photography.

To wide a topic to be addressed in posts on an Internet forum.

That makes perfect sense, thanks Kaa!

So, if I understand correctly, sRGB would actually be better at defining a gradient than Adobe RGB, as long as the gradient is within its gamut?

kylegehmlich wrote:
So, if I understand correctly, sRGB would actually be better at defining a gradient than Adobe RGB, as long as the gradient is within its gamut?

Correct, at least theoretically.

In practice, the limits of output devices and the limits of human ability to distinguish hues make this advantage pretty much irrelevant unless you're *radically* transforming an image.

Kaa

Bookmarked

The general rule is to keep the colours and profiles at maximum (16-bit, ProPhoto) while you are editing and working on the files. When you output them to JPEG for either web, or sending to a photo lab, etc... then you would save them as the maximum allowable. (8-bit, sRGB for web, and print depends on your photo lab)

joezasada wrote:
The general rule is to keep the colours and profiles at maximum (16-bit, ProPhoto) while you are editing and working on the files. When you output them to JPEG for either web, or sending to a photo lab, etc... then you would save them as the maximum allowable. (8-bit, sRGB for web, and print depends on your photo lab)

Actually, there is some debate as to whether this is always true. Since many modern image applications are web based, there are those who argue that it makes more sense to work in a colourspace that suits the output from start to finish. So if you know your output is web based then you probably should work in sRGB from start to finish rather than having to do a colour space conversion as a last step in the workflow. It has also been argued that you set your colour space to the smallest space that covers the device in your workflow with the smallest gamut (for most people that would be your monitor). It's argued that if you can't actually see the colours in a wider gamut than your display then what's the point of having them at all and that working in a wider space than your hardware can handle actualy may introduce more problems than it solves.

A well known example is the insistance of Lightroom that you work in the ProPhoto colour space. The ProPhoto (melissa variant) is so large that it is nearly impossible to "see" exactly what Lightroom is going to output prior to print or export. There are many who not only want softproofing but also want the choice of having a completely aRGB or sRGB workflow to minimise screen colours and output colours variations.

Gordon

We can't get printer and monitor to match exactly because its physically impossible. What color management does is rearrange the color so a 255.0.0 RGB value will be reproduced in the most saturated red the calibrated monitor can display and printed with the most saturated red the combination of yellow and magenta on the printer can produce.

Color management relies on the fact human color perception is highly adaptive. What we perceive is in large part what our brains expect to see based on previous experience. What's why a white shirt will look the same indoors and out; the brain rationalizes by comparison with other things the shirt must be white, so it adapts our color perception to make it white.

The same thing happens with primary "reference" colors . If we see a red Ferrari in a photo on the monitor or a print, and there is a not more saturated red in our line of sight, our brains convince us we are looking at the same saturation of red as the paint of the car. But if we compare car, monitor and print side-by-side the brain will key off the more saturated red in the car and the other two will look weak by comparison. Color perception is relative to the lightest / darkest tones and most saturated colors in our field of view.

When we profile a device a map of the maximum saturation the device can produce is created. Those maps are used by the computer to know when the max. red in one gamut is different than another, and how to move it. Its done automatically, but we need to select an editing gamut both the camera, monitor and printer gamuts can all fit inside of. The number of discrete colors in the various gamuts is the same and is a function of bit depth. See: http://www.chin.gc.ca/English/Digital_Content/Capture_Collections/Sub_sections/fnd_resolution.html

Color gamuts are three dimensional and their outer boundaries represent the most saturated colors. If you have a Mac you can use the ColorSync utility to compare gamut size and shape as rotating 3D wireframes. Here are some screen shots.

These two compare a 8/C HP printer gamut on high gloss paper the gamuts of AbobeRGB and sRGB:

Chuick,
Thanks for providing such detailed, clear, and illustrated explanations to a complex topic. Great summary!

Scott

flash wrote:
It has also been argued that you set your colour space to the smallest space that covers the device in your workflow with the smallest gamut (for most people that would be your monitor). It's argued that if you can't actually see the colours in a wider gamut than your display then what's the point of having them at all...

That argument is silly. I want the color space that covers the gamut of my final output device, but intermediaries can have narrower gamut.

Many contemporary inkjet printers, for example, have gamuts noticeably beyond monitor gamuts. Why should I limit my output? I might not see the intense blue-cyan sky, for example, on my monitor, but I will see it on the print.

Kaa

KaaX wrote:
That argument is silly. I want the color space that covers the gamut of my final output device, but intermediaries can have narrower gamut.

Many contemporary inkjet printers, for example, have gamuts noticeably beyond monitor gamuts. Why should I limit my output? I might not see the intense blue-cyan sky, for example, on my monitor, but I will see it on the print.

Kaa



Possibly it is silly. But I've heard convincing arguments that having the closest possible match between display and print is an effective workflow by people considered to be experts in the field. After all one of the most common complaints for enthusiasts is "my prints don't match the screen!!".

In addition is it highly unlikely that the majority would be able to "guess" what additional colours are available in print versus their screen. Those extra colours may, in fact, be too saturated or shift in an unpredictable and unflattering way. If I'm shooting a portrait do I really want my skin tones to be more saturated or a slightly different colour? Probably not. Of course, if landscapes are your thing then you may not care. But if you are shooting colour critical product work then you absolutely must have an accurate representation on screen as to what will appear in print. Working to the smalles colour space could be considered essential here. just because your printer can represent more colours than your monitor is not in itself a convincing argument that you should use all of them.

I would assume that most professionals and those serious about print quality would be looking to be moving to wider gamut monitors anyway, so the descrepency between print and screen is much smaller anyway. The whole idea of having a wide gamut monitor is so we can better represent on screen, what will happen in print. If screen/print matches had little importance then we'd never spend the money on wide gamut screens. Actually, why bother with profiles at all if we can just infinitely "visualise" colour?

It was correctly stated above that the eye is highly adaptive. But human nature for an "enthusiast" with an inkjet printer is to hold up a print next to their display, so direct comparisons are made and inevitably, one will be judged right and the other wrong. An inevitably the screen wins. We don't ofen see "my prints look great but my screen is off" topics here, do we?

I spent some time looking into this and I feel that for the majority working to the smallest gamut has merit. For a minority and professionals this is not the case. It's not as if a file can't be reopened and reworked later on anyway.

Gordon

flash wrote:
In addition is it highly unlikely that the majority would be able to "guess" what additional colours are available in print versus their screen. Those extra colours may, in fact, be too saturated or shift in an unpredictable and unflattering way. If I'm shooting a portrait do I really want my skin tones to be more saturated or a slightly different colour? Probably not. Of course, if landscapes are your thing then you may not care. But if you are shooting colour critical product work then you absolutely must have an accurate representation on screen as to what will appear in print. Working to the smalles colour space could be considered essential here. just because your printer can represent more colours than your monitor is not in itself a convincing argument that you should use all of them.

Well, the majority prints their 4x6s at Walmart and doesn't care. We are not talking about them. We are talking about people who are sophisticated enough to worry about the selection of the proper color space -- that's already a small minority of professionals and advanced amateurs.

Your portrait will not go out of sRGB gamut -- sRGB is well-suited to human skin. And if you do the color profile conversion correctly, nothing will signficantly shift tones or saturation, even if most of your workflow is in, say, AdobeRGB. I think that the fact that my printer will represent more colors IS a very convincing argument to use all of them. If you can't deal with that -- by all means stay away. But if you can, there is no reason not to.

And if you are shooting color-critical stuff, you often enough go by the numbers and not by what your screen shows you. If, for example, there is a Macbeth color chart in my shot, I can match the colors precisely and I don't care at all what my monitor shows (or doesn't show) me.

flash wrote:
I would assume that most professionals and those serious about print quality would be looking to be moving to wider gamut monitors anyway, so the descrepency between print and screen is much smaller anyway. The whole idea of having a wide gamut monitor is so we can better represent on screen, what will happen in print. If screen/print matches had little importance then we'd never spend the money on wide gamut screens. Actually, why bother with profiles at all if we can just infinitely "visualise" colour?

Wide-gamut monitors are useful, no doubt. But again, I think you should be guided by the gamut of your final output device. If you are uncomfortable with this, don't. But you'll miss on some colors which -- on certain images -- will make a difference.

flash wrote:
It was correctly stated above that the eye is highly adaptive. But human nature for an "enthusiast" with an inkjet printer is to hold up a print next to their display, so direct comparisons are made and inevitably, one will be judged right and the other wrong. An inevitably the screen wins.

I don't think so at all. Again, people sophisticated enough to be apprehensive about choosing the right color space should understand the inherent impossibility of comparing screen and print. And if print shows wider gamut and better colors, why no take the print as the "right" one?

Kaa