Digital Outback Photo
- Photography using Digital SLRs


Color Management for Photographers #006

Why Use the ProPhoto RGB Color Space?

note by Uwe Steinmueller (1/9/2005)



Since over 5 years we were using Adobe RGB as our color space. Today we switched over to ProPhoto RGB. Here is why?

Note: We followed the discussion about ProPhoto RGB for quite some time. Yesterday we had a conversation with a friend who understands color management and this made it pretty clear to us. We would never claim to be experts and do not even intend to get any expert status. But the arguments we heard were just too convincing.

All what we write here matters if your main goal is the fine art output (most likely prints).

What is Color Management (CM) all about?

Here is a very short definition we got:

"The purpose of color management is to translate the device-specific numeric values (colorant percentages) so as to keep the resulting actual colors the same as you move an image from one device to another. Each device will then display the correct color as best it is able to."

Very important is the phrase: " best it is able to".

Often the main purpose of color management is explained (we must confess we did that partly too) as:

CM helps to get exactly the same image printed that you see on your monitor.

Unfortunately this is not what you should expect from a proper color management and it is not even possible today.

In the following diagrams (all made with the profile editor of GretagMacbeth ProfileMaker 5.0) the colors mean:

  • light pink: Artisan Monitor profile
  • green: Epson Premium Semimatte paper with Bill Atkinson's profile
  • red: Adobe RGB (1998)
  • blue: ProPhoto RGB
  • yellow: Profile for Canon 1Ds II + Capture One made with the Camera Module of Profile Maker 5 using the new Color Checker SG target. Note: We do not claim that this is a perfect profile but we like it in our work.

We provide the diagrams to show how much one color space includes a different one.

Artisan vs. Paper

You can see that the monitor can display colors that cannot be printed and also the paper can render colors the monitor does not show. This mean if you want the print to look exactly like the image on the monitor you would not use the full capabilities of your printer.

You probably agree that it is not a good idea to limit the printed colors just because your monitor cannot display them.

Here is a better goal:

Print as much as possible of the gamut that was captured by the original image (especially digital cameras)

Paper vs. 1DS II

We can see that the camera can create a way larger gamut than the Epson printer but even here the printer has areas where the camera falls short.

You probably ask: what about the monitor? A good monitor will be used as a soft proofing device and will minimize trial and error printing.

The final proof is always only in the print.

Why is Adobe RRB a problem?

aRGB vs. 1Ds II

As you can see Adobe RGB does not preserve all the colors you got from the camera. Once the file is converted to Adobe RGB these extra shades of color are gone. Forever!

aRGB vs. Paper

Even the paper can render some colors that aRGB cannot show. Also think of future printers that may have broader gamut Adobe RGB would limit your data or other output to future projectors.

You can experience that Adobe RGB is a limiting factor in saturated colors like yellows where Adobe RGB would just level all kind of different shades of saturated yellow to a few limited yellow colors.

Why is ProPhoto RGB a better solution?

ProPhoto RGB vs Adobe RGB

ProPhoto RGB is a much larger gamut than Adobe RGB. Get us right we don't want to argue the "bigger is better" route here.

ProPhoto RGB vs. 1Ds II

ProPhoto RGB can hold nearly all the camera colors and more. Where the camera profile shows data outside of ProPhoto RGB we have probably some more or less profiling artifacts.

ProPhoto RGB vs. Paper

Here we have a situation that we can utilize all the paper colors from ProPhoto RGB. Colors that are outside the paper gamut need to be mapped using a "perceptual" rendering intent. If the colors of your photo mostly are in the the range of the paper you can also use the "relative colorimetric" rendering intent.

Paper vs. 1Ds II

Nearly all colors of the paper are inside the camera space. What does this mean? We don't want to give up any colors the printer can render by clipping data to a smaller color space like Adobe RGB.

How to get images into ProPhoto RGB?

Most raw converters will allow you to create files in 16bit that can be converted to ProPhoto RGB. Actually the native color space for Adobe Camera Raw is ProPhoto RGB.

Any Downsides of using ProPhoto RGB?

The main issue with ProPhoto RGB is that you need to use 16 bit images all the time or otherwise risk posterization.

Clearly 16 bit is the way to go anyhow although the performance and more storage requirements hurt. Since Photoshop CS and most tools work in 16 bit today there are only few arguments against using 16bit all the time.

Note: It does not make sense to convert JPGs to ProPhoto RGB. Why?

1. They were already clipped to 8 bit

2. Also there color space will be mostly sRGB or AdobeRGB. So the clipping of data is already done and cannot be reversed.

Why not use the native Camera Space?

This would be a valid alternative. This would even involve only one profile conversion from the camera space to the printer. The downside is that every image would have a different color space. We believe that a color space conversion to ProPhoto RGB will hardly change data and also prefer to have all the images in the same color space.

Here is also a second opinion on using camera spaces for your standard master images:

"Leaving your master image file in the camera RGB or Scanner RGB color space is usually a bad idea because those spaces are not gray balanced and not equal gamma for R,G,B. As you edit the master image file and raise or lower all RGB values the same percentage, you may unintentionally introduce a color cast or color crossover because the camera or scanner represents neutral colors as non-equal amounts of R,G,B. Much better to first use the camera profile or scanner profile to convert into a well-behaved editing space such as Lab or ProPhoto RGB."

What about the old Pictures in Adobe RGB?

If you like the printed results from these pictures in Adobe RGB just leave them. PS allows you to keep any image in its own color space. If you want to transfer older images to ProPhoto RGB then start with the original (e.g. raw files) and stay in 16 bit all the time.

Final Note on Monitors

Still the monitor is your main soft proofing device and needs to be as good as possible to make your life easier. LCD monitors are to be expected to improve in quality still over the next years.

All color spaces in 2D overview

Here you can view a 2D view of all the color spaces used in this article.

Some related links

A Primer on Rendering Intents
By Jack Flesher

After I read Uwe’s article on using the ProPhoto color space, I realized there may be a few readers who do not understand how rendering intents differ from each other. What follows is not intended to be a full discourse, but rather a brief overview on rendering intents. These explanations should help make it clearer when the use of each is appropriate.

In digital imaging when a color falls outside the gamut of our print space, we say that color has been “clipped.” If we did nothing about a clipped color, it would simply get left out of the final image. If we left these lost colors unattended, our final image would most likely come out looking flat due to the missing hues. Rendering intents give us methods of representing the colors that would otherwise be clipped when sent to the print space.

The first thing we need to understand is that the human brain is better at evaluating color relationships than it is at evaluating absolute colors by themselves. What this means is that as long as the relationship between hues remains constant, the final image will look close enough to the original that we won’t notice what has been lost. Rendering intents tell the system how we want those relationships managed.

The four main rendering intents are Perceptual (P), Absolute Colorimetric (AC), Relative Colorimetric (RC), and Saturation (S).

Perceptual compresses the colors in the wider source space to match the smaller gamut of the target (print or viewing) space. This is accomplished by altering the out-of-gamut colors and in-gamut colors so they appear similar when brought into the target space. The entire spectrum of original colors gets uniformly compressed in a way that maintains the overall relationship between the colors. By keeping the relationship between the colors the same, the overall appearance of the image remains the same. The downside is that we have little control as to how this compression is performed, and many of our in-gamut colors can get significantly altered form their original true color.

Absolute colorimetric reproduces the in-gamut colors exactly, but then clips out-of-gamut colors to the last reproducible hue. This usually alters color saturation and lightness; any color outside the gamut simply gets rendered as the last boundary color in the target space. This can leave lots of colors out of the target as well as big gaps in the relationships between some hues. The result is usually a flat looking areas in some parts of the target image with coarse tonal gradations in other areas. This intent can be useful for previewing how paper color affects printed colors.

Relative colorimetric is similar to absolute colorimetric. RC also scales the white point of the source image to match the white point of the target space, giving the source space and target space a common color point for us to “relate” to. This has the effect of making all of the colors appear the same hue relative to white even though they are not actually the exact same colors. Then clips out-of-gamut colors to the nearest reproducible hue. This usually alters color saturation and lightness; any color outside the gamut simply gets rendered as the last boundary color in the target space

Saturation rendering converts saturated simple colors in the source space to the same saturated basic colors in the print space. However it ignores differences in hue and lightness, so it rarely is appropriate to use photographic images and is really designed for rendering graphics that contain mainly simple, basic colors.

Generally speaking, in digital imaging for print output we will usually want to use either Perceptual or Relative Colorimetric intents when rendering our images. RC is best when only a small portion of colors from the source space will get clipped as the slight compression to the out-of-gamut colors is acceptable and the main body of colors is kept relatively intact. Perceptual is going to be the better choice when a large portion of colors from the source space are going to get clipped, as more compression is required to “fit” all of the colors in the target space and still maintain smooth tonal gradations.


For Comments post in our News Group

2000-2007 Digital Outback Photo