A lot of people still believe that when producing output for the web or because their screen is incapable of showing more colors than sRGB they need to work in sRGB. And then they wonder why the images from the raw files of their high end DSLR have crappy colors. A myth that is repeated over and over again is that if your screen can not display the gamut of your working space you are not seeing the correct colors and therefore it makes no sense to use a wide color space.

Your average DSLR today has a much wider gamut than your monitor. So you will never ever see exactly the colors your camera delivers. You will always see only an approximation of the colors that your camera delivers on screen.

Now what happens if you use a small color space like sRGB (or Adobe RGB which is barely larger than sRGB anyway) as your working space? (Let's assume you have a properly calibrated monitor that is capable of showing roughly the gamut of sRGB)

Your image gets translated from the wide gamut of the sensor to the small gamut of sRGB and then translated again to the characteristics of the monitor for display. As the image colors as delivered by the camera encompass a wider gamut than sRGB, colors get compressed into sRGB to deliver a approximation of the sensor's wide gamut as close as possible within your smaller color profile. To achieve that visual approximation, the image gamut will hit the borders of sRGB immediately when coming from the wide gamut of the DSLR. So you are already close to clipping. If you now change anything in the image that affects the colors (like saturation or exposure) you will reach clipping faster than you can blink and your images will suffer. As your monitor profile will be similar to your working space, not much will happen after that when the image gets sent to the display. The clipping has already happened and your screen will just display the clipped colors.

Let's compare this with a ProPhoto (or WideGamut) based workflow:

Your image gets translated from the wide gamut of the sensor to the even wider gamut of ProPhoto RGB and then translated again to the characteristics of the monitor for display. As ProPhoto is wider than the camera gamut, the images colors are easily accommodated. No dangers of clipping even when heavy image manipulation is applied as ProPhote has enough headroom to accommodate that. Now the image still needs to be displayed on your sRGB like screen. So the image gets translated from ProPhoto to your screen using the gamut your screen can provide to deliver a close as possible approximation. But as there are no image adjustments after the monitor conversion, using the full gamut of he screen will not drive your image into clipping, you just get the best approximation your screen can do.

The big difference between the two scenarios is that in the second one you see on screen a much closer approximation of what your camera provides than in the first scenario where an artificial gamut compression happened too early in the pipeline.

Or in other words, do you want rounding and clipping errors on your data during the processing or only some approximation errors in the final conversion to the output device?

What I described here, applies also for printing. Printers do have smaller gamuts than DSLRs but they will also benefit from keeping the image data wide until the last conversion.

A few years ago I made an experiment to see how much the choice of working space influences the final printed image. I went into the garden, took a properly exposed image of a red rose in raw, then used various working spaces and file formats to develop the image before sending it to the printer. Then I took the prints outside and compared them to the rose. The version where a completely wide path (ProPhoto RGB and 16 bit TIFF) was used to feed the printing program delivered the best representation of the real rose.

All of the above assumes working in your raw converter (which all use at least 16 bits per channel internally) or with 16bit TIFF files. Wide gamut profiles will lead to posterization when used in JPGs or other formats that have only 8 bits per channel, as the range of information that can be represented in 8 bits per channel is not big enough to encompass a wide gamut. Modern DSLRs deliver at least 12 bits per channel (14 is more common) in their raw files, so the source data from the camera is already using more than 8 bits per channel and it should be kept in 16 bits per channel as long as possible before converting to the final output.

But what about wide gamut screens?
If you have a screen that displays a wider gamut than sRGB and want to create sRGB output, what happens then? The above assumption about the screen being close to sRGB does not hold. So you either need to reduce the gamut of the screen in the monitor settings (but why have a wide gamut screen then?) or you need to soft proof. Soft proofing adds a conversion layer between the working space and the montor profile to simulate the results of switching to a specific working space. Now you can easily emulate the results of a smaller working space on a screen with a wider gamut.

Unfortuantely, soft proofing is not available in all imaging applications. But if your imaging application supports it, you should use it.

As a side note, while it is useful for simulating smaller working spaces, soft proofing is basically a must when trying to judge potential print results as the gamut of a printer is usually very much dfferent from a screen gamut.

Just to give you an idea about the gamut differences between a current DSLr and the working spaces mentioned above, here are two gamut comparions images created with ICC Examin: The first shows how the camera gamut (from a Nikon D700, in color) is encompassed by ProPhotoRGB (in grey): And now compare this with how small sRGB (in grey) is compoared to what the D700 delivers:

Further reading

Luminous Landscape: Understanding ProPhoto RGB

Digital Outback Photo: Color Management for Photographers #006
Why Use the ProPhoto RGB Color Space?