Monitor Profile - sRGB and blocked shadows

[tomczak]

How to explain it:

The working colour space is AdobeRGB; the monitor is not quite calibrated, but it has about right gamma; the Monitor Profile is set to sRGB; the image file is in AdobeRGB space.

On the monitor, the dark pixels below lower ~4% (pixel values 0 to 9, in 8bit) are all blocked to black. Changing the Monitor Profile to, say, SMPTE-C or PAL-SECAM, recovers most of the blocked blacks.

My suspicion is that it may have to do with piece-wise gamma equation in sRGB. But then, in less than ideal situation when the monitor is not calibrated properly, which colour space is best to assign to it, so that I could best estimate how the image will look like in prints and on other monitors? I don't expect colour translation to be ideal, but at least tones should be where I want them.

[JvdW]

I am not an expert in this field, and it is hard to diagnose these things from a distance, but I have a few suggestions:
- if you have no custom monitor profile and your monitor is not of a wide gamut type you should set Monitor Profile to sRGB (in general monitors are designed and manufactured to display the entire sRGB gamut)
- one thing that is often overlooked or misunderstood is the correct setting of 'brightness' and 'contrast' of the monitor (most people tend to set the monitor too bright with too much contrast), you might check if your settings are right (there are many test images on the net, like here: http://www.lagom.nl/lcd-test/)
- it is important to realise that color translations to colorspaces like sRGB, AdobeRGB and ProPhotoRGB wil allways use relative colorimetric rendering intent regardless of the settings. With sRGB as your monitor profile all the saturated out of gamut colors will be clipped off. One way to ensure good screen to print match when using a sRGB gamut monitor is to first convert colorspace to sRGB with relative colorimetric intent, and then print with perceptual rendering intent (or use default printer settings if you don't use a printer profile). The downside of this approach is that you litteraly throw away all saturated colors outside the sRGB gamut, while some of those colors can be printed on modern printers.

[tomczak]

Many thanks. I'm still not sure why almost black would be clipped when converting AdobeRGB to sRGB, even with relative colorimetric intent.

A related question - how to tell which rendering intent(s) are actually supported when converting between two colour profiles, and when preserving white point and black point compensation options can be used?

[JvdW]

I suppose real black should not be clipped when converting color spaces, but very dark saturated colors will. And you are working on an uncalibrated monitor. Did you check the pixel values in the blocked area's with the readout tool?

I am not sure about the black and white point compensation, but there are two kinds of profiles; matrix based and table based. Converting to a matrix based profile will allways result in relative colorimetric rendering. With table based profiles the options depend on which tables are available, in general there should be 4 options; relative and absolute colorimetric, perceptual and saturation. But I don't know how to tell which options are available without the use of specialist software.
One clue to make an educated guess wether a profile is matrix or table base is the size of the profile file on the harddisk. Matrix based profiles have a much smaller footprint than table based profiles. Matrix based profiles are <20kb (often just about 0.5kb) while table based profiles in general are much larger, 400kb-1Mb.

In general all working color spaces and most monitor profiles are matrix based while all printer profiles are table based. But there are exceptions! At http://www.color.org/srgbprofiles.xalter you can download a V4 sRGB profile that is table based and allowes different rendering intents. V4 profiles are not compatible with all vlavors of windows and applications.

[tomczak]

Thanks again! The V4 sRGB profiles: I read the ICC description and a white paper there and it's a minefield! Mathematically it's fascinating, but practically? - I'm not sure. How do you use the new sRGB profiles, the 'old' sRGB, the AdobeRGB and the printer profile (and the rendering intents between them) in some unison? Is is worth juggling several versions of sRGB?

Cheers!

[JvdW]

"Minefield" is quite an accurate metafore for color management :)

The way I look at color management and the use of different color spaces is to use them as tools to preserve colors I captured and to get the output I want.
I shoot RAW and develop to 48 bit TIFF in ProPhotoRGB, after some editing this becomes my 'original'.
I have a calibrated monitor (sRGB) and a custom printer profile (slightly larger than sRGB, its an old printer). To get a print that matches the monitor I use the previously mentiond workflow: convert colorspace to sRGB (rel.col) then print with perceptual intent. For about 99% of all images this is good enough for me. For some pictures that contain a substantial amount of saturated colors it is worth to use the V4 sRGB with the perceptual intent to make a 'working copy'. This image contains all colors of the original squeezed into the much smaller sRGB color space that I can see on the monitor and print (of course now the colors aren't 'accurate' anymore). Now I can edit this image without any clipped colors and print it with perceptual intent.
To reduce the amount of 'squeeze' when using the V4 sRGB color space I sometimes go back to the RAW file, desaturate some colors if neccessary, develop to AdobeRGB and then convert to V4 sRGB with perceptual intent.

As you can see this workflow is very much tailored to match the gamuts of my monitor and my printer. Modern inkjet printers have gamuts close to, or even exceeding AdobeRGB. In that case it makes more sense to convert to AdobeRGB (rel.col) and then print with perceptual intent.

As with most things in photography there is no best solution, it depends on the image and what you want to achieve.