A faster method of monitor calibration

[Frans Waterlander]

Matching the colors and brightness of images on the monitor and prints viewed in our digital darkroom lighting is easier said than done. The common approach is to starting with values for monitor brightness and white point based on  recommendations, past experience, gut feel, common wisdom, etc., followed by a sequence of evaluating results, making adjustments, evaluating results again, etc. etc. There are certainly starting ranges for both parameters that can be consider reasonable, like 80-150 cd/m^2 or nits for brightness (also called intensity) and 5000-6500K for white point, but that's just the beginning and no guarantee for coming even near anything reasonable.

There is a faster and potentially more accurate method. Rather than starting with some more or less arbitrary monitor brightness and white point values, we can use actually measured values, eliminating the initial guesswork and eliminating or significantly reducing the subsequent trial-and-error cycles.

Here is how it works:
1. Measure the color temperature of your print viewing booth or viewing area with the same sensor used to calibrate the monitor and use it for the monitor white point. This approach eliminates or greatly reduces factors that can impact the actual color temperature of the viewing light:
- difference between specification and actual color temperature
- impact of power supply variations on actual color temperature
- sensor inaccuracies
I use the NEC SpectraSensor Pro in Ambient Light Measurement mode (the sensor would overload in Colorimeter Window mode), pointed upward at two SoLux bulbs hanging from the ceiling.
2. Measure the brightness of the paper in the viewing booth or viewing area with the same sensor used to calibrate the monitor and use it for the monitor brightness.
I use the NEC SpectraSensor Pro in Colorimeter Window mode. I hold up a blank sheet of print paper in a position that I normally use to evaluate my prints and aim the sensor at the center of this paper at the same angle that I normally look at the prints, at a distance of about 2 inches to prevent the shadow of the sensor to fall within the sensor field of view.
3. Calibrate the monitor using the measured values.
4. Evaluate the monitor-print match and adjust in the conventional way if at all necessary.

In my case, the measured color temperature and brightness values were exactly the same as the ones I had come up with years ago going through the trial-and-error method. No further adjustments were needed.

I suggest that this method has the potential of being more accurate than the trial-and-error method because it avoids subjective human light perception and the uncertainty when having to decide if one set of values really results in a better match than another.

A word of caution about step 2: in hand-held viewing situations (like mine) the measured values of light reflected off of the paper will vary a lot when you change the angle of the paper relative to the light source. This is nothing new as that is exactly what happens when you view your prints. Observe the values as you change the paper angle and choose the value that represents your viewing habits best.

[Doug Gray]

Matching the colors and brightness of images on the monitor and prints viewed in our digital darkroom lighting is easier said than done. The common approach is to starting with values for monitor brightness and white point based on  recommendations, past experience, gut feel, common wisdom, etc., followed by a sequence of evaluating results, making adjustments, evaluating results again, etc. etc. There are certainly starting ranges for both parameters that can be consider reasonable, like 80-150 cd/m^2 or nits for brightness (also called intensity) and 5000-6500K for white point, but that's just the beginning and no guarantee for coming even near anything reasonable.

I suggest that there is a faster and potentially more accurate method. Rather than starting with some more or less arbitrary monitor brightness and white point values, we can use values that are based on actual measurements, eliminating the initial guesswork and eliminating or significantly reducing the subsequent trial-and-error cycles.

I think we all feel the same way and do it when we have the tools. If you watch Digital Dog's video "Why are my prints too dark" he walks through a very good measurement based approach quite similar to what you are suggesting (not just using a default 6500K). Take a look.

Here is how it works:
1. Measure, rather than assume, the color temperature of your print viewing booth or viewing area with the same sensor used to calibrate the monitor and use that value for the monitor white point. This approach eliminates or greatly reduces factors that can impact the actual color temperature of the viewing light:
- difference between specification and actual color of the lighting
- impact of power supply variations on actual color of the lighting
- sensor inaccuracies ; the same sensor is used for measuring the lighting and the monitor
I use the NEC SpectraSensor Pro in Ambient Light Measurement mode, pointed upward at two SoLux 5000K bulbs hanging from the ceiling; I ignore the "illuminance very high" warning; it's the color temperature I'm interesting in. I measured 5200K.
2. Measure, rather than assume, the brightness of the paper in the viewing booth or viewing area with the same sensor used to calibrate the monitor and use that value for the monitor brightness.
I use the NEC SpectraSensor Pro in Colorimeter Window mode. I hold up a blank sheet of print paper in a position that I normally use to evaluate my prints and aim the sensor at the center of this paper at the same angle that I normally look at the prints, at a distance of about 2 inches to prevent the shadow of the sensor to fall within the sensor field of view. I ignore the color temperature; it's the brightness or intensity value I'm interested in. I measured 105 cd/m^2.
3. Calibrate the monitor using the measured values.
4. Evaluate the monitor-print match and adjust in the conventional way if at all necessary.

In my case, the measured color temperature and brightness values were exactly the same as the ones I had come up with years ago going through the trial-and-error method. No further adjustments were needed.

I suggest that this method has the potential of being more accurate than the trial-and-error method because it avoids subjective human light perception and the uncertainty when having to decide if one set of values really results in a better match than another.

This method, of course, does not alleviate issues caused by poor-quality sensors, lighting, etc.


A word of caution about step 2: in hand-held viewing situations (like mine) the measured values of light reflected off of the paper will vary a lot when you change the angle of the paper relative to the light source. This is nothing new as that is exactly what happens when you view your prints. Observe the values as you change the paper angle and choose the value that represents your viewing habits best.

Good observation in your last paragraph.  In fact if you want to get the best match between a soft proof and a hard proof you need to illuminate the print at an angle such that you get no reflections. This varies from about 5 degrees off axis for glossies to about 25 degrees for Pearl and semigloss surfaces.  Technically, if you illuminate at 45 degrees you should always be safe because that's how printed profile sheet colors are read by spectrophotometers.  One of the great things about matte prints is little to no difference as to the angle of illumination. This freedom comes at the cost of reduced gamut.

[Frans Waterlander]

Andrew's approach is very different from the one I described. He doesn't measure the color temperature of the lighting or brightness of the paper to determine monitor white point or brightness, but goes through a time-consuming trial-and-error exercise. The angle of illumination of even matte paper has a big impact on the measured and perceived brightness, so that's why I advised some extra care.

[Doug Gray]

Andrew's approach is very different from the one I described. He doesn't measure the color temperature of the lighting or brightness of the paper to determine monitor white point or brightness, but goes through a time-consuming trial-and-error exercise. The angle of illumination of even matte paper has a big impact on the measured and perceived brightness, so that's why I advised some extra care.

Sure the angle of illumination makes a big difference in brightness. But not in color temp or xy coordinates.

Illumination, the scale of XY, and Z is reduced by the Cos of the illumination angle from the viewing (perpendicular to the print) axis.

You said you ignored the color temperature. Note that the xy coordinates determine not only the corrected color temp but  a green/magenta tint as well. But you said that this had already matched from a previous effort. Others might not be so lucky so the process needs to include that.  They will need to get the xy values of the monitor match to a blank paper target as well as the brightness to match the hard proof as illuminated. Andrew's process walks through that and it seems quite clear to me though he concentrates on the CCT as many tools do not allow setting xy monitor white point values. He does mention that though.

[Frans Waterlander]

Sure the angle of illumination makes a big difference in brightness. But not in color temp or xy coordinates.

Illumination, the scale of XY, and Z is reduced by the Cos of the illumination angle from the viewing (perpendicular to the print) axis.

You said you ignored the color temperature. Note that the xy coordinates determine not only the corrected color temp but  a green/magenta tint as well. But you said that this had already matched from a previous effort. Others might not be so lucky so the process needs to include that.  They will need to get the xy values of the monitor match to a blank paper target as well as the brightness to match the hard proof as illuminated. Andrew's process walks through that and it seems quite clear to me though he concentrates on the CCT as many tools do not allow setting xy monitor white point values. He does mention that though.

The point is though that whatever final tweaks may or may not be necessary, you get to that point a whole lot quicker. And personally, I put a whole lot more confidence in what you have at that point, based on measurements, than what you would have at that point, based on eyeballing and subjective judgment.

[Doug Gray]

The point is though that whatever final tweaks may or may not be necessary, you get to that point a whole lot quicker. And personally, I put a whole lot more confidence in what you have at that point, based on measurements, than what you would have at that point, based on eyeballing and subjective judgment.

It isn't all that hard but it does require some degree of eyeballing. This is because 10nm spectros can read colors from spikey monitor illuminants inaccurately but also people with normal color vision also vary. Each of us has our own color matching functions but, unfortunately, there is no way outside of a fairly sophisticated lab, to find out what they are. So we use the original CIE 1931 2 degree ones. These work well enough for most purposes but can and do provide inconsistent results with different monitor back illuminant sources. If you set an LED monitor and a CFL monitor to exactly the same xy coordinates and cd/m2 luminance they will often look different. Worse, some people will see bigger or no differences than others due to individual variation.

So here is what I do. It is very slightly different from what Andrew does but quite close.

1. Start with a good profile of a paper that is OB free.
2. Illuminate that paper on your hard proof viewing platform at an angle, preferably close to 45 degrees but always in such a way to avoid specular reflections from the surface. The brightness level from the paper is up to you but needs to be slightly below what you expect to run the monitor at.
3. Measure the xy coodinates of the illuminated paper.
4. Now profile the monitor for the same XY coordinates and a luminance value about 25 % higher than the paper's luminance. This will be changed later. It's needed so you can adjust luminance and color to the paper without clipping a channel at 255.
5. Open a new image in "white" in Photoshop. Select soft proof and Absolute Colorimetric mode in view proof.
6. Using curves or another preferred tool, drag each line (not a point within the line) down until the monitor's luminance, saturation, and hue match the illuminated blank paper on the hard proof viewing stand.
7. Once you have a match, reset soft proof so you have a normal, non-proof view but don't change any values in the now altered "white" image. Read the xy coordinates and Luminance of the monitor's display of this image.
8. Now use those values to create a new monitor profile. That is the final profile you will use.

That process will adjust for individual variations in color response as well as instrument variation and your soft proofing will match your hard proof (make sure you select view paper white).  Additionally, they will match different papers as some have slightly different hues and reflectance. No need to change for different papers.