4 Replies Latest reply on Jul 14, 2010 8:19 AM by Larry Tseng

    Musings on the Kruithof Curve

    Larry Tseng Level 1

      Hi All,

       

      When I first saw the Kruithof Curve at Wikipedia (http://en.wikipedia.org/wiki/Kruithof_curve), I wondered if it had anything useful to say about adapting prints made for one viewing condition to a different viewing condition without further processing and reprinting. Suspecting that a line drawn between the two points marked Tungsten halogen (MR-16) and D65 on the chart might produce a useful isoline of some kind, I prepped an image to look its best at the low end of the T-h/D65 scale, then took measurements at a number of points up the line (Argyll spotread with i1Pro ambient head), varying the distance of the print from various Solux light sources to arrive at the best illuminance for each CCT.

       

      Even though the exercise was as much tinkering around as anything else, I came away with a strong impression that you can use the T-h/D65 line on the chart to map between different CCT/Illuminance pairings that sit on the line. So, for example, take a print that looks good under 3600K @ 300 lux and you can have it looking just as good under 5000K by selecting the right illuminance, which is 3000 lux. But display it under 300 lux? Well, not so good.

       

      Having decided early to play with the monitor next, I didn't check to see if one might be able to do the equivalent with either the top or bottom curves of the chart, which could then open the way for a whole series of isolines for mapping any CCT and illuminance combination to any other combination.

       

      For the monitor, I substituted appropriate luminance values in place of illuminance values using the formula (lum = illum x reflectance)/pi. The measurement range was limited by how far I could push the NEC LCD2180-WG and SpectraView II. When I couldn't resolve some problems related to the low end of the T-h/D65 line, I opted instead to calibrate and profile for just three conditions that sit on the T-h/D65 line, then asked whether I would have arrived at the same points if I had a better way to vary luminance while keeping CCT constant. Relying on memory matches, I thought that I probably would have (arrived at the same points).

       

      While that workaround leaves much to be desired, what struck me as I was going over the figures is that images, prepped to look their best along the T-h/D65 line, cannot be displayed on a monitor calibrated to 5000K unless it can generate a white luminance of 812 cd/m2, the luminance equivalent to 3000 lux on paper (reflectance factor = 0.85). They'll do fine under lower luminance conditions, such as:

       

      • 4000K @ 162 cd/m2 (600 lux equivalent), if you can get it up that far

      • 3600K @ 81 cd/m2 (300 lux equivalent)

      • 3400K @ 54 cd/m2 (200 lux equivalent)

       

      Conversely, an image prepped to look its best under 5000K @ 100 cd/m2 (370 lux on paper) will look its best without further prepping under 3500K lights when illuminated with a very low 150 lux--I get this number by tracing the lower curve of the Kruithof funnel, assuming that one could do that. And lower still if one started out with a higher CCT like 6500K, or a lower luminance value like 80 cd/m2.

       

      Thoughts, anyone?

       

      Larry

        • 1. Re: Musings on the Kruithof Curve
          Tim Lookingbill Level 1

          Larry, I wouldn't know what to think about a chart that attempts to calculate how our eyes are going to perceive any given color temperature at any established luminance. I've tried to calibrate to 5000K from 6000K with my i1Display software but it's just too yellowish making all cool colors close to neutral look dull no matter how bright or dark I adjust the backlight. Probably the limitation of the software and/or my aging display. Not sure.

           

          I've included an image of what I'm talking about. Since trying to establish the exact appearance of any given Kelvin is pointless I've just called it a warmer looking color temperature from what our eyes consider as neutral.

           

          I do know the more warmer looking the display the brighter it has to be to induce adaptation more quickly. The more and faster photons reach the eyes the quicker we adapt. It's what happens when you go from a 2800K lit dim living room to bright outdoor daylight. The eyes quickly adjust for the blue cast as neutral. Stay outside long enough and go back into the 2800K living room and it takes a lot longer to adapt.Solux3500KemulatePNET.jpg

          • 2. Re: Musings on the Kruithof Curve
            Larry Tseng Level 1

            Hello Tim,

             

            The phenomena that I was exploring with the chart is one where color balance

            is affected by the lighting level of the image being viewed. I haven't

            studied the very complex CIECAM02 model, so I don't know if this particular

            relationship between CCT and luminance is accounted for there. It is

            definitely not something to worry about in the grand scheme of things. As a

            practical matter, the effect is small: take a print that you are familiar

            with, and under your favorite lamp, vary the distance in order to vary

            luminance. The further away from the light source you get, the cooler the

            print gets; move closer to the light source, the warmer it gets. Somewhere

            between cool and warm is an optimum point.

             

            As for not being able to calibrate to warmer CCTs, the image that you

            included in your post may actually contain a clue: viewed under 6500K @ 100

            cd/m2, it's an early morning shot; under 3500K @ 80 cd/m2, a late afternoon

            shot. If you had aimed for a late afternoon feeling, then I would say that

            your monitor is a lot warmer than your monitor profile says it is -- so warm

            that your eye refuses to accept monitor white as neutral white. Just a guess

            at this point.

             

            Larry

            • 3. Re: Musings on the Kruithof Curve
              Tim Lookingbill Level 1

              I don't see prints changing in hue from warm to cool depending on distance from the light source. That's not saying you don't see it.

               

              I've got a print of the PDI skin tone section on Epson Ultra Premium Glossy Photo Paper that looks quite neutral and matches my 6000K display under both GE Sunshine (Chroma 50) fluorescent tube and Solux 4700K task lamp. I move the print back and forth and it doesn't change color at all. It just gets darker. Maybe what you see is subtle as you say. I don't see it.

               

              Now if I stare at the print for a long period time after long edits on my 6000K display then I start to see a tinge of warmer color cast in the print but not from pulling it away from the light.

               

              Changing the brightness of the display doesn't behave the same as changing the distance of the print to the light source in my experience. However there is a noticeable color cast change when lightening and darkening the backlight on my LCD. My neutral looking 6000K turns slightly blue when darkening and warm when brightening.

               

              As for the intent behind my posted image I was trying to point out another aspect that sort of resembles a "double profiling" effect when calibrating a display to a warmer setting and the color errors that come about viewing images shot outdoors containing their own cool/warm color constance and color temp appearance properties previously established through edits on a comparitively more neutral looking display.

               

              You'ld expect the eyes to adapt and correct for the color error caused by the lights warmer cast on the print as seen in the jade blue waterfall on the far right going dull and colorless on the second image. I don't think display calibration software would correct for this in bringing back the vibrant blue color.

               

              So who and what fixes this? A printer profile? Or edits on a display calibrated to a warmer color cast?

              • 4. Re: Musings on the Kruithof Curve
                Larry Tseng Level 1

                I don't see prints changing in hue from warm to

                cool depending on distance from the light source.

                 

                Sorry Tim, I should have said more.

                 

                Imagine the following test setup: a luxo type lamp that you can aim into the

                distance, a warm 60-watt incadescent bulb, a dark or dimly lit room, and a

                colorful print. With the lamp behind you one foot or so back, and with print

                in hand, move closer to the lamp. The print should look brighter, warmer and

                more vibrant than before. Move away from the lamp and the print becomes

                darker, cooler, duller.

                 

                Next, imagine that you are adjusting this print to look its best as you

                would in Photoshop, except that the only thing that you can do to the print

                is move it closer or further away from the light source. When you do this,

                there should be some point between too bright/warm/vibrant and too

                dark/cool/dull, where an optimal balance is reached. This is where I would

                measure both the light intensity on the surface of the print and the

                correlated color temperature (CCT) of the light before moving on to a

                different set of readings involving a different light source.

                 

                What I found was that as you go up the scale from low CCT lamps to higher

                ones, light intensity levels need to go up as well, to retain balance. But

                there has to be some way of measuring light intensity levels to be able to

                see this. Argyll Spotread together with your i1Display, perhaps, if you want

                to give it a go.

                 

                Regarding the images that you posted:

                 

                ... You'ld expect the eyes to adapt and correct for the color error caused
                by the lights warmer cast on the print as seen in the jade blue waterfall
                on the far right going dull and colorless on the second image...

                I printed two copies of the top image to display under two different light

                sources--a Just ColorMaster  (illum 843.6 lux, CCT 4497K, CRI 92.4) and a

                Solux PAR lamp (illum 787 lux, CCT 3454K, CRI 91.1). When I stopped viewing

                both copies simultaneously, the differences between the two dropped

                significantly. If I had seen only the Solux version, I wouldn't have known that

                anything is amiss. If I had shot that particular scene and adjusted it to

                the way I wanted it under the cooler temp, it is quite possible that I would

                have trouble accepting the warmer version.

                 

                ... I don't think display calibration software would correct for this in

                bringing back the vibrant blue color.

                So who and what fixes this? A printer profile?

                Or edits on a display calibrated to a warmer color cast?

                 

                ProfileMaker can produce a printer profile that compensates for the effect

                of the warmer light. The adjustment is more pronounced with yellows and

                reds. In a side by side comparison however, I couldn't see any difference in

                the blues in your image.

                 

                I don't know to what degree chromatic adaptation is implemented in Photoshop

                to correct for source and monitor color temperature differences. The sRGB

                source that you provided had the feel of an early morning shot when

                displayed on a 6500K monitor, and a late afternoon shot on a 3500K monitor.

                I don't yet know what to make of this.

                 

                Larry