Does “Color Science” even matter?

JimKasson wrote:

yogi4fitness wrote:

JimKasson wrote:

yogi4fitness wrote:

just Tony wrote:

A question has been posed in this thread as to whether color response sensitivities can be zeroed out between one camera and another if we take control with custom user-built profiles as an equalizer.

I've been making profiles for each new camera that comes into my house using the ColorChecker SG card and LumaRiver Profile Designer software. I revisited that today with one layout of the chart and the lighting and made calibrations frames with a Z7, Z9, and GFX 100S.

The settings in LumaRiver were almost entirely middle of the road selections; if "Adobe ACR" in one place or another was offered I selected that because I use Adobe SW primarily for my raws. The one creative option I indulged in was the "Skin and Sky" Tone Reproduction Operator. I always use that because it references important memory colors, and my expectation was that they'd try pretty hard to achieve a uniform look on them.

Note that squares E2, J2, E5 and J5 are the corners of a set of 24 squares that are nominal counterparts to the CC24 card. But for those not yet familiar with the SG, all of the squares are glossy in order to make possible those pretty saturated squares. This makes the illumination a bit trickier than the matte surfaced CC24. The illumination here came from a pair of Solux 4700K halogen floods at 45 degree incidence angles (left and right). The card was held outwards from it's support on a long stick. The card "saw" only a very dark room and two very bright bare lamps.

The first image shows a complete frame. Both the central positioning and the use of f/11 provided a very uniform illumination. LumaRiver takes this a step further by computing a flat frame correction from the white patches.

After creating the profiles I processed the source raws in the profiles that was made directly from them. That insures the best possible chance of the colors landing exactly as the profile was designed to do. In Lightroom I zeroed out every adjustment except for selecting the associated profile for each camera, and using the White Balance eye dropper on patch G5.

Here are the images from the three cameras seen closer; note that they have been converted to sRGB for web-friendliness. That may render them unreliable for extracting measurements. The numbers in my spreadsheets below were collected under LR's working space which is similar to ProPhoto.

Z7, 105mm S at f/11 ISO 100 0.6 seconds

Z9, 105mm S at f/11 ISO 100 0.6 seconds

GFX 100S, 110mm GF at f/11 ISO 100 1.3 seconds

You might have spotted the difference in the exposure settings. I ran 1/3 stop brackets and selected the frames in which the brightest patches landed at L* = 92. However the Fuji persisted on requiring double the nominal exposure.

Why did Fuji require double the exposure at the same fstop and base iso? Does that mean the fuji is half as sensitive as the Nikons?

I have previously noticed that the base ISO 64 Nikon cameras are more sensitive than you’d expect.

How do they achieve it? Are their sensors more sensitive to light?

This is a matter of how the specifications of the camera are created. I think that product management plays a large role here.

Ah I get it. I was under the assumption that the iso scale is a standardized scale, meaning an iso 100 was the same across all brands. I guess iso 200 of one brand could mean iso 100 of another brand.

That's something to remember the next time you compare cameras on the Engineering Dynamic Range chart at photonstophotos.

In Lightroom I switched rapidly from one frame to the next. For the vast majority of color patches I saw no significant differences from one camera to the next. At least nothing that I would consider troublesome during creative color adjustments.

However the saturated reds and reddish purple were significant outliers. I recorded L*a*b* coordinates from several patches and computed the differences. Note that the delta E figures are merely the Pythagorean distances, I didn't use the rigorous formula.

I selected the 7 largest mismatches to show below as well as representative Skin and Sky colors, to see if the "Skin and Sky" look gave a uniform result:

I didn't expect those differences between Z7 and Z9. If people are going to toss around phrases like "Nikon Colors", their declarations won't be complete if they don't also mention which Nikon and which colors!

This chart shows the directions of the color differences:

So I'm seeing significant b* differences in saturated reds between the Z7 and Z9, while ironically the GFX 100S actually slipped right between those two! In b* results on saturated reds the Fuji is more similar to each of these Nikons individually than those are to each other. Did Nikon tweak the blue filters of the Z9 CFA? There is a pretty pervasive L* and a* difference for saturated reds between these Nikons and this Fuji.

The main takeaway for me though is the rather good convergence of these three cameras in most colors and in particular the skin and sky colors with due credit going to custom profiles that have matching goals.

-- hide signature --

https://blog.kasson.com

-- hide signature --

https://blog.kasson.com