Why isn't the linear response curve actually linear?

[kpz]

How do I get a linear response curve in Capture One? The "Linear Response" curve is not actually linear when I try to apply it, and I would like to fix that.

Under the heading "Blue Step Wedge example 1" on this page you can find a download link for a synthetic .DNG with varying tones of blue. http://www.markshelley.co.uk/Astronomy/Processing/ACR_Critique/acr_critique.html

The .DNG is constructed so that the ratios of the RGB color channels remain constant on each tone.

I opened the file in C1, set the curve to "Linear Response," and set the ICC profile to .DNG neutral.

The five lightest blue tones show the following RGB values:

138/185/250
98/134/194
68/96/144
45/67/104
27/44/73.

The 8th tone shows 5/9/18. One can easily check these ratios are not constant. I get similar (but different!) values from using the profile "Phase One Effects No color correction."

Could someone please kindly tell me how to render this .DNG accurately?

[Doug Peterson]

"Linear Scientific" (for cameras where it is listed) is 100% linear. "Linear" has a slight rolloff to provide hue uniformity into the highlights and useful/practical rendering of pixels that would otherwise clip.

Free discussion of this around minute 26 of in our Process Control webinar: https://dtdch.com/cultural-heritage-webinars/
More on this and related topics here (not free): https://dtdch.com/color-reproduction-guide-cultural-heritage/

[kpz]

Thanks Doug. I watched the webinar and it was informative. Let's ignore all values above ~200 due to the rolloff. I'm still having trouble getting a linear response even for the lower values.

For instance, the third-brightest tone has a G/B ratio of 96/144=.66. This decreases as the tones darken, to for example 9/18=.5 at the eighth step, and then even more for the deep shadows.

Opening the file in RawDigger confirms that the ratios should be exactly constant. Do you know what is causing the color shift?

[kpz]

There's definitely something not right with either the tone curve or the ICC profile noted above.

I used a gray step wedge to detect the problem. You can download it here:
https://www.rawdigger.com/howtouse/rawdigger-histograms-display-modes

After equation (6) here you can find reference values for the wedge:
https://www.rawdigger.com/howtouse/overriding-raw-converter-default-adjustments-settings

I get can the right values for the wedge both in RawDigger and Version 1 in Lightroom. (For the latter only in the brighter tones, but as noted in the article this is normal.) So the file is fine. But C1 is giving some weird values that do not appear on that table.

In the proof menu I switched back and forth between Adobe RGB and sRGB and that didn't help.

This is really baffling me.