Role of sensor in color rendering?

Started 3 months ago | Discussions thread
OP lélé Contributing Member • Posts: 595
Re: Role of sensor in color rendering?

JimKasson wrote:

lélé wrote:


  • I'm a hobbyist, with limited technical background on that particular topic...but eager to learn!:)
  • I'm not a native English speaker, so my wording may be inacurate not only because of my lack of technical knowledge, but also language/translation issues.
  • That said, I think I roughly understand the overall idea and differences behind 'colorimetry', 'color appearance' and 'pleasant colors'.
  • I don't want to (re)start any brand war!



I was playing around with Lumariver Profile Designer, an X-Rite ColorChecker SG color target an my Sony a7R IV. I was very surprised to see the color differences in skin tones I got with 'matrix only' profile. Those were backing up impressions I got, like hue shifting to more 'greenish' when in the shadows.

I also played around with other cameras files, in particular Canon ones, and this issue was less noticeable.

I was wondering how the camera sensor, in particular the CFA design, could impact that.

Mays you have more info' ?

The CFA dyes/pigments affect the color rendering. They are not the most important thing that affects colors in the final image; the raw development is more important.

If cameras obeyed the Luther-Ives condition, and saw colors in a way that can be directly related to the way that color-normal humans see them, cameras would all see color in the same way. But they don't.

Some background:


Hello Jim!

Thank you very much!
I've read all that with a lot of interest!

From a practical perspective, I'm struggling to create color profiles with 'pleasant skin tones' for my Sony a7R IV and my former Sony a7 III.

I understand that 'pleasant skin tones' is the confluence of personal preferences, subject, make-up (if any), context, lighting, illuminant(s), lens (although I'm not sure it has such an influence, outside of contrast drop and slight WB shift...but I may be wrong), camera sensor, camera matrix calibration, further color corrections through the profile hue/saturation/value tables and subjectives adjustments through the profile table. And I probably omit a lot of other factors...I'm no specialist, unlike you.

I also understand that skin is a multi-layer material, that light interacts with in a complex way, with specular reflections, some sub-surface scattering, etc. Unlike the solid patches of y X-Rite ColorChecker Digital SG.

Finally, It got a good idea - at least for the complexions I shoot the most - of the 'preffered colors' I'm aiming at and what subjective adjustements I can make to achieve that.

But I've noticed some 'strange' behavior when photographying people with caucasian complexion and my Sony cameras. In particular, skin tones turning to a "brownish and slightly' greenish' hue when in the shadow and/or viewed from an angle (in an environnement with no foliage or whatsoever to scatter green light, neutral colored clothes, etc.). This is partially reduced by the hue/saturation/value tables Lumariver Profile Designer creates from my color target shots (I tried with and without it to compare), but far from perfect. This is extremely difficult to correct through subjective adjustements without affecting other colors, like colors of some animal fur, wood, etc. or simply the skin tones of people with darker complexion.
I have also noticed skin tones seems hugely sensitive to the illuminant spectrum. For example, shooting with a Godox AD200Pro flash + soft box, I have VERY different skin tones, no matter I precisely adjust the white balance with a color target reference. I would be expecting such VERY large differences with lfuorescent or LED light, not flash...
I have downloaded numerous samples from various sources (in particular DPreview) and noticed similar behavior.

Finally, I have also downloaded samples taken with Canon cameras. I have created a color profile from the Canon EOS R DPReview studio scene and Lumariver. I know that their CC24 is (at least) from 2013, so it may have faded. I also don't know the exact illuminant they use. But the idea behind was to create a 'neutral' profile, with no subjective adjustment except the tone curve.
I carrefully inspected the Canon EOS R samples using the created profile and did not notice the issues with skin tones. They look less pleasant to me then with the in-camera profile (or it Adobe sibbling) subjectives adjustment, but no 'horrible' hue shift.
I know this is absolutely not a scientific method, but it raised the question : is the sensor causing the issue I have with skin tones?

I used to belive that the difference essentially lied in the RAW conversion and, more specifically, in the subkective adjustments made in the color profile. But I'm starting to think it's also a sensor thing.

And this is puzzling me out because I struggle to got the preferred skin tones I want, even spending hours doing subjective adjustements to the color profile.

 lélé's gear list:lélé's gear list
Nikon D300 Sony a7R IV Canon TS-E 17mm f/4L Nikon AF-S DX Nikkor 17-55mm f/2.8G ED-IF Fujifilm XF 35mm F1.4 R +13 more
Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow