Role of sensor in color rendering?

[J A C S]

This portrait looks nice to me. Are you seeing those greens or you are talking about the RGB values?

 

[lélé777004]

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 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.

Have you done the same for your Sony a7R IV and Sony a7 III ?

If the results are not to your liking, have you asked Lumariver technical support to help?

PS: I understand that 'preferred colors' is an extremely subjective thing. That getting these 'preferred colors' might require tons of subjectives adjustments. That there are tons of unknowns in the samples I checked, as well as in my own photos (illuminant, light scattered/reflect by surrounding objects, etc.). That my CCSG color target is not perfect. That my technique to shot it is not perfect. That the illuminant is not the exact same one. Etc.
But what is puzzling me is that in a most photos with caucasian skin tones I notice something that is unpleasing me. Namely a shift towards green of caucasian skin tones when view from an angle/in the shadow. And on the CCSG target shot I have F2 and D8 patches leaning towards magenta, and some darker ones leaning towards the opposite direction (especially I8). I don't know to which extent this is normal (might be caused by a lot of different things), but it seems consistent with my subjective observations.

 

[lélé777004]

This portrait looks nice to me. Are you seeing those greens or you are talking about the RGB values?

Seeing first!

But subjective observation confirmed by HSB values.

I know that is is perfectly normal that hue varies accross her face. But here there is something that looks (subjective) wrong to me. That's not the worst example I have through, it's worse on my children who have a different complexion...

 

[JimKasson]

But do you think that, for example, the hue shift towards a 'slightly greenish brown' of caucasian skin tones when viewed from an angle and in the shadow is really caused by cross-talking?

Brown isn't a hue. There's a green hue shift, and something else. What is that something else? Luminance? Desaturation?

Measure the skin tones in Photoshop in CIELab. Make the skin tones look right. Then measure again. Post the two sets of CIELab values.

When you say at an angle, do you mean that the lens axis is not orthogonal to the person's skin? Maybe you should post an example photograph.

Shadows are usually bluer than sunlit areas, because they are lit by the sky.

Jim

Hello again Jim!

Got the Sony a7R IV before lockdown, so limited pictures library for the moment. All pictures with skin tones I got are family portraits, so I won't post them on a public Internet forum.

That said, here is an example from DPReview samples that illustrates what I'm talking about:

https://www.dpreview.com/sample-galleries/0635778599/sony-a7r-iv-sample-gallery/8559074145

Here is what it's looks like converted with a matrix only color profile, then exported to sRGB:

View attachment d2e75f53ce6d459cae447c04b496b8dc.jpg

With color patches (from 11*11px average):

View attachment 04e3f9975636407ea40e0a24d3f4a98e.jpg

link to full resolution : https://ibb.co/JkYpfLy

Do that again and put in the Lab values.

Of course, I do expect lightness, but also hue and saturation variations accross her face. Due to natural variations in her skin color. Due to her makeup (she obviously has some). Due to the lighting (the main light may have a color cast). Due to the angles between light sources, skin surface and camera. Etc. In addition, her skin may be catching some indirect light with strong color cast, scattered/reflected by her surrounding. Finally, we don't know the particular lighting that was used for this shot.

With all these reservations about that particular example (please take it as an illustration, not more! ), I'm surprised how skin hue shifts towards green when viewed from an angle. This is something I noticed in almost all my familly portaits, no matter they were shot under direct sunlight, shadow, in halogen light (yeah, I still have halogen bulbs at home) or flash. And it's, in my opinion, unpleasant.

This is attenuated when using a color profile with LUT (very same than previous one except LUT, generated with Lulariver from a ColorChecker Digital SG shot under midday natural light). But in order to do so, I have to allow 'sharp bends' in the LUT (Lumariver has two settings for that), because light skin tones patches of the CCSG lean towards negative hues shift while dark skin tones patches leans towards positive hue shift. My understanding is that this may result in some posterization...

--

the last word the last word - Photography meets digital computer technology. Photography wins -- most of the time.

Photography meets digital computer technology. Photography wins -- most of the time.

blog.kasson.com

 

[JimKasson]

Of course, the chart may no reflect actual skins reflective spectra. There might be some errors in some patches (I don't have a spectrophotometer to check). There might be some operator error (it's a difficult target to shoot due to glare and I struggled at the beginning). But what Lumariver tells from the target shot is in line with what I noticed before I started making profiles...

Does this help you understand what skin spectra look like?



Three skin readings, face, palm of hand, back of hand, for two people, and two different CC24 cards.

Jim

--
https://blog.kasson.com

 

[lélé777004]

Disclaimer:

• 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!

=============

Hello!

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' ?

Yellow skin + bluish light to shadows -> greenish reflection. That's what will be a result from a matrix optimized for some colour reproduction accuracy. One can slightly reduce the effect by CFA design, but not without trade-offs.

Matrix transforms are usually not all what you want to use in raw conversion, because of this and other similar problems.

Thanks for your reply.

The reason why I checked pictures with matrix only profile is because I wanted to see what it looked like before LUT color correction. Of couse, outside of this particular testing confifuraiton, I use a profile with a LUT.

Unfortunately, even with a LUT, it's far from perfect. A least to my eyes. It seems that colors that are pretty similar have different shifts, so it's difficut to correct...

Another problem I ran into is these two cameras (a7 III and a7R IV) show a lot of variation in skin tones rendering depending on the illuminant. I was very surprised how completely off the skin tones looked with an AD200Pro flash, despite careful WB adjustment with a target.

 

[Iliah Borg]

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 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.

Have you done the same for your Sony a7R IV and Sony a7 III ?

Could you please answer this one?

If the results are not to your liking, have you asked Lumariver technical support to help?

Yes, but as for the Sony a7 III since I sold it recently I only have color Passport Photo target shots to work with. Not enough skin tones patches to do anything good...

I'll try to email Anders Torger (developper of Lumariver); he was very helpful on other topics. But I don't thinks Lumariver is what causes the problem.

The problem may be in how the shots were taken. It always pays to send the raw files and ask.

 

[lélé777004]

But do you think that, for example, the hue shift towards a 'slightly greenish brown' of caucasian skin tones when viewed from an angle and in the shadow is really caused by cross-talking?

Brown isn't a hue. There's a green hue shift, and something else. What is that something else? Luminance? Desaturation?

Measure the skin tones in Photoshop in CIELab. Make the skin tones look right. Then measure again. Post the two sets of CIELab values.

When you say at an angle, do you mean that the lens axis is not orthogonal to the person's skin? Maybe you should post an example photograph.

Shadows are usually bluer than sunlit areas, because they are lit by the sky.

Jim

Hello again Jim!

Got the Sony a7R IV before lockdown, so limited pictures library for the moment. All pictures with skin tones I got are family portraits, so I won't post them on a public Internet forum.

That said, here is an example from DPReview samples that illustrates what I'm talking about:

https://www.dpreview.com/sample-galleries/0635778599/sony-a7r-iv-sample-gallery/8559074145

Here is what it's looks like converted with a matrix only color profile, then exported to sRGB:

View attachment d2e75f53ce6d459cae447c04b496b8dc.jpg

With color patches (from 11*11px average):

View attachment 04e3f9975636407ea40e0a24d3f4a98e.jpg

link to full resolution : https://ibb.co/JkYpfLy

Do that again and put in the Lab values.

Of course, I do expect lightness, but also hue and saturation variations accross her face. Due to natural variations in her skin color. Due to her makeup (she obviously has some). Due to the lighting (the main light may have a color cast). Due to the angles between light sources, skin surface and camera. Etc. In addition, her skin may be catching some indirect light with strong color cast, scattered/reflected by her surrounding. Finally, we don't know the particular lighting that was used for this shot.

With all these reservations about that particular example (please take it as an illustration, not more! ), I'm surprised how skin hue shifts towards green when viewed from an angle. This is something I noticed in almost all my familly portaits, no matter they were shot under direct sunlight, shadow, in halogen light (yeah, I still have halogen bulbs at home) or flash. And it's, in my opinion, unpleasant.

This is attenuated when using a color profile with LUT (very same than previous one except LUT, generated with Lulariver from a ColorChecker Digital SG shot under midday natural light). But in order to do so, I have to allow 'sharp bends' in the LUT (Lumariver has two settings for that), because light skin tones patches of the CCSG lean towards negative hues shift while dark skin tones patches leans towards positive hue shift. My understanding is that this may result in some posterization...

Here it is: https://ibb.co/7tP6Tkq

 

[lélé777004]

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 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.

Have you done the same for your Sony a7R IV and Sony a7 III ?

Could you please answer this one?

I answered you here: https://www.dpreview.com/forums/post/63969539

Yes, but a7 III ColorChecker Passport only, while a7R IV ColorChecker Digital SG. Also I tried more illuminant for the a7R IV: direct sunlight, cloudy day, halogen...

a7III profile from CCP => no help

a7R IV profile from CCSG => does help, but does not solve the issue, plus I have to play with 'min. chromacity distance' and 'worst compression' settings (allowing more bends in the LUT if I understand correctly) to get a real improvement.

If the results are not to your liking, have you asked Lumariver technical support to help?

Yes, but as for the Sony a7 III since I sold it recently I only have color Passport Photo target shots to work with. Not enough skin tones patches to do anything good...

I'll try to email Anders Torger (developper of Lumariver); he was very helpful on other topics. But I don't thinks Lumariver is what causes the problem.

The problem may be in how the shots were taken. It always pays to send the raw files and ask.

I will follow your suggestion. Thanks!

 

[Iliah Borg]

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 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.

Have you done the same for your Sony a7R IV and Sony a7 III ?

Could you please answer this one?

I answered you here: https://www.dpreview.com/forums/post/63969539

I was referring to your "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", sorry if I'm misunderstanding your reply.

Yes, but a7 III ColorChecker Passport only, while a7R IV ColorChecker Digital SG. Also I tried more illuminant for the a7R IV: direct sunlight, cloudy day, halogen...

a7III profile from CCP => no help

a7R IV profile from CCSG => does help, but does not solve the issue, plus I have to play with 'min. chromacity distance' and 'worst compression' settings (allowing more bends in the LUT if I understand correctly) to get a real improvement.

If the results are not to your liking, have you asked Lumariver technical support to help?

Yes, but as for the Sony a7 III since I sold it recently I only have color Passport Photo target shots to work with. Not enough skin tones patches to do anything good...

I'll try to email Anders Torger (developper of Lumariver); he was very helpful on other topics. But I don't thinks Lumariver is what causes the problem.

The problem may be in how the shots were taken. It always pays to send the raw files and ask.

I will follow your suggestion. Thanks!

 

[lélé777004]

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 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.

Have you done the same for your Sony a7R IV and Sony a7 III ?

Could you please answer this one?

I answered you here: https://www.dpreview.com/forums/post/63969539

I was referring to your "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", sorry if I'm misunderstanding your reply.

Oh, sorry, I did not get you!

Yes, I made a profile from DPReview studio scene CC24 for the a7R IV. I assumed it was a 'pre 2014' CC24 since the studio scene was presented by DPreview end of 2013...
The color I got with the a7R IV profile generated from DPR CC24 vs. my CCSG were slightly different with matrix only profiles, but differences were subtle. The differences are more pronouced with a LUT, to the advantage of the profile generated from my CCSG shot, which deals better (but not great) with the issue I'm having with skin tones.

I did not try for the a7 III...

Yes, but a7 III ColorChecker Passport only, while a7R IV ColorChecker Digital SG. Also I tried more illuminant for the a7R IV: direct sunlight, cloudy day, halogen...

a7III profile from CCP => no help

a7R IV profile from CCSG => does help, but does not solve the issue, plus I have to play with 'min. chromacity distance' and 'worst compression' settings (allowing more bends in the LUT if I understand correctly) to get a real improvement.

If the results are not to your liking, have you asked Lumariver technical support to help?

Yes, but as for the Sony a7 III since I sold it recently I only have color Passport Photo target shots to work with. Not enough skin tones patches to do anything good...

I'll try to email Anders Torger (developper of Lumariver); he was very helpful on other topics. But I don't thinks Lumariver is what causes the problem.

The problem may be in how the shots were taken. It always pays to send the raw files and ask.

I will follow your suggestion. Thanks!

 

[57even]

Tri-stimulus colour spaces don't replicate human vision either

Please explain.

Are you talking about:

• Departures from color-normalcy?
• Variations in color responses within the color-normal range?
• Spatial effects?
• Adaptation?
• Non-physical primaries?

Or something else?

I am talking about the fact that a large range of colours we can see is not reproducable on any RGB or CMY based colour device, and on sRGB, about 64% of them.

So you are talking about the available devices, not tristimulus color spaces in general. XYZ is certainly a tristimulus space, and can encode all the colors that the standard observer can see. It does, however, have non-physical primaries.

I was under the impression that tristimulus implied three visible colours (CIE RGB) rather than unreproducable virtual colours that don't have a corresponding wavelength.

I've never used it that way. I've not seen others make that distinction. Here's an example of the usage that I've seen:

https://en.wikipedia.org/wiki/CIE_1931_color_space

A pertinant quote: "The CIE XYZ color space encompasses all color sensations that are visible to a person with average eyesight. That is why CIE XYZ (Tristimulus values) is a device-invariant representation of color.[5]"

For enough. I was misinformed, but I don't find it very satisfactory.

By the way, I don't think you meant to use the word "colours" after "unreproducible". If you can't see it, it's not a color.

I meant it is a radiation frequency that cannot be produced (visible or not) so there is no way to recreate the colours in CIE XYZ (even accounting for the invisible non colours) using any three actual physical means.

Good sensor colour science (closer to LI) is less import with smaller gamuts, but becomes more problematic with large ones, like the future Rec.2020 standard.

I wouldn't say that the color science of a LI sensor is in any global way better than that of the sensors that we have. The people who design those CFAs know how to get close to LI, but they choose not to because of the sensitivity and SNR issues of LI CFAs. Not sure I agree about the larger gamut making the situation worse, since the memory colors tend to lie within Adobe RGB.

While it's true that most OOG colours are cyan tones, red/blue response can have a fairly serious impact on colours along the line of purples.

Sure, it may not be a big deal, but AdobeRGB red (for example) is distinctly more orange than DCI-P3. Filter designs assuming AdobeRGB as an output space may not cut it in future.

I see no evidence of that, from looking at the numbers. But I can't disprove your speculation. There's a likely exception, and that's spectral primaries,

If the spectral primaries define the extent of the space and therefore the red channel hue, I would have thought that was a problem.

As far as I know, we don't have any RGB displays that use spectral primaries. It would be possible to build one, but I don't think most people would want to pay for it.

Jim

Close enough with Rec.2020. Using OLEDs rather than LCDs makes it easier I guess.

 

[Erik Kaffehr]

Disclaimer:

• 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!

=============

Hello!

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' ?

Hi,

Some thoughts...

Pleasant color may not be accurate color, though accurate color may be a good starting point.

Next, white balance is probably the most important factor in color rendition, as long as illumination is continuous spectrum. Flourescents add their problems of their own.
Regarding skin tones, Andrew Rodney has a nice video on correcting skin tones.

I don't have a lot of personal experience with skin tones. But I did some testing with color rendition of very different cameras, without seeing huge differences.

The P45+ is an MFD back using a Kodak CCD from 2007, A900 is regarded as one of the best compromises regarding CFA design and the A7rII is the camera I use right now.

Differences are better viewed in a stacked TIF available here .

All cameras were tested with profiles generated with LumaRiver using the ColorChecker in each image.

Obvious no skin sample... :-(, sorry!

Best regards

Erik

--
Erik Kaffehr
Website: http://echophoto.dnsalias.net
Magic uses to disappear in controlled experiments…
Gallery: http://echophoto.smugmug.com
Articles: http://echophoto.dnsalias.net/ekr/index.php/photoarticles

 

[lélé777004]

Disclaimer:

• 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!

=============

Hello!

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' ?

Hi,

Some thoughts...

Thanks for your feedback.

Pleasant color may not be accurate color,

I perfectly understand that!

That's why I use Lumariver to make my color profiles, so I can add subjective adjustments on top of the profile.
(For the adjustements that I know I will want in 90+ percents of my photos. What is 'picture specific', I do it in post-processing, picture by picture...)

though accurate color may be a good starting point.

That's my concern!

Next, white balance is probably the most important factor in color rendition,

I completely agree!

That's why I keep my ColorPassport Photo in my bag, so I can get accurate WB.

as long as illumination is continuous spectrum. Flourescents add their problems of their own.

Once again, I completely agree!

I usually try to avoid fluorescent and LED.

Actually, even at home, for my daily pleasure, I'm going back from LED to halogen because I do not like how people and food look like with the LED, despite claimed 'good' CRI...

Regarding skin tones, Andrew Rodney has a nice video on correcting skin tones.

Thanks for the link...unfortunately it won't help with the issue I have.
My problem is not to determine a target color. This is okay.

My first concern is mainly skin color variations accross the face.
I understand it is perfectly normal that skin color varies accros the face: naturally, because of the makeup, depending on the angle of view, depending on the lighting, due to indirect lighting scattered/reflected by the hair, torso and surrounding environment, etc.
But on many, many, of my shots the skin has a 'greenish cast' on caucacian skin tones when viewed from an angle/in the shadow (I don't know if it's linked to illumination, the angle, or both). This does not look nice at all to me. People look sick to me on some photos. And this is a hassle to correct because it involves hue corrections in very selective parts of the image...
As for the color profile, adjusting manually the profile through Lumariver 'Look' tab proved complicated because I have to set opposite adjustments to colors which are very close to each other. So white balance has to be absolutely perfect. So it does not work with mixed lights. So, more broadly speeeking, it may work for one image but not the other. I also tried hue compression, but with unatural results: it get rid of slight color variations accross skin that I do want to keep (or it looks unatural to me) way before the issue is corrected. The best I could do is selective 'look' adjustement that also use lightness as an input (basically adding red and increasing saturation to the darker areas of caucasian skin tones). It works quite well, but it is very exposure dependant (if for some reason the face is more less exposed, it does not word) and cause problems on certain complexions. In a word: it's complicated and likely to end with a picture by picture full adjustment.

The second concern is how skin tones color rendering varies depending on the illuminant. I'm not talking about LED or fluorescent, whose spectra, I know, are real roller coasters. No, I'm talking about thing like daylight vs. flash. Or one flash vs. another one.
Of course, I do expect color temperature and tint difference. That's why I use a CC Passport for WB. I also expect some color rendering differences...but they seem exacerbated by the a7 III and the a7R IV!
For example, I have a Godox AD200Pro and Godox V1. I did a few autoportrait to get familiar with them. I was really suprised how skin tones, despite careful WB adjustment on reference, looked VERY different between the two. And VERY different from daylight. Of course, those are flash tubes light, not a 'perfect' illuminant'. So I expected differences. But not to such extent! I also planned to use both in a two lights setup...but that's impossible because the areas lit by one or the other look to different. There is a tint difference I could get (mostly) rid of with a colored filter (V1 more green), but I'm affraid it goesway beyond that.

Both concerns make using these cameras puzzling for skintones: sometimes they look quite good, sometimes they look horrible. Like if the cameras were ultra-picky about the illuminant when it comes to skin tones. Problem is that I'm no studio photographer: I essentially do familly photos and bought the Sony a7R IV because now that my children are getting older I plan to redo more photography in the near future (portraits/maybe street/macros/cityscape/lanscape), like I used to do when I was a student. At least that was the plan before COVID crysis!

I don't have a lot of personal experience with skin tones. But I did some testing with color rendition of very different cameras, without seeing huge differences.

The P45+ is an MFD back using a Kodak CCD from 2007, A900 is regarded as one of the best compromises regarding CFA design and the A7rII is the camera I use right now.

Differences are better viewed in a stacked TIF available here .

I have no problem with colors except skin tones.

Also, I see people complaining about the reds, greens, or blues: my experiences is that, at least for my usage, those are pretty easy to ajust to taste.

Skin tones it's more complicated, because slight change are more noticeable, what works well on one complexion may not on another...plus there are tons of things that have colors quite similar to skin (wood, sand, hair...) so when adjusting skintones it's difficult not to harm the rest of the image.

All cameras were tested with profiles generated with LumaRiver using the ColorChecker in each image.

Obvious no skin sample... :-(, sorry!

Best regards

Erik

Thanks for sharing.

 

[JimKasson]

For example, I have a Godox AD200Pro and Godox V1. I did a few autoportrait to get familiar with them. I was really suprised how skin tones, despite careful WB adjustment on reference, looked VERY different between the two. And VERY different from daylight. Of course, those are flash tubes light, not a 'perfect' illuminant'. So I expected differences. But not to such extent! I also planned to use both in a two lights setup...but that's impossible because the areas lit by one or the other look to different. There is a tint difference I could get (mostly) rid of with a colored filter (V1 more green), but I'm affraid it goesway beyond that.

I'd suggest using a spectrometer or at least a colorimeter to measure the lighting and help with the gel calculations.

JIm

 

[D Cox]

When you say "Here is what it's looks like", you mean "Here is what it's looks like on my monitor". Monitors vary.

 

[lélé777004]

When you say "Here is what it's looks like", you mean "Here is what it's looks like on my monitor". Monitors vary.

My desktop monitor is calibrated.

You have Lab values in my answer to Jim KASSON: this is objective measurement, not subjective assessement.

What is questionnable - and I'm seeking for answers - is: are these kind of variations 'normal' or resulting from the sensor CFA design?
(Not especially in this particular example, that is nothing more than an illustration I choose for convenience/privacy reasons...)

 

[JimKasson]

Here is what it's looks like converted with a matrix only color profile, then exported to sRGB:

View attachment d2e75f53ce6d459cae447c04b496b8dc.jpg

With color patches (from 11*11px average):

View attachment 04e3f9975636407ea40e0a24d3f4a98e.jpg

link to full resolution : https://ibb.co/JkYpfLy

Do that again and put in the Lab values.

Here it is: https://ibb.co/7tP6Tkq

The hue angle shifts in the darker of the skin sample pairs are towards yellow, not green.



That could be due to the lighting. I don't find them offensive. Are you sure you're color-normal?

Some skin reflectivity hue angles that I measured with a spectrophotometer in M1 mode:









Jim

--
https://blog.kasson.com

 

[lélé777004]

Here is what it's looks like converted with a matrix only color profile, then exported to sRGB:

View attachment d2e75f53ce6d459cae447c04b496b8dc.jpg

With color patches (from 11*11px average):

View attachment 04e3f9975636407ea40e0a24d3f4a98e.jpg

link to full resolution : https://ibb.co/JkYpfLy

Do that again and put in the Lab values.

Here it is: https://ibb.co/7tP6Tkq

The hue angle shifts in the darker of the skin sample pairs are towards yellow, not green.



That could be due to the lighting. I don't find them offensive. Are you sure you're color-normal?

Jim

Not affended at all.
My professional activities include training, with a vast majority of male worker, so this is something I'm precautious with.
(When you realize your trainee cannot read your graph... )

Yes, I'm am color normal. I was checked several times, including for my former work, it was mandatory for some of my activities.

In this particular example, it may be the lighting. Too many unknowns.
Like I said, this example is used as an illustration only.

What is puzzling me is that I noticed such shift in a vast majority of my shots. With direct sunlight, in the shadow, in extrior, in interior, with flash...

 

[JimKasson]

Here is what it's looks like converted with a matrix only color profile, then exported to sRGB:

View attachment d2e75f53ce6d459cae447c04b496b8dc.jpg

With color patches (from 11*11px average):

View attachment 04e3f9975636407ea40e0a24d3f4a98e.jpg

link to full resolution : https://ibb.co/JkYpfLy

Do that again and put in the Lab values.

Here it is: https://ibb.co/7tP6Tkq

The hue angle shifts in the darker of the skin sample pairs are towards yellow, not green.



That could be due to the lighting. I don't find them offensive. Are you sure you're color-normal?

Jim

Not affended at all.
My professional activities include training, with a vast majority of male worker, so this is something I'm precautious with.
(When you realize your trainee cannot read your graph... )

Yup.

Yes, I'm am color normal. I was checked several times, including for my former work, it was mandatory for some of my activities.

In this particular example, it may be the lighting. Too many unknowns.
Like I said, this example is used as an illustration only.

The reason I asked is that I would not consider the darker skin samples to be greenish, and it they don't measure greenish. In fact, they are within the range of hue angles that I measured with my i1Pro3:

What is puzzling me is that I noticed such shift in a vast majority of my shots. With direct sunlight, in the shadow, in extrior, in interior, with flash...

--

the last word the last word - Photography meets digital computer technology. Photography wins -- most of the time.

Photography meets digital computer technology. Photography wins -- most of the time.

blog.kasson.com