Colour calibration process Capture One - x-rite colour checker

[justjay]

I use a x-rite colour checker card and I have recently switched to using Captureone.

I noticed some minor color (probably tint?) differences when I captured the photo through a glass panel and without a glass panel (above the photo to flatten it). I used the x-rite color checker during both instances and created 2 different ICC profiles (with and without glass) and ensured each image was using the correct profile. Can someone explain why I am seeing colour differences when each image is supposed to have been colour corrected? Is there a better process I should be using?

My process for creating ICC profile in CaptureOne is as follows (I follow instructions on youtube):

1. Capture color card (Color checker passport)
2. Change ICC profile to "No colour correct" and change curve to "Linear Response"
3. Set white balance using the square second from bottom right (above white square)
4. Export as tiff ensuring camera profile is selected
5. Use colorchecker camera calibration software to create ICC-TIFF profile
6. Restart Capture one and apply ICC profile to image
7. I then copy and paste the setting from one image (with glass) to another (without glass). Usually this is tone curve and some highlight/white/ black adjustments

I use 2 xAD200 Pro in softbox on the left and right side of camera (Nikon D750). I thought the reason for using a colour card is the get consistent colour when the lighting/camera/ lens changes but in this case everything was constant, with the exception of using a glass panel and not using a glass panel. Appreciate any thoughts or links to reading material.

[Manfred M]

Most glass has a tint (unless you are using high grade optical glass), so no surprise there. Most commonly used glass is float glass, which tends to have a green tint due to the iron oxide it contains.

The ICC profile that you are creating is for the camera sensor and should give you a slightly better colour response curve than with the built-in default with Capture One (C!). It has nothing to do with colour correction.

I have built a profile one for daylight with my ColorChecker Passport that I use with my D810 with Adobe Camera Raw (ACR). Initially I also built one with flash, but there was not noticeable difference so I don't bother with a separate flash one any more.

I did notice it made a difference when I had the D800 (the reds were a bit strong with the custom profile), but there is virtually no detectable difference with the D810. I built a profile for Capture One many years ago, but never bothered when I changed cameras, I use C1 when doing tethered capture, but find ACR gives me a better workflow than C1 does, so I rarely use it as my raw convertor.

There should be some small colour differences as my Godox Witstro AD-360 has a ±200K tolerance on the base 5600K colour temperature, so in theory the light output can be between 5400K and 5800K, so that could explain some of the colour temperature difference. In practice I tend to see this more when I vary the power level and the flash is quite consistent when I shoot at the same output setting. I believe the AD200 has a similar colour temperature spec. Your soft box can influence colour temperature as well, depending on age, cleanliness, which baffles you use, etc.

When I pull a white balance, I tend to use the middle gray swatch, not the darker or lighter ones. The colours can change on the card as it ages or gets dirty too. When I shoot in the studio, I really don't get a noticeable change between shots, so I can't even guess why you are.

[dje]

I use a x-rite colour checker card and I have recently switched to using Captureone.

I noticed some minor color (probably tint?) differences when I captured the photo through a glass panel and without a glass panel (above the photo to flatten it). I used the x-rite color checker during both instances and created 2 different ICC profiles (with and without glass) and ensured each image was using the correct profile. Can someone explain why I am seeing colour differences when each image is supposed to have been colour corrected? Is there a better process I should be using?

My process for creating ICC profile in CaptureOne is as follows (I follow instructions on youtube):

1. Capture color card (Color checker passport)
2. Change ICC profile to "No colour correct" and change curve to "Linear Response"
3. Set white balance using the square second from bottom right (above white square)
4. Export as tiff ensuring camera profile is selected
5. Use colorchecker camera calibration software to create ICC-TIFF profile
6. Restart Capture one and apply ICC profile to image
7. I then copy and paste the setting from one image (with glass) to another (without glass). Usually this is tone curve and some highlight/white/ black adjustments

I use 2 xAD200 Pro in softbox on the left and right side of camera (Nikon D750). I thought the reason for using a colour card is the get consistent colour when the lighting/camera/ lens changes but in this case everything was constant, with the exception of using a glass panel and not using a glass panel. Appreciate any thoughts or links to reading material.

My initial guess is that it's a white balance issue. Have a look at the two color checker images you took (and white balanced with the color checker) and take note of the Kelvin and Tint values for each. They are probably a bit different. Now have a look at your comparison images and manually set the Kevin and Tint values to match those from the color checker images. My assumption here is that the glass may be changing the color temp and tint of the illuminant.

Another thing to bear in mind is that camera color profiles are not precise and represent a best compromise in color accuracy (DeltaE) across the color patches. Also check how the two profiles you created compare when you apply them to the color patch images.

Dave

[justjay]

Most glass has a tint (unless you are using high grade optical glass), so no surprise there. Most commonly used glass is float glass, which tends to have a green tint due to the iron oxide it contains.

The ICC profile that you are creating is for the camera sensor and should give you a slightly better colour response curve than with the built-in default with Capture One (C!). It has nothing to do with colour correction.

I have built a profile one for daylight with my ColorChecker Passport that I use with my D810 with Adobe Camera Raw (ACR). Initially I also built one with flash, but there was not noticeable difference so I don't bother with a separate flash one any more.

I did notice it made a difference when I had the D800 (the reds were a bit strong with the custom profile), but there is virtually no detectable difference with the D810. I built a profile for Capture One many years ago, but never bothered when I changed cameras, I use C1 when doing tethered capture, but find ACR gives me a better workflow than C1 does, so I rarely use it as my raw convertor.

There should be some small colour differences as my Godox Witstro AD-360 has a ±200K tolerance on the base 5600K colour temperature, so in theory the light output can be between 5400K and 5800K, so that could explain some of the colour temperature difference. In practice I tend to see this more when I vary the power level and the flash is quite consistent when I shoot at the same output setting. I believe the AD200 has a similar colour temperature spec. Your soft box can influence colour temperature as well, depending on age, cleanliness, which baffles you use, etc.

When I pull a white balance, I tend to use the middle gray swatch, not the darker or lighter ones. The colours can change on the card as it ages or gets dirty too. When I shoot in the studio, I really don't get a noticeable change between shots, so I can't even guess why you are.

Thank you for your reply Manfred. Yes, you are correct that the glass I was using has a green tint which I assumed would alter the colours slightly. However, I thought that using a colour card would allow me to "reset" colours to point that they would be similiarish to a capture without the glass. I also do not understand the second para where you say that the ICC profile is for the camera sensor, not for colour correction. I always thought it did both.

The AD200 Pro is fairly consistent with its colour temperature. I am not seeing differences in temperature shifts when I do multiple shots. The only change in colour I was talking about was the differences in colour (AFTER I have created and applied ICC profiles for each) when shooting through glass and shooting without glass.

You also mentioned that when you did an ICC profile for the D800, the reds were a bit strong. How do you know they were strong? Were you comparing the RGB values against a different ICC profile? Many years ago, I shot some red coloured cookies. Think red velvet cake kinda red. I profiled my colour card as I normally do (this was using lightroom back then), and the reds on my cookies looked horrendous. It was overly saturated and the hue was completely off. I have a decent monitor (99% sRGB) and it is colour calibrated. I did some research back then and found that cameras have an issue with red blowouts. I didn't really know how to get an accurate red back then (still don't) so I just adjusted the reds until they looked good on my screen. The cookies were then printed on packaging on the cookie on the box looked nothing like the cookie inside lol (the client knew I had zero experience with packaging work). I don't do much product photography work anymore but I am curious about your process for calibrating reds...

[justjay]

My initial guess is that it's a white balance issue. Have a look at the two color checker images you took (and white balanced with the color checker) and take note of the Kelvin and Tint values for each. They are probably a bit different. Now have a look at your comparison images and manually set the Kevin and Tint values to match those from the color checker images. My assumption here is that the glass may be changing the color temp and tint of the illuminant.

Another thing to bear in mind is that camera color profiles are not precise and represent a best compromise in color accuracy (DeltaE) across the color patches. Also check how the two profiles you created compare when you apply them to the color patch images.

Dave

Thank you. I'll give that a go. Do you adjust the white balance before you create the ICC profile or after?

[dje]

Thank you. I'll give that a go. Do you adjust the white balance before you create the ICC profile or after?

You set the white balance on the color checker images before you create the profile with them. What you outlined initially in your first post seems correct in terms of creating the profile. What I'm suggesting is to take note of those wb settings in Capture One and then apply them manually to the sample images you are comparing (with the new profiles in place or the standard profile for that matter). My theory is that when comparing the sample images, the white balance needs to be correct for each image and this is probably more important than the custom profiles.

Dave

[xpatUSA]

I use a x-rite colour checker card and I have recently switched to using Captureone.

I noticed some minor color (probably tint?) differences when I captured the photo through a glass panel and without a glass panel (above the photo to flatten it).
.
.
6. Restart Capture one and apply ICC profile to image
.
.

Jay, by "apply" do you mean "assign" or do you mean "convert using your new profile"?

May we assume that you are observing these differences on you monitor with exactly the same room lighting and monitor settings and not using a proofing function? And that the profile is of type 'mntr'?

If so we can assume that the differences are in the PCS' XYZ (or Lab) values and I agree that the difference is perhaps a poor white reference transformation. Have you looked at each profile with a Profile Inspector? Are you able to post the .icc files?

Would appreciate any thoughts or links to reading material.

https://ninedegreesbelow.com/photography/xyz-rgb.html

Heavy going but Elle Stone makes her stuff very clear.

None of CMS is simple and ICC's writings are virtually unintelligible.

Here's another:

https://photosauce.net/blog/post/mak...lors-carefully

Elle gets several mentions there, too.

[Manfred M]

Thank you for your reply Manfred. Yes, you are correct that the glass I was using has a green tint which I assumed would alter the colours slightly. However, I thought that using a colour card would allow me to "reset" colours to point that they would be similiarish to a capture without the glass. I also do not understand the second para where you say that the ICC profile is for the camera sensor, not for colour correction. I always thought it did both.

The AD200 Pro is fairly consistent with its colour temperature. I am not seeing differences in temperature shifts when I do multiple shots. The only change in colour I was talking about was the differences in colour (AFTER I have created and applied ICC profiles for each) when shooting through glass and shooting without glass.

You also mentioned that when you did an ICC profile for the D800, the reds were a bit strong. How do you know they were strong? Were you comparing the RGB values against a different ICC profile? Many years ago, I shot some red coloured cookies. Think red velvet cake kinda red. I profiled my colour card as I normally do (this was using lightroom back then), and the reds on my cookies looked horrendous. It was overly saturated and the hue was completely off. I have a decent monitor (99% sRGB) and it is colour calibrated. I did some research back then and found that cameras have an issue with red blowouts. I didn't really know how to get an accurate red back then (still don't) so I just adjusted the reds until they looked good on my screen. The cookies were then printed on packaging on the cookie on the box looked nothing like the cookie inside lol (the client knew I had zero experience with packaging work). I don't do much product photography work anymore but I am curious about your process for calibrating reds...

Your glass acts as a filter and changes the light that the camera sensor records. To some extent what you are saying is correct and in most circumstances what you suggest will work reasonably well. The problem is that we don't understand the underlying algorithms and how they are impacted by abnormal spikes in the light. A green filter will reduce the impact of magenta in the image. The ColorChecker Passport uses 24 discrete colour samples to create a profile, but if there is something strange with the light, the resulting profile could be impacted.

I tried to create a profile for fluorescent lights some time ago, hoping that I would be able to get stronger images, but the results were very poor. The algorithms took the spectral spikes into account but the results caused other issues. The green spike was particularly problematic as were the lack of reds in cool white lamps.

How did I know that the reds improved with the custom ICC profile? I could tell just by looking at the image with / without using the custom profile. My colour printing comes from the wet darkroom days were we could see the issues without using complex tools and worked on exposure and colours through test prints. It was just obvious that the reds were more saturated with the custom profile. Reds are still problematic with some saturation levels. If you look at the CIELAB chromacity diagram, you will notice that the human visual range can pick out colours that are out of gamut for both sRGB and Adobe RGB. That is effectively the issue that you are seeing. My 99% Adobe RGB compliant screen has trouble accurately producing some of the bright green hues, whereas my 100% compliant screen handles them quite well and the difference is actually quite noticeable.

[justjay]

I did what you suggested and noticed that when I was comparing images (with and without glass), I only copied the kelvin setting, not the tint.

There is a large variation in the WB settings, particularly for tint.

no glass - 5672, 0.8
with tinted glass - 6009,3.2

Manfred's explanation of the green glass filter changing some of the tones makes more sense to me now. Thank you both for your help.

[justjay]

Jay, by "apply" do you mean "assign" or do you mean "convert using your new profile"?

May we assume that you are observing these differences on you monitor with exactly the same room lighting and monitor settings and not using a proofing function? And that the profile is of type 'mntr'?

If so we can assume that the differences are in the PCS' XYZ (or Lab) values and I agree that the difference is perhaps a poor white reference transformation. Have you looked at each profile with a Profile Inspector? Are you able to post the .icc files?



https://ninedegreesbelow.com/photography/xyz-rgb.html

Heavy going but Elle Stone makes her stuff very clear.

None of CMS is simple and ICC's writings are virtually unintelligible.

Here's another:

https://photosauce.net/blog/post/mak...lors-carefully

Elle gets several mentions there, too.

I changed the ICC profile (under Base Characteristics) in Capture One. I didn't convert to any particular colour space if thats what you mean.

I cant upload ICM files on imgur but I have posted 2 tiff files of the colour card I captured (with and without glass) that has the ICC profile applied. I initially created 2 separate ICM files off these images and then re-applied the corresponding ICC profile in C1.

https://imgur.com/a/sq4mcyZ

Thanks for the links. I ll have a read later. =)

[Manfred M]

Jay - I guess I should have asked at the beginning, why are you trying to do this? Are there issues with the profiles that come with C1 that you need to correct for?

[xpatUSA]

I changed the ICC profile (under Base Characteristics) in Capture One. I didn't convert to any particular colour space if thats what you mean.

I cant upload ICM files on imgur but I have posted 2 tiff files of the colour card I captured (with and without glass) that has the ICC profile applied. I initially created 2 separate ICM files off these images and then re-applied the corresponding ICC profile in C1.

Ii looks like "the corresponding ICC profile" was identical for each image, see below ..

https://imgur.com/a/sq4mcyZ

Thanks for the links. I ll have a read later. =)

Thanks for the two images. Each one has a completely identical Adobe RGB (1998) profile with XYX PCS including identical black and white references and simple, identical, 2.2 gamma tone reference curves.

With identical profiles I would expect the RGB numbers to be different betwixt the two but, for example, the green patch numbers are:

no glass: 163, 239, 109. with glass: 162, 239,108.

Makes no sense to me at all, sorry. Either the RGB numbers should be different or the profiles should be different.

Summat's up or summat!

[Manfred M]

Ii looks like "the corresponding ICC profile" was identical for each image, see below ..



Thanks for the two images. Each one has a completely identical Adobe RGB (1998) profile with XYX PCS including identical black and white references and simple, identical, 2.2 gamma tone reference curves.

With identical profiles I would expect the RGB numbers to be different betwixt the two but, for example, the green patch numbers are:

no glass: 163, 239, 109. with glass: 162, 239,108.

Makes no sense to me at all, sorry. Either the RGB numbers should be different or the profiles should be different.

Summat's up or summat!

Ted - the ICC profile is all about correcting the camera's colour response and is related to the spectral curve.. When I did my first profile, I made one for daylight (shade) and one for flash. One would have had a colour temperature of around 9000K and the other at around 5600K, yet the camera profiles I produced were identical. When I replaced the D800 with the D810, I only went with one ICC profile, based on my past experience.

When I tried to come up with one for different types of fluorescent lights, they were all over the place, which I suspect was related to the emissivity spikes of the phosphor mixes used by the various tube manufacturers. I had planned to try some mixed lighting profiles (two different light sources) as the x-Rite software supports this, but never got around to it.

My suspicion is that the green / magenta bias introduced by the float glass isn't strong enough to impact the profiling. When I was playing around with profiling my camera I also tried lenses from difference lens manufacturers (Nikon, Tokina and Sigma) to see it there were any biases there and found no difference either.

[DanK]

I only copied the kelvin setting, not the tint.

That could be a substantial source of the problem. You need values on two axes to locate a color in a color wheel.

[xpatUSA]

... You need values on two axes to locate a color in a color wheel.

Actually, in a color wheel, a color is located by angle and distance, i.e. is a vector. Your dual axes, on the other hand, exist in chromaticity diagrams such as CIELUV, CIELAB. etc.

[xpatUSA]

Ted - the ICC profile is all about correcting the camera's colour response and is related to the spectral curve.. When I did my first profile, I made one for daylight (shade) and one for flash. One would have had a colour temperature of around 9000K and the other at around 5600K, yet the camera profiles I produced were identical.

Manfred, it sounds like we are taking at cross-purposes. By "camera profiles" were you talking about Adobe DCP files of type .dcp. Reason I ask is that can not see how two identical profiles of type .icc/.icm can correct two difference color casts.

As we should know, ICC profiles have an input side and a output side. While it is reasonable that the output sides would be the same for given output device, it is much less reasonable that the input sides would be the same for two different color casts or, in your case, two widely-differing Kelvin lighting values ...

[DanK]

But can’t the end point in any vector be identified by the aliens on two orthogonal coordinates?

That’s what the blue-yellow and magenta-green adjustments represent, no?


Sent from my iPad using Tapatalk

[Manfred M]

Manfred, it sounds like we are taking at cross-purposes. By "camera profiles" were you talking about Adobe DCP files of type .dcp. Reason I ask is that can not see how two identical profiles of type .icc/.icm can correct two difference color casts.

And, as we should know, ICC profiles have an input side and a output side. While it is reasonable that the output sides would be the same for given output device, it is less reasonable that the input sides would be the same for two different color casts or, in your case, two widely-differing Kelvin lighting values ...

Sorry Ted, you are right. I have been doing too much printing lately and have ICC profiles in my head. That is different than the DCP profile.

[xpatUSA]

But can’t the end point in any vector be identified by the aliens on two orthogonal coordinates?

Obviously, but why confuse less knowledgeable folks by mixing the two together in a single sentence?

That’s what the blue-yellow and magenta-green adjustments represent, no?

Adjustments where and to what? I'm thinking not ...



No orthogonal blue-yellow and magenta-green axes to be seen here either:

https://www.handprint.com/HP/WCL/labwheel.html
.

[justjay]

Jay - I guess I should have asked at the beginning, why are you trying to do this? Are there issues with the profiles that come with C1 that you need to correct for?

Mostly because I took 2 sample photos of a clients photo album. One with a glass panel to flatten a photo, and one without a glass panel (the photo rolls up slightly) to show her the difference. After I processed them and assigned the corresponding ICC profile, the photos had a obvious colour difference. One was more green and the other was more magenta. I didn't know how to explain the differences to her to I tried to colour match them instead -___-"

I couldn't match them closely so I gave up and processed the whole album without colour edits apart from the ICC profile. She didnt ask questions.