One thing with the CC correction…It will show you an exposure value. I think that the values and visual representation are valid ie the delta E values so long as you set your exposure to that value…not sure how that will change your results…maybe you have been doing that all along anyway… It will be a good experiment and test of how using legacy wb compares to what you get with using CC with and without correction… If you do an onsite custom wb then that should help and then you can from there do legacy vs CC and also look at both with the CC channel mixer corrections…
I tried adjusting exposure and black level to values suggested in Normalization section with no effect on colors. I even tried to adjust tones with Tone curve module since I have absolute values for every patch. No effect as well.
Looks like Color calibration module goes ‘mad’ with my raw file.
I think more that they look right when you add that exposure or are at that exposure…rather than anything changing the calculated deltas from the process but I could be wrong
The suggesting algorithm of CC module works a bit weird in this part. It gives absolute values but does make mistakes. Sometimes when I set exposure and black level to suggested values my white patch goes 100lab instead of 96lab. The same story with the dark patches. It’s a tricky process and it doesn’t add a lot to the final result.
In my previous experiment with other light setups I found that it’s better adjust the tone curve to absolute values of the card (to official values). In case of Spydercheckr I’m using the first vertical column of the right part of the book (From 1E down to 6E).
Also I’m thinking of doing some exposure bracketing in camera next time with 0.5 or 0.75 EV steps. I feel like this may help me more.
Should it be 96… I thought xrite stuff was closer to 92 and datacolor was 96 for white??
Edit sorry I went back further… but I was confused… when you said this…
You want me to shoot a proper white card (I have a color checker passport with WC and an old Kodak grey card) and set the WB in camera.
But your using a spyder it seems???
Some cameras offer a custom white balance setting where you set the white balance to a white or neutral card for the lighting setup you are using. In the past this has worked well for me to get the nice white balance to reproduce artworks. It might be worth trying this method so that then DT can use WB as shot.
Ya that is what I meant above I guess it wasn’t clear…
I have both a spydercheckr 48 and color checker passport. I tried both version when doing reproduction and I like spyder48 a little bit more honestly.
Should it be 96… I thought xrite stuff was closer to 92 and datacolor was 96 for white??
And yes spyder48 has 96 white path. I’m talking about 1E patch. Here’s he proof
https://cdn-docs.av-iq.com/other/SpyderCheckr%20Color%20Data.pdf
I can confirm that once I have 1E patch 96 things get better. Also if I do tone curve for 1E…5E overall tones of the picture become more realistic (which is important for me). But 6E patch is another story. May be because of the type of a paint or so this patch acts strangely in different light situations. I avoid using it. The same with color checker passport if I’m not mistaken.
But your using a spyder it seems???
Yes. But the spyder doesn’t have a white card except small patches. Actually it has grey cards on the other side of its two papers. And it’s a pain to over turn them every time. So for WC I usually use my CC passport. This device is much smaller and more handy.
You’re right. Looks like I need to be more careful about WC and exposure.
1 Like
Yes, I got the idea. I’m gonna try this method first. Probably I’ll also try to use the Spydercheckr’s gray card as WC.
I’m sort of going through a similar exercise with my editing space. I recently got device to calibrate my screen. I thought it would be pretty strait forward and maybe it is… the calibrite (old x-rite) software is pretty basic but doesn’t support ambient light on the version that I bought where as displaycal does… But then you can go down the displaycal rabbit hole with all the potential settings and profiles that you can make. This coupled with my lack of experience for example with the software and if I should worry about ambient light levels and temperature has led to me creating quite a few profiles with no easy way to decide what is the best… at least yet… My best approach would be to control the light in the area where my computer is… At the moment this would be only two options… lights out and go with ambient… During the day I get some light but it would be fairly dark and pitch black at night… I could turn the lights on but they can’t be directed or dimmed so the old fluorescent tubes combined with the yellowish beige color of the room gives a light measured in and around my desk at around 3900K and maybe 250 for brightness. I am looking into replacing them with some 5000k lights or 6500K lights that might also ideally be dimmable …but maybe dimming causes issues as many lights that report to be “spectrum” lights aren’t dimmable… But that project will have to wait so I might do some more exploration and try to compare profiles correct for the ambient light vs those that don’t. The problem is that each profile or often wants you to tweak the rgb channels of the monitor so once you have a few profiles to compare you would need to change all those to match… I’m likely over thinking it and playing whack a mole… It was interesting for one profile I did enable the gamut mapping options in displaycal so that I could look at the difference from perceptual to relative and also there were some setting to try and do a CAT for scene to display… in any case switching the mode from relative to perceptual made a substantial change in the histogram output so the issue of having data go through the display profile which should not impact the output data is still there for certain profiles with LUT’s…
Yes.
Make your ambient light ad consistent as possible, run the calibration, load the ICC and be on your way.
1 Like
Ya I think its the sort of the yellowish environment that bugs me and provokes me… I have converted most of the lighting in my house to nice bright LED lights… the old holdovers are my shop and office in the basement… guess where my computer is 
… Time to enact KISS… keep it simple stupid
Yesterday I shot some more oil paintings with my halogen light setup. I decided to play with all these calibration things again. I was sitting in a relatively dark room. At the adjacent room I put some oil paintings. I put them in row so that I could see them and they were all lit with the light from the window. I re-calibrated my monitor to window light’s color temperature.
What I found is that
- Using white balance isn’t enough for the ‘correct’ (exact) color representation. The reds tend to go to yellow. Magenta doesn’t show up correctly.
- With DT the only way the color calibration module gives me a reasonable quality in my 3200K setup is when I set the WB to ‘As shot’ and then enable CAT and pick up WB again for CAT from 5E patch from my spydercheckr. This patch is almost pure grey on this card.
- Because I set WB in both modules I see these red warnings in both modules about WB being set up twice. Why is this an error at all? The names of saved presets invisible in this case. They’re replaced with error icons (just a small annoying drawback of interface)
- When I prepared spydercheckr card shot I tried to tweak the exposure more aggressively. And finally I managed to find a way to minimize the Delta E. It just a matter of luck I believe.
- Once I started to play with optimization presets I realize that it’s better to save all generated calibration matrices (presets) and give them proper names
- Then I started to compare art works with their shots by applying calibration matrices to images one by one to see the difference. I was just sitting and moved my head from the paintings to its images on the screen and back again. A really nice exercise for the neck )
- I still feel like I have some problems with the reds. It’s like I see more shades of red on the painting then on the screen.
My conclusion is that among all the color calibration 8 presets for each image you’re almost guaranteed to find 1 that will make shot looking 90% close to the real picture. Great news. Now a funny thing. It’s still pretty random process. You never know which preset would give you the right result. It’s like squeezing a balloon with water with your hand ))) You never know where the balloon appears.
So thanks to everyone. You gave me some useful tips.
P.S. At some point of my comparison exercise the weather changed. Light clouds disappeared from the sky and… the paintings started to look different again )))
I think that my problem with the reds is just because sRGB doesn’t have enough space for reds.
Now the question is. Will I have any gain in shades of red if I switch my camera to AdobeRGB?
Does anybody use AdobeRGB as the input color space when working with DT?
The colour space set in the camera only affects the jpg files produced by the camera, not the raw files.
For raw files, your input colour profile should be set to standard matrix (unless you have a better, custom profile in icc/icm format), working profile Rec2020.
Seems like an ICC input profile tuned for this temperature of light might be helpful?
1 Like
You get the error because the CAT applied by CC module is expecting D65 reference values provided by the program if you use any other then you get that error. If you are bothered and aware of why you get it you can disable the error notification in preferences.
You could try one other thing just for comparison and then see if it helps you could try to figure out how you could do something similar in DT… There is a free program that runs in Windows called Picture Window Pro 8… it has a few neat tools that I play with from time to time… It has a transform/module that it called multi color balance… You can use it to correct color casts or add them… What you can do with it when you have color checker images is use it in correction mode and select each one of your tonal patches so as you cancel all 6 you get a set of rgb curves that apply a global correction. Depending on the lighting this might give decent result or help… I am not sure if this would end up being something you could do in DT??
WB_Multi.pdf (1.2 MB)
Thanks for your help, guys. Pretty informative and clear.
Not sure about Picture Window Pro though I’m on linux.
If you want to send me or post a random shot I can try to see what it does … It was more just sharing an idea… The software is authored by one of the original lotus 123 developers. The UI looks a bit dated in many places but it has some clever tools including reference matchIng images like you can do in DaVinci. It has full masking and a nodal/branching type workflow so that you could do parallel processing and mix and merge the outputs from either branch… into the other… It was just a curious musing
Sure, it would be interesting. I can’t send you shots I did yesterday. They’re not mine and copyrighted. But I can send you a raw file of my daughter’s study I shot last week. It’s an oil study painted with red cadmium and shot under the same above mentioned light conditions. I still think I can’t archive perfect reds with it. Although I’m pretty close. In this file white balance was set in camera with WB CC passport. I’m gonna send it to you in a personal msg
1 Like