This topic discusses extracting ICC input profiles from Nikon CaptureNX software. I played with this a while ago.
Does anyone do this? Does it really make a big difference?
I tried it ages ago but never used it - I usually adjust my colours so much that slight profile differences are of zero consequence.
I don’t know if they could be more useful for accurate reproductive work, but you’d really want a custom profile there anyway.
1 Like
Yeah, for reproduction work I’d want to make a profile from the scene’s illumination. The Adobe DCP workflow attempts to approximate that using two profiles and an interpolation between the two based on a calculated color temperature, but I cannot imagine that gonkulator even coming close to a target-shot profile from the scene’s illuminator.
Not particularly related to Nikon .icm profiles, just wanted to say that… 
I also tried them once, but the difference between them and a well-constructed target-shot profile like what the softwares contain is negligible.
Is a well constructed target-shot profile one created with the colour calibration module and a colour target, or do you mean something else?
I was actually thinking of an ICC input profile, generated using the same target shot but with dcamprof or similar, but the colour calibration profile works exactly the same way.
1 Like
There are a number of ways to do it, but the common denominator is a shot of a color target under a particular light source. Most are done under mid-day sunlight, roughly corresponding to the D65 color temperature reference. The predominant target is the X-Rite (now, Calibrite) 24-patch ColorChecker, but there are others with various patch populations. What’s needed for the target is a reference file with its measured XYZ values for the software to use in comparing to the patches in the target shot.
For software, I also use dcamprof to make my .icc files. I gues dt has equivalent software built-in. X-Rite (Calibrite?) also has software, but it costs. The fellow who wrote the open-source command-line dcamprof program also has a GUI version with more capabilities, LumaRiver, also costs.
1 Like
I am not sure if a ICC profile from Nikon’s software would translate accurately to another program including DT. However, I created my own Canon R7 style in DT that would effectively achieve the same desired result. I created this by selecting a well exposed daylight picture with skin tones neutral tone areas and parts of the image which were strong colors of red, green, blue, yellow. It was actually a street scene from Japan with a couple of girls posing for a shot.
First I opened the camera’s JPG and took a snapshot. I then opened the RAW and used the snapshot as a comparison.
I set the exposure slider to match the JPG
I set color calibration to as shot in camera
I applied basic denoising and sharpening
I applied default local contrast
I increased contrast in the sigmoid module to 1.7
I applied shadow and highlights module at default values
I did some tweaks in color balance RGB module to adjust shadows and contrast
Finally I did some saturation tweaks of various color zones to lift the saturation of the reds in the image to better match the color look of the JPG.
This style was then saved and with one click I get a great starting point for my edits. I plan to do this for all of my cameras. BTW, a landscape style might need to lift the saturation of the blues and the greens where this Canon R7 style needed more warmth in the skin tones and hence the color zones adjustment.
1 Like
Darktable uses ArgyllCMS I beleive. I have a colour checker passport and have used it to profile. I just wondered how usefull the Nikon ICC profiles are in actual practice.
I have looked into this a little further today and my understanding is that using a Nikon ICC profile, extracted from CaptureNX, as an input colour profile is not the same as calibrating with a colourchecker via the colour calibration module.
The ICC profile achieves a specific look as a starting point, being the profile used to transform into the working colourspace, whereas calibrating via the colour calibration module achieves accurate colours compared to the colour checker.
I might try the ICC profiles again now that I have a better grasp of Darktable and I (sort of) have a better understanding of what I am doing.
When you do I would be interested in if the Nikon ICC profile is actually useful or accurate when applied to DT. What information or instructions are included in an ICC profile that can be translated correctly by DT?
I honestly don’t know the answer to this. I just know that it is described in the RawTherapee documentation:
https://rawpedia.rawtherapee.com/How_to_get_Nikon_ICM_profiles
I am not unhappy with the results from darktable, it is just an itch that I want to try and scratch again. I have heard that some users get better results when using an ICC profile from Nikon.
Last night I extracted ICC profiles for “Neutral”, “Standard”, “Vivid” & “Monochrome” Picture Controls and I plan to have a play with this at the weekend, provided life doesn’t take over. Do you have any suggestions as to meaningful tests? Anything I should look out for?
Well, one thing to sort through is the difference between “Neutral” and everything else. The primary purpose of these profiles is to provide the information to pull the camera data’s gamut and tone into the space capable of the rendition media. What the Neutral profile and the dt-internal data do is that, colorimetrically, that is, remaining faithful to the measured colors. The other profiles add on subsequent alterations to affect the “look”, departing from “colorimetric”.
I personally don’t like the “look” approach, clutters the original purpose of these profiles.
Thanks, I did assume the non neutral profiles affected the look, but I extracted them anyway as it was easy enough.
Is it your understanding/expectation that the Nikon ICC profile is better in any way and will allow for a better result (“better” being subjective off course), or is it simply a different starting point and the same final output will be acheivable, just with different levels of processing adjustments.
I prefer to start with the “colorimetric” image, that is, the image interpreted to the CIE 1931 color-matching reference. I get that from an ICC profile created from a target shot, or in some cases a spectral sensitivity dataset for the particular camera. My eyeball test tells me the Nikon Neutral profile is closest to that, compared with my target-shot profile, so I just use mine. Or, mostly, I just use the software-provided data, as that is just another target-shot source.
Really, I think the “look” profiles are only of use if you plan to use the out-of-camera JPEG. There, you’re making processing decisions before you take the shot, and they’re all you have to subjectively influence the rendition. If you’re raw processing, I think starting with the “neutral image” is better for bespoke tweaking.
Whatever icc profile you use then the process in the CC module is running some math to come up essentially with a set of channel mixer settings to make the color more accurate. So its specific to that icc profile. If you switch then in theory you would need to rerun the CC process as well…
Are these not really just profiles that will help to make a better match to a jpg that used one of those modes in the camera…so if you chose vivid then for the jpg created there in camera you would end up being closer out of the gate by using one of those profiles I would guess in theory but I think there is no relation to actual color accuracy but more of a look match …at least that is how I would see using them…
Agreed. So does that mean the choice of input colour profile is arbitrary, as the colour calibration module will achieve the same (or very similar) result?
I wondered the same. But the DT manual says that a colour matrix specific to the model of camera is applied, which makes me wonder if one from Nikon is a better match than the one from DT.
Just read on the Capture One support page that there is a Flat Nikon picture control which should retain a wide range of tones. ( I don’t use Capture One, just doing some reading). I will extract that one aswell.
Not .icc profiles, but .dcp from Adobe for Nikon worked really well for me in RT/ART. Quite often colors were better than with the standard camera matrix. I miss DCP support in DT. I tried converting .dcp to .icc a couple of times but it never worked for me.
I think the harder that the CC module has to work to correct the color then there are likely going to be more compromises …but again I am not talking about any look matching ( I think there is a chance that if a profile is biased to achieve a certain look then using CC might actually work to remove some of that)…if you have a color chart in the photo and then for a given icc profile that is used then the CC module is going to try to improve upon the color accuracy that results from that color profile…in the end its applying a matrix via the channel mixer to push color accuracy to be a bit better but having played with profiles in something like lumariver where you can tweak it patch by patch …at some point some patches get better and some worse so it becomes a bit of a trade off …for example I think you have profiles with perhaps a focus on skin tones as that will be the priority for the images and its not so important if there are some loss of accuracy elsewhere or some other version made with an intent…bottom line is I view the CC routine as a fine tuning of the whatever the current profile used delivers… I’m not sure how well it performs if you ask it to make large adjustments…Maybe someone else has spent more time with it and could comment…
I spent an hour or so on this tonight. I used the Nikon neutral ICC profile as the input colour profile, enabled unbreak input profile, set a low linear and had a little play with the gamma. Ended up settling on about 0.55.
For me, no matter how I processed, I couldn’t get anywhere as good as when using the standard input profile. The image looked generally washed out, particularly highlights. Very hard to tell if colours were better, they were slightly different but it’s all subjective.
As I said originally, just scratching an itch out of curiosity with this. This is the second time I have tried this and I come to the same conclusion. I will stick with the standard profile.
I noticed that the unbreak input profile module works in display referred, wonder if that’s not ideal for a scene referred workflow. I also have no real way of knowing if the Nikon ICC profile that I extracted from NX studio is actually a neutral profile, NX studio could have added anything to that recipe before spitting out the ICC profile. I just blindly followed the instructions from RawTherapee.
1 Like