Hello,
can someone point a download source to a input profile for Darktable for halogen quartz lamp (not just a tungsten for 2900K regular bulbs but for halogen 3200K)?
Thanks in advance.
An input profile describes the colour space used to encode an image. It could be a general-purpose, device independent profile like sRGB or Adobe RGB, or one specific to the device used to digitise an image (a camera or a scanner).
The illuminant you would approximate either via setting the white balance module (for example, by sampling a grey card), or by setting white balance to camera reference and picking a grey spot in color calibration, or by shooting a colour reference card (e.g. ColorChecker) and using that to create a more detailed correction. See darktable 4.4 user manual - color calibration.
1 Like
you need to take a shot of a colortarget to generate one with e.g. GitHub - Beep6581/dcamprof: DCamProf is a free and open-source command line tool for making camera profiles, and performing tasks related to camera profiles and profiling. if you want to use it in the colorin module.
So an arbitrary input profile doesnāt give improvements over color calibration.
Even if you have access to a supported color target then using this in color calibration is a much simpler job than profiling the camera for a specific illumination.
Sorry. Wrong question.
Let me clarify my question. Iām now experimenting with 2 halogen 3200K photo bulbs for reproduction purposes (oil painting). I use cross polarization setup (homemade) based on these 2 1kW bulbs. I also use spydercheckr and colorchecker to fine tune the colors in DT 4.4.2 using Color calibration tool. The results are promising.
For WB I tried standard and tungsten (build in DT). Tungsten gives me better results. I wonder if the result would be even better if I had a proper WB 3200K profile for my polarized 3200K light.
Sorry guys. My mistake. Iām talking about WB profile for halogen 3200K.
What Iām doing now step by step.
- Shoot spydercheckr48 under polarized light and put the file into DT
- I set up WB module for tungsten (or Camera reference, or others) mode
- In color calibration tool I either select CAT space or just ānone (bypass)ā
- Then I set up calibration tool and create save configuration (by choosing one of the āOptimize forā
The problem is in WB. It looks like polarized light under 3200K gives me some strange results. I have some problems with reds without proper WB.
I tried to set up WB for CAT adaptation algorithm using its color picker while setting up generic WB modules to different modes like As shoot, tungsten, camera reference) etc.
My question is more about Color calibration tool and the correct workflow for it rather than profiling I think. I renamed the thread.
I think the process is supposed to be used with white balance = camera reference.
See this video from the moduleās author:
1 Like
Yes, Iām aware that many recommend setting WB module to Camera reference and after that set WB for CAT adaptation module choosing a particular grey area by picker. However my reds tend to be pink with this method probably because of polarization.
After hours of playing with all these algorithms I found that if I set main WB module and ātungstenā and then select WB area for CAT module, I will have better reds (despite the fact that DT gives me error message in both WB and Color calibration modules). So better reds without loosing shiting other colors.
Now Iām trying to understand what Iām doing wrong.
Probably I need to set up WB in camera with my white card and choose āAs shootā or probably I need to create a kind of WB profile for my weird light?
If you have a color checker could you since your lighting is fixed and controlled could you just use the grey card and do a custom WB in your camera assuming it lets you ā¦ then just let DT use that as shot in as a starting point??
Yes, this might sound as a solution. Iāll try this.
But isnāt the same setting camera WB and choosing white patch on color checker. I mean choosing āWB from Image areaā?
Iām a bit lost with these 2 WB options in main generic WB module and CAT module.
Are you shooting the color target with the polarizers on? If so, can you share a color checker raw file?
Sure. This is the RAW file with polarizers on.
Iāve just attached the jpg from camera. The reds in it tend to yellow/green. But itās Ok.
Of course I do not have any problems doing calibration with this checker in normal light conditions
Well I just suggested this way as it would bake the wb into the exif data that DT would read. if you do it post that would be a correction on whatever wb you end up with when you shoot. I think actually you wb off the most neutral patch which might not actually be the white patchā¦ I think one of the grey patches is suggestedā¦ I was actually thinking even of a grey cardā¦
And I would start with legacy wb just to see how that goesā¦. Later if you want to use the CC module you could try. But then you are taking a chance that you have good D65 reference values as CC illuminant from those values with the CATā¦. I would keep it simple at firstā¦ You can still use the color checker correctionā¦. Its just a channel mixer matrix so you could make a further correction with it ā¦ just do it in bypass or as pipeline mode I think would be okayā¦
To rephrase you to my newbie language.
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.
In DT I have to use the main WB module choosing āAs shootā option. Right?
Later if I wanna fine tune the colors more I have to enable āColor calibration moduleā and set ānone (bypass)ā option for CAT optimization. And then play with all these pre-sets like Sky, Skin, Average delta E etc. to get a proper result. Right?
Cool, thanks. The other question is, have the camera or light source moved from the color checker shot?
My understanding is that the Color Calibration should do what it needs to to shift the colors in the raw file to match the actual color checker colors.
No. Camera and lights do not move.
The color calibration works. it does shift colors. But I was expecting a better result. The reds get strong pink tint. The problem is probably in initial WB.
The pink may come from filmic, if itās enabled.
2 Likes
Just my suggestion but you can do an intermediate step with no CATā¦. So in bypass as I understand it CC is a pure channel mixer. If you use the same as D50 pipeline then it will not apply a CAT but it will do the channel mixing or other tabs in the selected Cat colorspaceā¦ what all that really means is not something I could really comment on but as I see it you can just try to get a correct Wb on scene with your card. Then if you want based on that then just editā¦ or try to make it even better by doing the color calibration color checker in bypassā¦ so that you just get that feature and donāt have to use the CAT workflow and once you have tried all that you could use CC as intended with your wb set to reference and set the cc module to as shot and then try to do a correction with the internal color checker again from that workflowā¦ I just suggested the first way as a reference for doing it this final way with the d65 reference values and using the CAT in case these introduce a changeā¦which could be a good change or maybe not good or maybe in the end you would be the sameā¦. Basically in his video AP demonstrates the use of the correction tool in CC module. He then says if you want to use it as a general profile then set the mode to as shot and save a preset ā¦when you do this you get a set of channel mixer adjustments that you can apply to your current input profile and as shot wbā¦ I guess you just have to experiment and see what works bestā¦ good luck
No. I disabled Filmic and Highlight reconstruction modules.
I have
- Input color profile: standard color matrix with linear rec2020 (I guess these are default)
- Exposure set to get E1 path to 95-96%.
- WB set to either As shoot, Camera reference or Tungsten (works better)
- Color calibration is set to CAT or bypass and I played wit misc optimization methods
Thank you. A good starting point for me. I will try all these things over the next few days and get back here with an update.
1 Like