Hi there! I’m starting out with darktable for the first time. I am using the tool to generate an EXR file from my Sony RX100 M6 that is as close to the captured scene’s tristimulous as possible for scientific use. Said differently, I am not interested in producing a good looking image.
My specific use is comparing computer monitors’ presentation in different output modes and configurations. I also have an LS-100 luminance metre that I use for my actual measurements, but for the article I am writing, it is helpful to also have some photos that have minimal processing applied to them.
I believe I get good results if I disable all default modules except for the “lens correction” and “white balance” modules. Are there other modules that I should have active for this purpose?
Ideally I would like to use D65 that matches both the monitor that I’m photographing and the files that I am saving to, so I am currently using 6504 K with a tint of 1.0. Maybe there’s a better setting I can use? I see if I set the white balance to “camera reference”, it uses 6502 K with a tint of 1.0… I don’t understand the 2 K difference, but this is negligible and seems to suggest my approach is good(?)
I believe that camera RAW files do not have truly linear data, so I presume that some sort of nonlinear transformation is needed to produce the linear RGB saved in the EXR file… Is that being taken care of automatically or as a part of the white balance module?
The article will describe the behaviour of HDR mode on monitors and how different monitors have different factory calibrations.
One example would be comparison photos of SDR mode and HDR mode of a monitor: the HDR mode uses a different transfer function and calibration and also has a black level that has been raised due to the LCD backlight being increased to a higher intensity when HDR mode is enabled.
Another example would be comparison photos of Windows and macOS in HDR mode, demonstrating that they both produce the same appearance when providing an HDR10 signal to an external monitor.
Another will be a comparison of powered-off monitors and how glare affects different monitors.
The monitor will be showing some test gradients as well as a photo that includes dark and saturated colours (a photo of a building with neon lights). But this is fairly arbitrary.
If the monitor is D65 calibrated, the most accurate white balance would likely be achieved by following the instructions under “Caveats” here: darktable user manual - color calibration
I have never seen raw data described as anything other than linear, so not sure where you got that idea from. In any case, as far as darktable is concerned it’s linear and will be treated as such.
Regarding the linearity of RAW data: I saw a slide on a deck from what I thought was a presentation linked from color.org, but I can’t find it. So maybe I misinterpreted what was on the slide, since I was just glancing through it at the time…
Presumably the publication will be using sRGB, so if you’re capturing an image of an HDR monitor and need to map it down to SDR, you need to ensure that tone mapping is on luminance and doesn’t cause a hue shift. Same with gamut mapping.
Almost correct when being close to full photosite maximum level some sensors surely will not be linear any more. There are even exif tags for this but nothing to be used by algorithms.
when being close to full photosite maximum level some sensors surely will not be linear any more. There are even exif tags for this but nothing to be used by algorithms.