Being a beginner at Darktable, and apparently having some highlight clipping issues stemming from my camera’s very Adobe specific RAW implementation, I don’t know if what I’m seeing is due to my equipment not being fully supported or me not understanding Darktable enough:
So the Darktable portion on the above screenshot only has highlight reconstruction at 0.95 (per the previously linked thread), chromatic abberation and lens correction activated beyond the defaults. But notice how blotchy the blue LED lit areas are, how the lens glare shafts are missing (not added in post, I swear), and the weird miscolored green LED light blown areas. Also, there’s white areas near the copper heatsink which for some reason are black in Darktable…
Anyone familiar with Darktable willing to see if they can achieve smoothness and correct colors in the highlights using their regular workflows, before I start looking at tutorials, or if I need some special workflow to adapt to the irregularities of this weird camera?
I have made only one change from darktable’s standard opening procedures:
I switched filmic|options|color science from v6 (2022) to v5 (2021).
before I start looking at tutorials
It is always wise to study tutorials. There is quite a lot of information about the different color science modes. Use the forum’s search utility to follow the discussions.
Trevligt, tack för det! Att dra ned kontrasten i “look” hjälpte mycket också.
Very interesting! So, things don’t always get better then? Sigmoid also wasn’t able to do the highlights as smoothly as V6 and those are the two newest options, if I understand things correctly.
preserve chrominance
no means that the ratios between the RGB channels are not preserved. This will tend to saturate the shadows and desaturate the highlights, and can be helpful when there are out-of-gamut blues or reds.
CC module has a gamut compression slider …use it liberally and it can help… there are many ways to attack things in DT… LR is also not the default for a good image …in many areas often it handles things poorly as well so you just have to decide and target all the aspect of your image you want to develop and use the tools… which if you spend a bit of time you will master with some nice control… v6 of filmic has some gamut control. It is meant to be an improvement but for some images it is like a set of handcuffs so then v5 or perhaps sigmoid will work…with time you get to know… I almost always use v5 filmic. I like the controls and the the lack of the handcuffs… in the case of filmic for me at least v6 is not simply better because its a higher number it just is the next iteration that took a different approach…
With only exposure and sigmoid … more like your LR look… just used HLR as LCH .
You have a number of good replies here already. For my 2 cents worth if you want to chase the lightroom look filmic version 4 seems close. I did add the shadow and highlights module at default values to bring out the details in shadows more.
A custom version of sigmoid that lets one modify the primaries used for applying the per-channel curves. I hope to publish a version for testing some day.
The colorfulness control of the color calibration module does an admirable job here. Well done. Makes sense to “prime” the signal before applying the per-channel curves (sigmoid) and then restore chrominance in the midrange areas. This is coincidentally very similar to what happens in AgX.
A possible filmic v6 rendering. I set the white point manually, enabled reconstruction in filmic, and played with the params. L1082187.DNG.xmp (10.8 KB)
This is my usual sigmoid workflow, with the preserve hue slider in sigmoid down to zero and the blue ‘colorfulness’ slider in color calibration pulled back a bit - a good option for this sort of issue. Also tweaked the highlight reconstruction threshhold as needed with this camera.
I do find those blues hard though!
There are at least a couple of issues at play in this post Ie highlights and the gamut of the blue light. I think one thing that is key to convey to the OP if it is not already apparent, is that there are many ways to handle this and many looks or outcomes… Some people totally blow it out and some people look for gradients. I have not explored it but I recall Aurelien said that one of the main uses for the guided Laplacian module was to reconstruct these gradients so this too could be experimented with esp now that it is much faster.
In the case of this image if you set the image defaults to be none so that you just have as shot legacy wb and use the default HLR which is now inpaint you will get this…
Not far off a few of the edits… and still no tweaks to exposure or tone mapping etc
Switching back to default HLR and using CC with the gray tab and blue set to 1 and blended in average at about 50% opacity gives this… blending in normal gives a slightly different look and in case they can be tweaked by opacity…
I will upload the pictures later… I seem to be having issue with slow or cancelled uploads to Pixls the past week or more… Could be on my end… anyone seeing this too…
Linear, unbounded Rec2020 output (no filmic etc.), run through my gamut compressor experiment. This version was created using the modern chromatic adaptation method (white balance = camera reference, color calibration enabled, using the default gamut compression = 1).
And this one is just plain old white balance = as shot, without color calibration and of course gamut compression:
