Aah, that was it! After I cleaned the /opt/darktable directory and reinstalled, it worked. Thank you
Negadoctor is now merged in darktable master and planned for release in 3.0.2 (in 2 to 3 months from now).
@anon41087856 For some images, after I set the color film base, when I click on the DR picker, the image gets very dark, even if I redraw the square to leave the unexposed film part outside. Screenshots and raw file below:
IMG_7461.cr2 (11.3 MB)
This file is licensed Creative Commons, By-Attribution, Share-Alike
EDIT: I played with the picker by selecting smaller regions and it seems to work, except if I include parts of the cactus
A darkening Dmax, which values goes through the roof, usually arises when you have dust on the film, so the measure is completely wrong. You need to select a clean bright part to measure the Dmax. Otherwise, simply setup the Dmax manuallly to make it look good.
Hum, I think you’ve mentioned that on your video.
Even if we read tfm/see tfv, we still forget things. With the lack of the underlying knowledge, it’s as if we really don’t connect things and end up relegating some important details to oblivion.
Dust and stratches removal is on my to-do list, but it won’t be easy.
That would be great, but the way negative inversion works now is already awesome.
I wonder if the dust&scratches future code could also be used for slides, since they’re both subject to the same diseases (dust, scratches, but also fungus). Then maybe it would fit better inside imagedoctor, isn’t it?
Could dust and scratches just be detection, then use a mode from retouch to fill them all in?
Time for someone to submit fungal samples in the mail.
My research documented in Scanned image scratch removal with “ICE” shows that you need some pretty advanced inpainting techniques to get reasonable results. Even the dedicated code from vuescan does not do a good job here, only patch based algorithms lead to acceptable results so far IIRC. That’s probably not to far from the healing/cloning in retouch module, but the difference is that a manual source patch is selected there.
A sample image is available at the end of the thread above, and I’ll update the original post soon to include some more samples. Only dust and scratches, as this is detected by an infrared scan, unfortunately, fungus is not and requires manual work to recover.
Wow, coming back to pixls.us after two weeks, I had 54 posts and one 20 minute video to catch up in this thread… Thanks a lot, @anon41087856, for your work on this and on darktable in general.
In case there are some people following this thread who have missed it so far, I’d like to point out that you might want to consider supporting @anon41087856’s work: aurelienpierre's profile - Liberapay
Personally, I’m sitting on 3900 scanned negatives and still have a few thousand to scan. Currently, I’m using some scripts to batch process them, which is great to get a preview and to ensure that the scans are okay, but so far I have still been missing a way to manually editing single important pictures. I am really happy that this is changing and hope to find time to play around with it soon!
Well, aren’t we all lucky you’ve got such a useful sense of humor? I’m pleased as punch with it and have been using it exclusively to process my negatives. I have a feeling it’s only gonna get better and am over the moon that I now have a reliable workflow for my film post processing. Nicely done!
Happy customers everywhere
Is there a Windows build where I can try out the new neg module? I moved away from darktable because of problems with the old negative module, so this is of huge interest to me.
Unfortunatelly I think negadoctor isn’t part of 3.0.1 release so you need to wait for 3.0.2 or play around with unofficial build.
Yeah, I was wondering where dev builds for Windows were these days, if there are any.
I have 3.0.2 (Ubuntu 18.04, darktable from OBS), but I think I don’t see the module.
Is this correct or I’m missing something?
EDIT: sorry for the spam… I found out the answer my self with just a simple search in the forum…
There is a thread with windows (dev) builds around here, around 20 days old. It contains negadoctor (have been playing around with it last few days… But got distracted by filmic v4 :))
Hoping this may be of interest to some. Comparing inversion methods in darktable using invert, tone curves and rgb curves.
Image is a .png [not a negative], input profile sRGB.
Working space is linear rec 2020.
Output profile is sRGB.
Blend mode is normal for all.
Original.
Inversion methods below in order of best to worst:
Invert Module.
Tone Curve Module (preserve colors: none).
RGB Curves Module (preserve colors: none).
Tone Curve Module (preserve colors: average RGB - default).
RGB Curves Module (preserve colors: luminance - default).
Comparing tone curve to rgb curves, we see tone curve doing a nicer job with luminance, and rgb curves producing brighter colour. When preserve colors are set to defaults, the inversion effect is poor (hue remains mostly the same, black background remains black), but when preserve colors is set to none the effect is much better. It is somewhat immaterial though, as invert module is best.