Using darktable to adjust scans of colour reversal slides

It has just suddenly occurred to me that all the reading and watching on darktable that I have been doing has almost always (by implication) been about processing raw images from modern digital cameras. But, for my circumstances, the processing of TIFFs, from the scanning of colour reversal slides , is a very large part of my ‘work’. These scans have been made over a long period using various software and hardware configurations in an environment of knowledge, understanding and skill which has varied from ‘non-existent’ to ‘reasonable’ over that period. However I am still far from fully appreciating how to produce produce the best scans possible using, say, Vuescan or Nikon Scan 4 with either Nikon CoolScan or Epson V700.

As a consequence many of the earlier scans will need to be replaced, so much of my work plans involve handling raw images which have not come from a digital camera. How do I need to adjust my darktable preferences, or key module settings, to reflect this image portfolio? Or, are any changes required?

I have only worked with scanned negatives, so I can’t talk from experience with colour reversal, but I would imagine that the only thing you should do with darktable is turn off any of the default workflows and make sure you start with “none” in preferences > processing > auto apply pixel workflow defaults. You don’t really want to apply filmic or a base curve to something that has already been developed.
Other than that, I would imagine most of the care needs to be taken with the actual slide scanning.

If you want to handle scanned negatives in darktable, you should use the negadoctor module. Instructions are here:

I worked with a large number of copies of color negatives and diapositives. Part of them created with a scanner, another part created with a DSLR. Here a supplement to the reply of europlatus for working with diapositives :
Most important is to take care of “correct” scanning parameters to avoid strong color shifts. How to do this, strongly depends on the light source your scanner uses. I can not give a general recommendation because the devices and the associated software vary widely. I prefer copies created with a DSLR, as light source a bi-color LED flex-panel, color temperature and intensity adjustable. I started with a test series of about 50 shoots to find the lighting parameters and DSLR white balance settings to get almost neutral colors (for different film materials).
Having optimized these parameters I mainly use the modules

  • White Balance
  • Color Calibration
  • Exposure
  • Color Balance

to get excellent results. Again: it’s highly relevant to start with “good” copies (means low color shifts). With scans I got from another person, taken with a professional scanner but having a strong magenta cast, I was not able to get satisfactory results.

1 Like

Thanks for this - it goes directly to two of my major concerns about my workflow. As my understanding of dt slowly increases, I had begun to think that I should be setting ‘none’ in those relevant preferences as you have stated, so I’m encouraged to see your confirmation. The bigger concern though is in the second point you make, about getting good scans. Many writers on the internet make it too - as does the responder ‘pehar’ in a following post - but trying to find examples of Vuescan or Nikon Scan settings on the internet is like prospecting for hen’s teeth. I even have Sascha Steinhoff’s two books on the subject. They succeed in creating a forest of settings and their descriptions which, in aggregate, result in one ‘not being able to see the wood because of the trees’.

Yes, I strongly concur - n a theoretical sense. The difficulty comes in discerning what these ‘“correct” scanning parameters’ are, in which scanning engine, that will produce the required ‘“good” copies’. I am wondering if the settings advised on the ColorPerfect web-site, to produce ‘linear scans’ are relevant ?

I was confronted with the same questions when I started to think about digitizing all my diapositives and negatives. My main problem with professional scanners was, that most manufacturers do not provide any information about the light sources they use. They just say : “we use LED’s”. And every software package for every scanner differs… And most software packages are not usable on linux… And as you say : where do I get valid information about parameters to start with… And what to do if scans have strong color shifts…
I asked serveral friends and colleagues for scanned images (positives and negative), I got some, developed them using dt, and I was disappointed. This led me to the decision to use my Camera to create the copies. I bought a macro lens, a holder for the filmstripes or slides and the above mentioned tunable light-panel. This gives me a high level of control on the lighting conditions which I found to be extremely important (as always in photography). The results I get with this setting are better than every result I got using copies from a (professional) scanner during my “exploration phase”.
I have no experiences with ColorPerfect (I suspect you mean the PS plugin), I don’t use Windows and Photoshop.

Over the 2+ decades that I have been fooling around, in my spare time, with scanning the huge collection of negatives and what I think of as ‘slides’ (a most inaccurate word with which to name what you call ‘diapositive’), I have had just a few (say 0.5%) results from scanning (on either Nikon Coolscan or Epson V.700) which have been most pleasing to me. This suggests that the basic hardware - especially the Nikon device - is capable of producing a good scan. The really overwhelming factors which prevent this hardware configuration from consistently producing acceptable results are:

1). the skill of the original photographer in getting a properly exposed ‘slide’, That was almost always me, meaning that only the minority of the slides were good to start with - 60 years ago I knew even less than I know now.
2). the degree of care exercised by the developing lab - which has been unknown, obviously variable, and out of my control
3). the choice of scanner settings - which we have agreed is pretty important and the subject of an equally important lack of guidance.
4). the settings of the final (raw) image processing software (now darktable).

None of these will be dramatically improved, in my opinion, by my making a further investment in getting a ‘DSLR’ scanning set-up, such as you describe. I find it difficult to believe that anything I put together will be superior to my Nikon Coolscan V ED. By the way, does it have to be a ‘DSLR’? Any reason why my Fuji mirrorless camera wouldn’t work?
Specifically what information do you have from your ‘DSLR-scanning’ equipment which you didn’t have when using a ‘professional service’ and how does that help you to adjust your settings in darktable? For example, although it is available (I think), I have never been aware of the need to know what the colour temperatures of the scan LEDs in my scanners are. Even if I did know it, where would I use it ?

I asked about the advice on the ColorPefect web-site not because I wish to use PS or a plug-in to PS, but because of the suggestions the web-site owner makes on getting good scans from Vuescan before they are processed by ColorPerfect. I was wondering if those Vuescan recommendations would assist in darktable.

As a rule I prefer to work in Linux too; the only barrier to my working totally that way is that Nikon Scan (withdrawn by Nikon at least 10 years ago) is, obviously, a Windows-only application. It is, however, functionally equivalent, better documented and so easier to use than Vuescan, under either Linux or Windows.

I don’t say that it is not possible to get pleasing result with (professional) scanners. I think it is possible. But for me, using the mentioned setup, was easier, faster, and less expensive. It’s only my personal experience. If you already own a scanner, then, of course, you should try to get the best out of it.


I don’t care about the value of the color temperature. Assigning a “color temperature” to the spectrum of a LED is questionable anyhow. I love having a tunable light source, allowing to “adjust” my setting as decribed above. As far as I know this is not possible with most scanners (but may be I’m wrong).

I can relate, but I would recommend not to get too bogged down in finding the perfect set-up. My first negative scanning set-up included a roll of gaffer tape, a pile of CDs and a smartphone (as light source) and my DSLR. I actually got very decent results. It allowed me to digitize a whole bunch of very average negatives for zero money. Even though the results weren’t amazing, it was better than having no digitized negatives. Perfect can be the enemy of good, after all.

Software and modules like negadoctor can work wonders, so you can get away with sub-optimal white balance / exposure at the scanning stage. It really depends on what end result and intent you have. If you’re archiving prints for a museum or you intend to display these works, then it’s obviously worth investing money into the best scanner, a copy stand (if using a DSLR/mirrorless), slide/negative holder, strobes, color checker passport, etc. If you are just digitizing old snapshots for personal record keeping, you can use a more modest set-up, like I did.

If using a camera like I did, it’s important to find a way to get the negative/slide as flat as possible, use a macro lens, use the sweet spot aperture of your lens, have the scanned medium a suitable distance from the light source, and expose as perfectly as possible with no clipping (on the exposed part itself, clipping is ok outside this part). Focus is also critical at this stage. Other than that, I have been able to correct most white balance and colour problems in software, but your mileage may vary. As I mentioned before, I only have experience with negatives, not slides/colour reversal.

Ah! So true; I repeatedly fall into this ‘trap’. Your words do, in addition, assure me that, even though my current equipment approach (using consumer level scanner and well-established scanner software) will be able to give me scans - both negative and reversal - that meet my needs, which are personal only. Nobody else is going to be interested in these images, aside from (possibly) some alien explorers/historians who arrive long after we, as a species, have succumbed to a surfeit of advertising, scamming and malware (for which we can thanks the custodians of the internet).