If not the slide, possibly the backlight. There are all sorts of discussions (many on this forum) about which backlights are best for the job.
Just a data point but this is a professional scan (done by the lab that developed the film) and the second one is homemade using my DSLR + RT :
I think mine is better. Granted they had an option for better quality scans, but this was quite expensive already. I tried to edit it in the same style, but the RT version could be taken in any direction. Keeping dust out is difficult I found.
5 Likes
I’d hope that a 4700K Solux halogen with 98 or 99 CRI would be about as good a light source as possible. Having said that I do see some merit in the argument that Kodachrome was designed for slide projectors so the lower colour temperature of the (also halogen) projector lamp may be better. So far once adjusted for white balance I have not seen a difference but re-doing the grassy slide using the projector lamp is on my ‘to do’ list. It will need to wait a little while because I will be away from my slide digitising setup for the next week.
Would be very interesting, but hard to achieve! One would need the same slide(s) being scanned by various methods.
Here is a comparison of what I meant concerning Kodachrome and ICC-profile.
Center is without any profile, right is with a profile derived from a Kodachrome target and left is what I would call my optimum results after quite some work:
But one still has to be careful since my images were rendered into AdobeRGB, displayed on a calibrated monitor (EIZO) but the screenshot does some conversion of colour space.
Hermann-Josef
3 Likes
I was referring to the difference between say, full spectrum white light, and narrow band rgb lights, of which there is some debate which is best. In many cases yellow and orange may prove most problematic, which could perhaps explain why your grass goes from green-yellow to green-blue. Just one of many possibilities. See @NateWeatherly graphs here: Digitizing film using DSLR and RGB LED lights - #5 by Thanatomanic
I am using darktable for processing my slides and still heavily struggling with that (already got some help from your community). My focus with digitizing is not find the top five slides of the last 20 yrs and get them perfect, but getting as many as possible slides digitized with reasonable effort and a decent quality. My technical setup is shown here. I still ow the community the “processing part”, but after reading this thread here full of intimidating professionality, maybe I never dare to make this second “tutorial” …
1 Like
The IT8-calibration will take care of any differences between the standard illumination D50 and the light source actually used.
Unfortunately the transmission data for my Kodachrome target are not available from the manufacturer. But the argument holds for all types of slide material. I have re-calculated the reference table for my Ektachrome target for tungsten illumination. Here is the difference, D50 at left, tungsten at right):
Hermann-Josef
1 Like
Please don’t be put off. What matters is that you get results that you are happy with. Some of the calibration technicalities are beyond me.
I had not really given it much thought but guess my ‘focus’ is twofold - firstly there are a lot of slides that are still in my scanning backlog and have not been on any scanner yet. Many are very average photos but of subject matter that is now gone and that I care about. I will be cherry picking because life is too short. Secondly there are a high proportion of scans from the OpticFIlm that are ‘fails’ because it did not handle the shadow areas well.
The recipe that I posted with the colour balance module and a parametric mask works for putting some life into dull greens. I’ve tried it on a few more slides now. It should work for other colours too. Which colours you should adjust, in which direction and by how much… is a more difficult question. You can rely on your judgement (and looking at the original slide of course) or you can go down the calibration route… or mix the two.
1 Like
Thanks. I must admit that I’m not sure what to make of the two calibration target scans. The yellows do look noticeably warmer in teh right hand one. For the Venice photo I don’t see a huge difference. Both look good but I slightly prefer the right hand one. Not much green or yellow in it though.
Oh wow, very well explained. That would be a reasonable setup for my father in law, who has about 15000+ slides to digitize. Thank you very much for the effort of doing such a detailed video.
I have still one question: Why did you remove the collimator lens? In theory, it should give you more even lighting. And did you take a flat frame to judge the evenness of the lighting? Unfortunately, darktable does not yet support flat frames for correction …
Oh, please, go forward. You can hardly find material as detailed and well done as your part 1. That does not mean that your setup is for everybody, but the documentation is extraordinary, and the setup is spot on for its purpose: Best quality in zero time 
.
1 Like
The left one was rendered with the standard D50 illumination used to calculate the reference file, the right one used the tungsten spectrum.
In a “real” slide, the differences are not easily recognizable, depending on the colours present in the scene. The colour of the water, e.g. is quite different in the Canale Grande image. However, there are a lot of differences seen in the colour patches of the target, e.g. compare columns 15 to 19.
In the end, the IT8-calibration is made to make the colours in the image the same as if you viewed the slide either in sunlight (D50) or with a projector lamp (tungsten) – if viewed on a calibrated monitor, of course. By and large it is also a matter of taste.
However, the comparison also illustrates the power of the IT8-calibration. Kodachrome scans in most cases end up with a blueish colour cast, which can only successfully be removed via the appropriate ICC-profile. This colour cast results from the difference in illumination (e.g. LED vs. sunlight) and in the different spectral sensitivity of the detector (CCD) vs. human eye.
I can also refer you to the experience of another Kodachrome-user reported in the German scanner forum (If you do not speak German, you might want to use google-translate, which gives reasonable results.).
Hermann-Josef
1 Like
Thanks for the friendly reply. In fact I got already advice and have shown some sample results here: Looking for "automatic" darktable style for processing digitized slides
1 Like
Thanks! Happy it helps. As below I got advice from community here. And no: I did not use flat field correction as darktable seems not to be capable of this. But I learned that disabling the base curve is essential.
@chris With respect to the collimator lenses: I had exactly the same thought and tried it with them: but you can stop down aperture heavily and still see the glowing lamp as a spot. So the small sheet of teflon realy does the trick (and you shoudln’t overestimate the quality you got in former times from projection - that means the projector-method to me seems quite comparable to projecting. Scanning will be of course superiour in quality).
1 Like
Thanks. I am having a few days away so digitising and post-processing will be on pause but lots of time to read and watch tutorials
Hm, actually I thought on using both together, but this may be difficult due to available space.
If you download the demo for Babelcolor Patch Tool and open the application file contents there is spectral patch data for a Kodachrome IT8. Who knows how close your slide color is to the one that was measured, but I just thought I’d let you know!
Many thanks! I will give it a try. There are, as far as I know, two different targets around, one produced by Kodak and one manufactured by LaserSoftImaging.
Hermann-Josef
A really simple idea occurred to me last night so I gave it a quick try. When comparing the slide on the light box with the image on screen the need to refocus my eyes and account for differences in illumination made things more difficult. I’ve seen articles suggesting using a dimmable light box (mine is not dimmable but I might modify it). This idea is simpler and I think better. I doubt I am the first to think of it.
You may be able to see that I did not quite do it perfectly and get a clue about what I did. Here is a bigger clue…
Basically… darken the room, tilt the screen back by about 30 degrees, run an app with a blank white window (notepad in this case) and position it on top of Darktable so that the screen backlights the slide. I used a junk slide to allow the slide being checked to sit higher up the screen. Putting a card frame around the darktable image being compared would also be a good idea I think.
Hm, I did a lot of research during the last couple of days and found that there is an interesting twist to the use of collimated light. There’s a wikipedia article on the Callier effect. From the wikipedia article it is hard to understand what happens, and the article is even a bit contradictory. But what I take from the example images there and from other sources is, that the local contrast, and for some media also the global contrast, of a photographic film image increases with degree of collimation.
I am not sure what would be the better approach, but it seems that you are emphasizing the film grain and dust/scratches with collimated light rather that improving the actual image contrast.
It seems also that professional scanners such as the fujifilm frontier and noritsu are working with diffuse illumniation (judging from the resp. patents).
It is clear that a diffusion layer is required in the scanning setup. Placing it before or after a collimator lens could change the appearance of the scan result regarding micro contrast, global contrast and visibility of imperfections. It would be interesting to test this in a slide projector setup. Unfortunately I do not own a slide projector anymore, but maybe, @seume, you have the chance to test this …