I’ve been trying out a 1970s Nikkormat mechanical SLR that I bought last year. After shooting a few black and white films and being pleased with the results I’ve recently shot a roll of Kodak Ultramax and have been scanning the negs.
I’ve figured out a way to get results that don’t look awful but it seems to me I’m using Darktable in a slightly odd way.
I’m using Vuesdan and an OpticFilm scanner. Having tried both Vuescan’s negative inversion and Negadoctor in DT, I much prefer Negadoctor so I’m scanning in Vuescan as ‘image’ and saving as a 24 bit TIFF.
I think the negatives are reasonably well exposed and the histogram in Vuescan looks well spread out with a spike at the top end where it’s seeing some of the unexposed part of the frame. The DT histogram looks the same as the one in Vuescan if I set it to linear (but the screenshot below is showing a logarithmic histogram).
Trying to lighten the image, either using the exposure bias slider in Negadoctor or the Exposure module produces a very washed out result - this one is with Exposure but the other way is much the same.
I’ve had good results by using Negadoctor to lighten the image until the blacks just stop clipping and then using Color Balance RGB to increase the global perceptual brilliance by 25%. Then I duplicate the Color Balance RGB as many times as needed to get the image to a sensible level of brightness (this image needs four instances).
Having four instances of Color Balance RGB seems wrong to me.
I’ve tried changing ‘curve low’ and ‘curve high’ in Vuescan - by default these are 0.25 and 0.75 so I guess it is altering the image. Setting them to 0 and 1 (or as near as poss) changed the results but not in a good way - the image after Negadoctor was still dark and also had a magenta cast.
I was a bit surprised that Exposure produced such bad results - I’ve not noticed a problem when using it on DSLR images or DSLR slide captures.
So I think I’m doing something wrong but I’m not sure what. Are my negatives more underexposed that I’d thought? Have I got Vuescan set wrong? Is there a better way to brighten an image in Darktable than duplicating the Color Balance RGB module multiple times?
It looks like a false Dmax setting.
You may scan the primer part of the film so you get both Dmin (no light touched the film) and Dmax (received maximum light) on the same picture.
For the blacks :
when you select the Dmin color, make sure the red channel fall between 90% and 100%
if it’s too low, that means you need to increase exposure from the scaner (I don’t know how) or increase exposure in the exposure module and select the Dmin color again.
Note that you may keep the same parameters for all the picture you’re scanning from the same film in vuescan or else, so you just have to copy/past Negadoctor setting to your scanned picture after having setup the “film properties” tab.
For the whites :
Select the Dmax part of your film with the eyedropper.
you may have better results like this
edit :
This is what I get using you screen cap of the negative picture with some adjustment in the other tabs:
Thank you. That seems to do the job without needing the duplication of CBRGB modules. I haven’t had chance to go through the film and find somewhere where the film is fully/over exposed but that seems like a good plan. I’m already copying the sampled unexposed values between frames - easier than sampling for each frame.
I’m not sure if Negadoctor tries to autodetect these values but if so then maybe it is seeing the opaque scanner frame and assuming that black area represents the densest black on the negative.
Ah yes - I’ve just looked again at the negs. The first ‘shot’ is only half a frame. I’d always assumed it was because there was a leader with no emulsion but I guess it’s normal film but was (fully) exposed when I loaded the camera.
Now I just need to fetch the scanner out again and scan that part of the film - I didn’t bother with the first half frame because I thought it was junk.
I’ve now scanned the first half frame and have sampled both the fully exposed part and the between frames unexposed part.
The Dmax sampled was 1.5 (the default value in DT is 2.05).
Using the value of 1.5 on the sample image improves the result - I now need less adjustment with CBRGB and the resulting colours look more correct for early evening with a fairly low sun (which is how the light was). A side by side comparison - the DMax 1.5 version is on the left
I still need a fair helping of adjustment though - typing the value restricts the maximum per instance to 100%. Some frames are ok with less than 100% but others need a second instance to achieve enough brightening.
I think that VueScan is doing some auto adjustment on the brightness of each scan which suggests to me that the negadoctor settings from one frame may not be entirely reusable for another - certainly the brightness of the unexposed part of the scanned frame seems to vary a lot. I might be able to fix this by locking the exposure in VueScan but then I’d need to rescan the whole film… I’m not keen on that idea.
Thanks. There is a check box that I think will do the job. Finding the checkbox is easy but the snag is that I’d need to rescan the whole film. I think I’ll save that as a lesson for the next film and make adjustments individually for the current one.
The words online also suggest that the preview operation sets the exposure. I did just one preview and then just hit ‘scan’ for each new frame… so it may have actually ‘locked’ the exposure anyway (but I’m not 100% sure on that)… only snag is that I did the scans in three separate sessions.
Anyway… probably going wayyy off topic for Darktable.
