VueScan: Camera RAW vs RAW DNG vs calibrated TIF

Flatbed and Film Scanners
1

Hey everyone,

I’ve done some experiments lately to find out which scanning method leads to the best results and I’ve had some unexpected results.

I have access to a Nikon Z6 mirrorless camera with a Sigma 50 mm f/2.8 EX DG Macro and to an Epson V600 flatbed scanner. The Nikon Z6 is undoubtedly better for 35mm film, but the flatbed scans seem similarly sharp to the mirrorless scans for the larger medium format film. However, I have mostly preferred using my Nikon Z6 because the flatbed scans always had some weird/less appealing colors after initial conversion with NLP that proved much more difficult to fix.

I have since acquired an IT8 calibration transparency and made an ICC profile for my scanner and found out that the default colors were extremely off. I can obviously only make use of the ICC profile when exporting normal 48-bit TIF files from VueScan, so I decided to give that a try. Up to that point, I had only been using RAW DNGs from VueScan as per the NLP instructions, but it seems that using 48-bit TIF files recorded with the generated ICC profile leads to scan that are much easier to work with (at least for me). I converted all the files in Lightroom Classic using NLP v3.1.1 using the Basic Color Model, a Pre-Saturation of 3, a Border Buffer of 5%, ToneProfile LAB-Standard and WB Auto-Neutral (for the two RAW scans I also used the Color Picker to WB off the film edge before conversion) and did no further editing and here are the results on some Portra 800 (merged into one picture because I’m only allowed to post one picture per post):

scans
scans1920×698 1.66 MB

I thought I’d share my results since this (at least apparently) goes against the normal wisdom of using RAW DNGs for best results with VueScan. Maybe one could still theoretically get better results with the RAW DNGs, but it would need way more messing around with the colors to get it to look good. I’ve tried the same process for the whole roll and all RAW DNGs had the similar issue of being weirdly desaturated (especially in the reds) and sometimes a bit too blue/magenta.

2

Welcome to the forum @kobbolt

…even in practice, one can get better results doing one thing or another…

The word “better” usually triggers the following:
NLP’s approach to convert negatives is to adjust images in order to get neutral whites, grey and blacks. Depending on what is in an image conversions can take on some cast or be off completely (I exaggerate) or almost perfect, but again, what does perfect mean.

For the sake of one’s peace of mind, it helps to assume that NLP delivers a starting point that requires more or less tweaking in order to provide a result that meets one’s expectations.

With all this in mind, your image of an ICE seems to be good for you if done as you describe. Also, your “recipe” will probably be useful for similar images, but what if you come across a “Freccia Rossa” or a steam loco in lush greenery? Your recipe might match or not and some changes could be necessary in the latter cases.

My take is that a spot landing can be achieved, but not in the sense of a silver bullet that works in all cases. This means that it’s useful to pursue a once established scanning and converting procedure and to get off that beaten path if an image or set of images asks for it. This might feel discouraging at first, but after all, we’re not dealing with computational photos that have been optimized in camera to look like what a common sense of beauty might require. And still, NLP saves a lot of time converting compared to anything we can do manually. Photography is an art and requires some skills, and converting negatives is adding to the pile.

If you test your method against a bunch of images that have variations in colors, contrast, lighting etc. and find that your method delivers in all cases, we’d certainly be happy to read about it here.