I’m scanning with a Minolta 5400, and the colors I’m getting with NLP are really inconsistent. Maybe 1 out of 10 look correct. I’ve used both Vuescan and Silverfast, and while the scans they produce look quite different, they’re both very off.
I managed to get an emulated Windows XP VM with the original Minolta software running, and the colors there look fine, though it’s not really a usable workflow.
I’m wondering if the problem is that the Minolta doesn’t usually include much if any of the film frame. I was using a Plustek 8200 before and it left a hefty portion of the frame visible in its scans, and the colors look much, much better in NLP.
Just to determine whether this is the problem could you maybe do a scan with the film rebate between frames showing and then do your actual scan with the film rebates masked off by the holder? Not necessarily suggesting this as a workflow but just to see if NLP gives you correct results with an accurate sampling of the film rebate, though no reason why you couldn’t work this way I guess if it’s successful.
From the NLP Guide:
Before converting, use the white balance selector in LR to sample off the film border (you can do this once per roll and sync across photos). If no film border is available, or if Lightroom says that it is too bright, use ‘Auto WB’ setting in Lightroom.
Thanks for the suggestion! I scanned a blank frame from the same roll, and used that as the basis for the white balance for all images. It did work a little better, though there’s still something off about the colors in most cases.
That does suggest that perhaps it is some other factor rather than sampling the rebate. There are many expert NLP users on here so if you post an example I’m sure they can help.
It could be the scanner. That model produced the sharpest scans of the lot of them, but after a number of years’ use could also produce “off” colours regardless of which software was driving it.
The scans are extraordinarily sharp, for sure. Just kind of odd color. And when the color looks correct, it’s a little desaturated compared to my other scanners (Plustek 8200 and Epson v700).
Please feel free to email RAW samples to me at nate@natephotographic.com and I will check it out. That’s the easiest way for me to see what the issue is.
Yes, I used the Minolta software (via a Windows XP emulator), Silverfast 9 and Vuescan. The color from the Minolta software seemed fine, but the process of using the emulator is cumbersome enough that I didn’t use it that extensively.
The scans from Silverfast and Vuescan look quite different from each other, which seems surprisingly since I’m getting RAW files from both.
That’s right, I’m using an M1 Mac. I could probably run Windows 10 emulated, if Dimage works better in that environment. I’d assumed it would be best in XP though.
Tiff is fine for NLP. I just scan as positives with the largest color gamut available in the minolta software. That’s the end of it.
Works perfectly fine with NLP once in Lightroom. Obviously the lastest refinement of NLP (like roll analysis) may not work as automatic exposure isn’t disengageable IIRC.
I don’t use Vuescan for two reasons : One, its implementation of ICE isn’t working as well as Minolta’s. Two, its calibration process brings artefacts. In my experience, “hot” pixels are badly discarded and you can end up with doubled or even tripled lines along the scanning path. Never had any problem with Minolta Dimage.