I’ve been using NLP for quite a few years now. First with my Plustek and Vuescan. Now with a Valoi Easy35. Throughout all the different versions I have found that NLP can be quite inconsistent with WB, whatever scanning method I use.
Currently my main 2 points are:
Auto-Warm is probably 99% a total orange mess.
The newer Auto-Mix is an improvement on Auto-Neutral but often introduces a notable green cast to the image, especially in the highlights.
Does anyone else find this?
NB: My NLP is correctly set up and I know about WB off the film border.
Getting correct white balance can depend on a variety of factors. Do you develop yourself or use a lab? I have found that even minor processing errors can have an effect on the auto wb interpretations. Even the best negs/scans will probably need a touch of manual white balancing here and there.
I always use auto neutral as a baseline and then work off that. I’ve found the other auto options to be relatively useless. But if even after tweaks to wb and tint things still look way off you might have a conversion interpretation issue or processing problems.
@mattdoerr , I work the same way as @winslow … and I don’t care whether WB is accurate or correct, as long as I can make an image look how I want it to look. Unless it’s about reproductions, “correct” or “accurate” are concepts that are best left behind imo.
Selecting one of the WB presets on NLP’s second tab can bring an image closer to target or take it away from it. And from what I understand from NLP’s inner workings, this is to be expected and dependent on chroma and luma characteristics of the source image.
If we look at the tone curves of a converted image, we can see flat and/or steep curves and a slight change of balance can make a big difference. If we’re lucky, NLP will match our intentions and more often than not, it requires additional adjustments - at least that is what I see with the negatives I have. Most of them are about 40 years old, taken under various lighting conditions and using all kinds of film.
If you want the cold hard truth. Best workflow is to buy a narrowband light capable of individual RGB strengths. Calibrate the film base so the histogram lines up, make RGB equal strengths. Then all you need to do is literally invert your scan without any software algorithms. You essentially delete the orange mask at the physical level, not software level.
This makes it so if you shoot daylight, there is no white balance needed. For sunsets/mornings you need to perform color timing with RGB curves or color grading to boost warm/cool tones.
Hi Matt, something to note also is that film will produce color casts/shifting depending on your exposure, film stock, developing, etc. For example, Portra is known to shift to a green cast, Ektar suffers from excessive reds, etc. Even if you hypothetically have a “correct” white balance, there can and will still be color shifting that can’t be fixed by making a “global” color adjustment. Accurate colors with film will more often than not require specific “local” color adjustments to different tonal ranges (highlights, midtones, shadows).
I used to always wonder why when I tried to WB my scans they always looked “too cold” or “too warm”, but never “correct”. The reason is because “global” adjustments will not fix “local” color shifting.
Scanning also produces color shifting around the edges of the negative because of vignetting from the film holder. After inverting the digital negative file, there will be excessive blues in the blacks/shadows around the edges of your image (this is because the negative gets darker/warmer in the vignetted areas around the edges, which produces a lighter/colder “washed out” tones when inverted). The reason only the blacks/shadows shift to blue and not all of the tones is because the brightest areas of your film negative are the most transparent, which means they are most affected by pure orange film base color, which means there will be more blue/cyan blacks around the vignetted edges when the digital negative is inverted.
Some solutions would be to use masks and/or specific tone curve adjustments. LR has a great color and tonal selection masking tool that I use all the time. It’s great to locally select specific color shifted areas of the image and adjust only that. Another solution is to research the specific color characteristics of your prefered film stocks and learn how and why they color shift at different exposures, development, etc.
I think you’ve got it in a nutshell not that I’ve tried it, high quality film scanners have always worked using those principles, the movie industry options like the Arriscan XT work like that and they’re not going to mess around with something that is only second best. The problem is that there is only one panel that is sporadically available to buy though you can make your own to his plans I think.
The other problem is that it is very difficult to demonstrate how one solution is better than another when you’re working with colour negative as there is no original to refer to. I think this is probably why no mainstream companies like Valoi or NS have gone down this route (yet?).