I’ve had a love/hate relationship with NLP since I started shooting film last year. I love having many conversions come out fine, and that NLP offers a lot of ways to correct conversions and adjust them to my liking. I hate that many conversions come out with odd color shifts, and require too much time tweaking to get right. Last weekend I tried Film Lab for the first time and was surprised at how it “just works” with very little tweaking necessary. No sampling the film strip to fix the white balance, no having to worry about areas outside the image, and only a few simple controls to adjust density, white balance, and few other things. You set the film stock, your camera, the light source type, and the automatic conversions are pretty close to a final image. I’ll admit I was tempted to just switch over to Film Lab going forward.
I did want to give NLP another go however, and properly test out Roll Analysis. I shot a test roll of Portra 400 on 35mm, and included shots of a Calibrite ColorChecker. I then did a number of conversions to compare the results:
A. Single image analysis, where I shot the negatives in aperture priority mode with no exposure compensation.
B. Film Lab.
C. Roll analysis, where I shot the negatives in manual mode and exposed far to the right (ETR).
D. Single frame calibration using the image of the ColorChecker. The image was taken in sunlight around mid-day. Also shot the negatives in manual mode and exposed far to the right (ETR).
All were shot using the Negative Supply 4x5 Light Source Basic MK2 (99 CRI).
For NLP I used the basic color model, pre-saturation 3, the lab standard tone profile, WB none, and HSL natural. I didn’t adjust any other settings. For Film Lab I used the auto suggested settings. The results were interesting. Here’s a comparison of 4 images with the above conversions:
Note the issues with single image analysis (A). Image 1 has a weird yellow/green cast. Image 2 a magenta cast in the mid tones. Image 3 is over-saturated, too contrasty, and also has a yellow cast in the mid-tones. Image 4 has too much magenta in the shadows. Can all of these be corrected in NLP? Yes. But this is the kind of fiddling that drives me bonkers, especially when other conversions seem to come out fine.
Note the Film Lab conversions (B) are mostly OK. Image 4 also has some magenta in the highlights and shadows. But overall they’re a good starting point for adjustments in LR almost all the time.
Now look at roll analysis (C). These are looking pretty good! Not quite as good as Film Lab but better than single image analysis. The exception is image 3 of course. I did read that this can happen with some images in NLP and the solution is to lower the BlackClip. While not shown, lowering it to -10 resolved the issues.
Where things get interesting are the conversions where I used a single frame calibration of the ColorChecker chart (D). While this is subjective, I like these the best of all 4.
Which brings me to a question and feature suggestion for @nate . Why can’t NLP come with some pre-built single image “rolls” based on color chart images taken with different film stocks, digital camera types, and light sources? Similar to what you specify in Film Lab. I’m not expecting you to create all of these permutations and combinations. But one idea is to crowdsource this from the community. I’m sure lots of folks would be willing to chip in and help. Myself included.
Thoughts?