A nontechnical, experience-based review on whether it's better to digital camera scan your film with a DSLR or mirrorless body.
My new mirrorless camera is amazing for taking photographs, but how will it scan film?
Out of the box, my new mirrorless camera was amazing. It was so easy to capture my daughter running and playing and moving, in focus, at f1.4 without any effort. I loved the fact I found myself using TLR-height perspectives since I could seamlessly shoot with only my live view. But the question remained- how did it stack up against my DSLR when scanning film? I use my digital camera for this... woof, probably more than I use my digital camera to actually make photos. Since this was such a big use case for me, I needed to do some side-by-side comparisons.
Gear
Before I kick off with my methods and results, I want to prime you with the gear I used. The mirrorless and DSLR cameras I use are roughly similar in fanciness (highly technical term). The DSLR is older (2017 v 2020), however they've got the same resolution sensors and are both considered to have great dynamic ranges and good in low light conditions.
DSLR - Nikon d850 (full frame, 45.7MP camera)
Mirrorless - Nikon z7ii with FTZii adapter (full frame, 45.7MP camera)
Lens - 60mm 2.8 macro (old, for SLR's so I use the converted on the mirrorless camera to attach and use it on the Z body)
Copy stand
Shutter release cables (DSLR cable is longer than the mirrorless cable)
Film carrier: Valoi 360 (120/ 35mm without sprockets masks)
Software: Lightroom and Negative Lab Pro v 3.0.2
Lightsource: 95CRI
Experiment
Process & Settings
Scanning
For each scanning instance, I measure the total time it takes me to scan a roll of uncut film (I don't measure cut film since it always takes longer to do a cut roll and there is more human variability injected in the process) and total height the camera has to be mounted on a copy stand above a flat surface. I also have a qualitative discussion on the general feel and ease of scanning each format with both types of camera bodies. The light source was set at 5300K, 100% brightness. All film was scanned in identical orientations so there should be little variation that can be attributed to a spatially inconsistent light source.
Converting
Since the software I use for conversions, Negative Lab Pro (NLP), algorithmically changes its color approximations based on the conversion area selected for an image, I manually went through and cropped the full frame for each image to the best of my ability without standoff. I did this so that when doing Roll Analysis, the opportunity for a like-for-like comparison of NLP results is more attainable. Is there some variability? Sure. But yano, at some point, we've all gotta make a call, and this felt like the most intellectually honest way of approaching conversions.
BW base conversion in NLP was done with the B+W Profile. Color was done with the Noritsu, Pre-Saturation 3 Default. I made sure all the Lightroom WB was identical for both DSLR/ Mirrorless passes prior to converting with NLP.
Camera Settings
Each camera was set in Aperture Priority mode, ISO 100, f10, filetype: large RAW for the scans. Addressing the elephant in the room on the front end: yes, I am aware NLP Roll Analysis depends on stable aperture and shutter speed settings to accounts for exposure variability in the results. I don't use this feature. There is always an error message that shows up during Roll Analysis telling me the exposures vary too much and I can either cancel the conversion process or continue without exposure data. I do the latter. I have found I get better results for skin tones this way. Accurate skin tone is very important to me, so I do what I do. You can do what you do.
Film Tested
35mm Color
I scanned the same cut roll of Kodak Gold 800 35mm film from the 90's. I also scanned a black and white roll to get the total time it took to scan them with each camera, but since color is where the rubber meets the road, I am just reporting my results from the color roll. The color roll has a wide variety of frame densities and is set at the beach (a notoriously challenging environment for NLP because of the vast white beaches and huge swaths of blue sky). I did this intentionally to see if there was variability in each body's ability to resolve detail in over exposed and under exposed images.
120 B&W
I scanned the same roll of uncut, home developed HP5 using HC110B developer shot on a Pentax 645n.
120 Color
I scanned the same roll of uncut, lab developed Kodak Gold 200 shot on a Pentax 645n.
Takeaways: No winner
TLDR: It's easier to focus on the DSLR, Mirrorless scan time was shorter, conversions are mostly a wash
The scanning process is faster with a mirrorless camera, however it is easier for me to focus on the grain in a frame using my d850's live view. I didn't need appreciably more or less copy stand vertical bar height to scan with either body, though the heights did vary slightly by format. Generally, the Z mirrorless setup required more copy stand height but that was largely, because I'm using an SLR lens and require an extra inch or so for the FTZii adapter.
I intentionally compared direct conversions without tweaks with identical Lightroom WB temp/tint settings. I found the colors on the DSLR to be better, but not by much. I actually think this might be attributed to the fact I have WB set to Kelvin on my DSLR and Auto on my mirrorless camera more than anything, because all I have to do to make the images look identical are extremely minor temp/tint adjustments in the NLP panel. I don't have the heart to redo these right now with my mirrorless in exact same WB setup. This is probably an important variable. I'm aware.
My biggest takeaway here is use whatever digital camera you have. I know this is a lame takeaway, but I think it's important to do this analysis so you know for sure you're giving your scans the best you can. In my case, I feel like I don't need the d850 anymore if I don't want the body. I considered keeping it just for scanning, but my mirrorless does as good of a job and does it faster. If anyone's in the market for a used d850, I know a copy *wink.
Nitty Gritty
Copy stand camera heights: Mirrorless setup was always a bit higher, but not significantly so such that I'd need a better copy stand for it. This wouldn't be the case if I didn't need the FTZii adapter.
35mm DSLR
35mm Mirrorless
120 645 DSLR
120 645 Mirrorless
Scan times: The DSLR takes me about 1.5x the amount of time to scan a roll of uncut film relative to the same roll with a mirrorless camera.
Focusing on the grain: Easier with the DSLR Live View
It's hard for me to explain in a post, but I found it far easier to focus on the grain with my d850 than my z7ii. Part of this is how far I could zoom into the photo. I also think this is influenced by just how electronic the Live View on the z7ii feels relative to the d850. I don't know what the technical explanation for this phenomenon is, but observationally, it's like the update rate was smoother (maybe even slower??) with the d850 than the z7ii. The z7ii moved a lot and felt a little more electronic and looked noisier making it more challenging to be ultra-confident the grain was tack sharp. I found myself turning on focus peaking with the z7ii to be sure I got the grain in focus and using it as a secondary checking measure (still not my primary). I didn't notice any sharpness differences in the final, converted positives though, so I clearly was able to get them both in focus, even if I was less confident with the z7ii in the moment.
Image Results
120 B&W
I mostly did B&W to compare sharpness. The top row are DSLR images, the bottom row are mirrorless images. I didn't see any appreciable sharpness differences, and tones came out identical between cameras.
120 Color
Left are mirrorless images, right are DSLR. There's a huge issue here with the pinks. I think this surfaced because I was in Auto WB on my z7ii (whoops) and had my DSLR set to match the Kelvin of the light source. It didn't take hardly any NLP panel adjustments to make them match, and they were all in the WB section, no color grading, so I feel pretty confident this is responsible for most of the color shifts I see in these results.
You can see it didn't take any adjustments outside the WB panel in NLP to match the two images 99.9%, so I feel pretty confident most of the added pink is from AWB of the Z7ii. In any case, it took very little effort to adjust them, and worked as a global adjustment to the whole roll.
35mm Color
Images on the left are from my mirrorless camera, right from my DSLR. As I mention in my process section, I intentionally let this be Roll Analysis built images in NLP with NO TWEAKS. I always tweak my final scans in the NLP panel. Always. But that's not a useful comparison for someone trying to see if the colors shift a lot between bodies, so I didn't for the experiment (and it pained me). Densities and contrast came out the same/ if there are differences they're not appreciable.
