This weekend while organizing I pulled out my whole stack of developed film that I keep in a cabinet. I remembered there was a film frame that I wished had been better scanned at the lab, so I pulled it out, with the intend of just doing a regular scan using my Nikon Z6 Ⅲ. I did that, and the result was, fine…
But then I remembered that the Z6 Ⅲ is capable of pixel-shift super-resolution. I hadn't considered that feature for this purpose until now, but it seems perfect for the job.
For context: pixel-shift super resolution is a feature in some recent cameras that makes use of the vibration reduction actuators on the sensor to take several photos with the sensor at slightly different positions, just enough to capture light for the exact same spot from different sensor pixels, especially sensors behind different Bayer color filters. The different images are then combined to producer a higher resolution picture.
One of the drawbacks of this tech is that if anything moves in between the different shots of the series — even something as minimal as leafs on a tree swaying in the breeze — that motion will cause artifacts in the final image which look like RGB stripes and other such artifacts.
However, for scanning film this shouldn't be an issue, since the subject is already a static picture. Given enough light from a light table or scanning rig, this should be golden.
The setup I used was:
- Film scanning rig with lens ring adapter and 5000K light panel;
- AF-S Nikkor 105mm ƒ2.8 VR;
- 1/125s ƒ8 ISO 100;
- White Balance fixed at 5000K;
- Pixel-Shift mode set to 32 shots;
- Images combined, then color-inverted and -corrected on Nikon NX Studio, with resolution-priority noise-reduction setting;
- Combined 96MP image exported to Affinity Photo, where I fixed dust spots with the clone tool, and further tweaked white balance and levels to taste, then downscaled to 24MP;
- 24MP image exported to Apple Photos so same color adjustments that I had used on the lab scan could be applied, then both exported to JPG, which are the images you see below.
The results:
| Lab Scanner | Nikon Z6 Ⅲ |
|---|---|
|
|
|
|
I'd say it was a success! Although digital cameras are not putting dedicated high-resolution scanners out of business anytime soon — especially since it's laborious to take photos and process them one by one, where the dedicated scanners can go through a whole roll of film in literally seconds — it is nice to be able to get something at a similar or higher level of quality for those slides/frames that deserve it.
Subjective differences aside (yes, color balance is subjective; yes, even on film once scanned; yes, you lab operator color-corrects all your scans by eye), I think the important differences here are sharpness and noise. However, I can't say for sure that the lab scan is not post-processed. That noise looks smeared to me, and I'm guessing it involves some downscaling and JPEG-ing.
That's one benefit of scanning it myself: I can decide how much noise I tolerate. That being said, I was hoping for less digital noise from a super-resolution composite, but I'm still happy with the results. It definitely looks sharper.
I'll certainly be using this technique to scan film in the future, and if I refine it further, I'll report on my findings.