Untitled photo

Really, that first photo is the entire article. To digitize film with the new Fujifilm XF 30mm f/2.8 Macro just get yourself an LED light / tracing table, and rest the lens, hood-down, on the frame you want to photograph. No tripod. No additional setup. No additional accessories, unless you want them.

If you’re good with that then the rest of this post is just elaboration and repetition.

To continue giving the conclusion first, I own a Nikon LS50 35mm film scanner, a Canon 9000F flatbed, and a Fujifilm XF80 macro lens. Until recently, I also worked at a photo lab with a commercial scanner. I’ve used all of those methods to digitize both 35mm and 120 film.

The XF30 is as good as the best of them, and is far easier to use.


Before writing this I went on the internet looking for thoughts on how to get analog film into a digital computer, and let me tell you, dear reader, a lot of it isn’t very good. Frequent advice was to use a long lens, manual focus, all you need is cheap adapted vintage, whatever: nope. They're wrong. And you can tell that they're wrong because they call photographing film with a digital camera "scanning".

Scanners scan. Cameras photograph. They are different tools with similar purposes that use different methods – an easy comparison would be hammers and screwdrivers. Scanning and photographing are different; being hammered and being screwed are different. Life is easier when you keep these distinctions clear.

Happily, the word “digitize” exists. It means 'making a digital version of analog media', regardless of what method is used. Scanners digitize. Cameras can also digitize. Neither tool is inherently superior, functionally or morally, but their methods and compromises are different. Choosing between them is a lot of what I’m doing from here on.

First, a detail shot to show quality. This is a 1200 pixel crop from Ilford Delta 3200. I'd say that we're resolving the grain fairly well, pulling about as much detail as possible from the film. If you click to open the image in a new tab and magnify it to 100%, that should be about 1:1.

For best practices I set the camera – I'm using the 26Mpx Fujifilm XT30 – for electronic shutter and 2-second delay. (And a custom white balance, which absolutely does not matter for black and white.) It's actually a bit easier with the cable release, but the joy is that the XF30 does a great job without much "stuff" to stuff around with, so that's what I wanted to show.

But we're just looking at the digitization quality here, not a whole-system sharpness test – the original frame was recorded handheld with my rangefinder, and probably using my compact 40mm lens. And it's Delta 3200. (But just look at the definition in that grain.) At some point I'll try out sharper film from a better setup and see just what it can do.

I've tried comparing a few negatives that were digitized with both my Nikon LS-50 35mm film scanner and this lens, and both methods were about as detailed as each other, although the XT30 put more pixels on the frame. That's an excellent result. I'll also say that I like the camera tonal range better. And did I mention how fast and easy the process is?

But nothing is perfect. The hood is just a bit too short to let the camera cover a  35mm film frame edge-to-edge. So here's more about that:

Untitled photo

The emphasized frame is a single capture from the XF30 with the stock hood. (An aftermarket hood with a different length gives different magnification, which we'll get to later.) By comparing the dimensions of the stitched frame to a single file, I'm seeing ~93% coverage by length and height, or about 86% coverage by area. That's not perfect, but remember when Canon used to say that using legacy film lenses on smaller-sensor digital cameras was a feature because you'd "use the best part of the lens"? This is something like that.

It is worth noting that many scanners don't go all the way to the edge, and neither do most non-electronic camera viewfinders. A single frame from the XF30 is often enough, and it's both simple to produce and very high quality. But we can do better if we need to.

For the full image above I turned the camera "vertically" to span the width – height? – of the film, where it almost covered the entire filmstrip, sprocket to sprocket. This means recording multiple images, stepping along the frame, to merge into a panorama in post. Two can be enough, but software can get confused, so having more overlap with three images is better. This method is more work, but gives full coverage of the exposed area, with the corresponding 16% increase in pixel count. Here I'm using a 26Mpx sensor, so my full 35mm film frame is about 30Mpx.

By recording two rows of photos, overlapping with the top and bottom of the 35mm frame, it's possible to cover more than the film strip. (For the photo below I used four images, recorded horizontally.) This is handy for cameras that expose the entire film, which a few do and are otherwise very challenging to digitize. This might prompt me to get my Sprocket Rocket cameras back out… but probably not.

Untitled photo

But beyond needing more work, there's a tradeoff to stitching multiple photos. Lightroom, like Apple, thinks that it's smarter than the people who use their products, despite some obvious shortcomings. So Lightroom doesn't have a "just arrange them" stitching option and doesn't understand flat-field tiles. This is why the overlaid "danger" images don't quite align – Lightroom has auto-distorted the merged file.

Other software can do a better job stitching together these images, but that's more work than just using the catalog+edit program. So there are three workflow options. Keep it simple with a single shot and possibly don't get the originally intended composition; I can see what the frame will look like through the viewfinder, so I can decide case-by case. If that's not enough coverage then I can record multiple images to merge later. And if Lightroom biffs the blend, then escalate and move the files into different software until I get something sensible.

Happily the distortion that merging files (potentially) introduces is usually minor, and mostly only apparent when doing an A/B comparison, as here. I'll probably never do that again. But it is a reminder that "computational photography" is different from "photography".


Untitled photo

Here's where things get a bit more wild.

This is a medium format 6x9 negative, and this crop that includes the film border is 207 megapixels. The image area itself, when I crop it down comfortably, still comes out to about 170 megapixels. That might be a bit excessive for a pinhole photo, but it really is amazing to be able to zoom in and marvel at the lack of detail.

(For a camera with a lens, though, watch out. With the XT30's files I can see the grain in Ilford FP4, which isn't easy to do. My next step will be to get my big Fujifilm GX680 out of storage and put a couple of rolls of Acros through it. I've owned that camera for twelve years, and will finally get to see what its photos really look like.)

To include the entire film area I photographed across the negative in three rows, and being generous with my overlap produced 23 files to merge. Excluding the border, and taking a bit more risk by only recording two rows with less overlap, let me merge another frame from 14 photos. Either way, that's a substantial amount of storage space, and the merged files are in the 600-900 megabyte range.

If only there was some easy, inexpensive way to decrease the magnification a bit…


Untitled photo

…so here's where we talk about using a different lens hood.

The stock lens hood – all dimensions approximate – gives the camera a working distance of 26mm (from the front of the lens) for a coverage area of 33x22mm. 135-format film is 36x24mm, which is why I've gone through hoops about merging photos. But you can use a different hood if you want to. That online bookstore sells them fairly cheaply, just make sure they're a barrel design, not that vented rangefinder style.

A shorter hood gives more magnification and less coverage. I don't see much need for that, but perhaps super-sharp film and exemplary working methods might deserve it. A longer hood will let a single frame cover more area, in exchange for lower resolution, because it's putting down fewer pixels per millimetre. Fortunately, the XT30+XF30 has a whole whack of resolution, so that's the direction I want to go.

I happen to have a random metal screw-in hood kicking around, and enough stepping rings to make it fit the XF30's petite ø43 filter thread. Other random hoods will undoubtedly be different, but mine gives the camera a 44mm working distance, and covers (approximately) 47x32mm. That's a single frame from the rehooded XF30 above. The 26Mpx camera still puts about 15Mpx on the image area, which is plenty for everything but very large prints. So simply by changing hoods I can choose between wider coverage and higher resolution as needed, still with one-shot convenience.

For digitizing 120 format there's also an improvement in convenience and decrease in resolution. Now covering a 6x9 frame can be comfortably done with just eight photos, and a 6x6 with only four. That's still producing 85Mpx for the wide frame, and 58Mpx for the square. That's an abundance for basically anything.


Untitled photo

I've been dissatisfied with my scanning options for a while now.

The Nikon LS50 that I have is excellent, and it's very nice to feed it six frames at a time. But it only understands 24x36 frames, which falls down with sub-frame and panoramic cameras, not to mention the continuously exposed ribbons that become my Convolution series. But for what it can do, it's great.

Scanning 120 negatives is another story. I have the Canon 9000F, and it's… not great. And there's really nothing substantially better – flatbed scanners haven't advanced in over a decade. (Even pinhole cameras have made significant improvements in that time.) And loading film into a flatbed is a journey all on its own. Cutting the negatives to fit the little bed, needing three hands to get them into the holder, and keeping everything clean and dust-free is the pits.

And for camera-work I own the XF80/2.8, which is a fantastic macro lens. I tried digitizing 120 film with it once, but the more complicated setup for the longer lens introduced all kinds of out-of-plane defocusing and was a massive hassle with the merged file. Photographing 35mm was less punishing, because it's just a single frame with fewer opportunities for things to go wrong, but it was still an unreasonable amount of work. There are tricks for aligning the camera to the lightbox – use a mirror, when you're seeing the centre of the lens reflected back into the viewfinder you're square – but I needed to futz around with two tripods and a focusing rail. The setup time was substantial, the non-image area on the lightbox needed to be masked off, and actually photographing the film had to be done in near-dark. And it was still never quite square. Pfft. Never doing that again.

XF30: just rest the camera, hood down, on the lightbox. Done.

My flatbed Canon may never come out of storage, and I'm not worried about my Nikon scanner breaking any more. I'm not saying it's worth buying the XF30 just to digitize film… but it's a good lens for other things, too, so maybe it's worth considering.