A few weeks back, I wrote an extensive post detailing my attempts to develop a DSLR/mirrorless digital camera film 'scanning' workflow. Over the holidays, I had a lots of opportunities to try out some new things, and throw some new gear at the problem. I've also had a chance to find out the pros and cons of the method. What follows is an update of the original post, which can be found here.
For those who would like the executive summary, here it is:
This method works. Well. Very well. It's not perfect, but most of the putative 'cons' of the process are related to the post-processing workflow required (and in particular to the inability of ColorPerfect to do batch conversions). However, from a pure image quality standpoint, this method produces outstanding 'scans', that easily rival the best film scanners I've ever used. In some respects, the method isn't cheap, but it will save you a great deal of time, and because you're using a camera and not a scanner - which spends most of its life sitting idle on your desk - to digitize your film, there is a huge opportunity-cost benefit to this method (i.e. you can use the camera as a camera, too).
First, some updates.
1. I sold my Fuji X-Pro1 (which I seldom used, anyway), and purchased a used Nikon D800 and a Tokina 100mm f/2.8 Macro. The D800 gives me a 36MP image of a 35mm film frame at 1:1, and that gets reduced down to about 30MP, once the non-critical part of the image is cropped out (sprocket holes, film leader/border, etc).
2. I'm using the Lomography Digitaliza (120 version) to hold my medium format negatives flat. This works really, really well (I'm actually a bit surprised, given the manufacturer). For 35mm, I'm using the holders from my Polaroid SprintScan 120 (for now...I've ordered some generic Plustek 35mm film holders for use once the Polaroid is sold).
3. Given that I've moved to the D800 (36MP) from the X-Pro1 (16MP), I no longer feel the need to stitch as much as I once did. For a 6x6 frame, I'll generally stitch 2 images; for 6x7, I'll do 4, and for 4x5 film, I'll do 10 (in each case, I give ~20% overlap where possible). This gives me more resolution than I am ever likely to need, and avoids the issue of stitching artifact, which can become problematic the more images you attempt to stitch. For 35mm, I don't stitch at all; 30-36MP is already beyond the point at which additional details can be resolved from 35mm film, especially the 400 ISO film that I'm typically using.
4. Autofocus works! I was expecting to have to manually focus every shot, which of course adds a lot of time to the process. Well, using the Tokina macro, I don't. Autofocus in live view works just perfectly. All of the images you see below were taken using autofocus. This is a significant addition to the workflow, and one that will cut off 5-10 minutes per roll.
Next, let's address the positives of the method.
1. If you already own a DSLR and a macro lens, this method costs you very, very little. Beyond these two things, all you need is a few pieces of software (which, again, you may already own) and a back light source. If you don't own a digital camera and/or a macro, the cost of course goes up substantially, but then again, you're buying a camera, not a scanner; a camera that has other uses, too.
2. The results are really, really good. Here's a 4-image stitch (totaling 46MP) of a Fuji 400H negative, taken with a Mamiya 7 (excuse my son's silly face. It's a phase...)
...and here is a 100% crop of the same 46MP image.
Now, the Mamiya 7 is, of course, one of the best cameras out there (*the* best, in my opinion, with respect to ultimate IQ), so the sharpness here isn't surprising. But still, this is impressive. Grain is clearly evident, suggesting that the focus of the macro lens is spot on. I have *no* complaints about this scan, at all.
3. I have a DSLR and a LOT more desk space. I keep repeating this, but I see this as a major plus. Not only do I have a film scanning workflow that is faster than my old one (and produces better results, too), but I've also got a camera that I can use to photograph my kids, take on vacation, etc. This is of huge value to me, since I don't always want to shoot film just to photograph my son at the park (though I have...).
4. The colour conversions are very good. My major concern going into this method was how ColorPerfect was going to handle the linear TIFF files generated by the D800 and MakeTIFF; there has been some discussion that this doesn't work well. I'm here to tell you that it does. It's absolutely no different than the linear TIFF files generated by VueScan.
The method isn't perfect, and it's only fair that I detail the potential cons.
1. Lack of preview. If you're shooting reversal film (as I do in the summer), this isn't an issue; your preview is the slide. If you're shooting negative film, things are different. When you use a scanner, you can generate quick previews (with rudimentary colour corrections), and then decide which frames to scan at high resolution. With the DSLR method, you pretty much have to scan the entire roll, perform the colour corrections, and then decide which ones to keep. But you've already invested the time and effort (and disk space) to scan. This isn't such a huge deal with medium and large format film, where you're not taking huge numbers of shots per image. But with 35mm, it takes 15 min per roll to digitize, and then about 30-40 min to do the colour corrections and cropping in Photoshop. The solution to this problem is an updated version of ColorPerfect that allows batch processing; as it stands, you've got to do each image manually (although you can recall your previous settings with one click). I'm told that the makers of ColorPerfect are working on this for the next version. We can only hope.
2. Cost. As I said above, this method isn't cheap. That is to say, it's not (necessarily) cheaper than a film scanner. If you want resolution that approximates a true 4000dpi film scanner, you're going to need a DSLR with >20MP *OR* you're going to need to stitch a lot of images together. This necessitates a fairly costly DSLR and/or a very good stitching program like AutoPano Giga. Plus you'll ideally need a macro lens. And a lightpad. And a reasonably powerful computer (which you'd need anyway with a scanner). Altogether, I've paid about as much as I would have for a Plustek OpticFilm 120. Of course, I can use the D800 for other things. And I can scan large format film with it, too. I don't know about you, but I'd *much* rather have the D800 kit.
That's it. That's really all the cons I can think of.
Bottom line: my Pakon is gone and my Polaroid SprintScan 120 is for sale (message me if you're interested). I'm committed to this method, and, needless to say, I highly recommend it.