Depth of Field for scanning 35mm slides on flatbed scanners (Epson)

Question

I have both an Epson V750 and V850 scanners and am trying to digitize a large amount of slides.

The software for scanning does great for color restoration (slides dated from 50s on) but the issue is focus. Slides are from film that ranges from PolaChrome, KodaChrome through Agfa and others. Mounts range from "factory cardboard" through "factory plastic" through after-market snap slide mounts.

I mention all this because all these factors seem to impact slide position in a scanner as far as focus and DOF (differing thickness, clamping, centering) as well as how "warped" the image is in the mount.

I used a set of 12 varied images for some testing in both and found that the DOF in both scanners appears to be less than the thickness of two layers of film. I can find no way to enlarge that and that indicates to me that this is way more of a issue than I had hoped. If that is true, then I think my only option is to create a test matrix for each scanner that cross references - film type - film age - mount type - film warp amount - scanner holder setting - and then use that to scan based on those parameters.

Am I missing something?

Just to scale the "job," this is for 22,000+ images accumulated over the last 60+ years

---

I had used the 750 several years back for a 200+ slide effort for a friend and even with the "Unsharp Mask" the results were not up to the originals but worked for the low resolution video we produced, so the base question still remains, what can be done to improve the DOF for the scanners? If nothing, then I will again abandon the scanners. As far as dedicated film scanners (Nikon-Coolpix/etc.), I have tried at least five over the years and all with limited success (jams every night, no batch film warp compensation, batching scans by mount, film type, age, etc. creating a post scanning organizational issue) for any volume.

Since these are great slides with considerable detail, I have recently invested in a 50+mp Canon with a top of the line Macro Lens and tried to duplicate that way, only to have to send the body in for maintenance as the focus prism and the film plane did not match. So am awaiting return of that for a second round of tests. Did not want to bring that into the scanner topic but thought I would respond to BobT.

The Epson scanner does such a great job on initial scan correction range, but if the focus is still going to be a hit and miss based on the limitations of the technology, then I really would like to "put that to bed" since it appears to be unsolvable at volume.

I appreciate the input of others, but if the scanners are "doing their best" and that is not acceptable, then I will put my efforts on the the camera (to achieve the best focus) and eat the time for "all the corrections" needed for the "age range and variations" of the originals. I will post again when the body comes back and I can run yet another test.

Answer 1

I digitized 6000+ slides in early 2018 which were mounted in a variety of cardboard and plastic mounts. The slides dated as far back as the early 50s, with many of that age exhibiting significant curvature.

A number of options were considered, but I couldn't readily find any commercial service that offered comparable quality to what is achievable with modern photo sensor technology. Many are still using scanners that were state of the art in the early to mid 2000s, but are now entirely obsolete. Frankly, I was also unwilling to entrust the sole copies of these photos to Fedex.

Ultimately, the best solution in my case was something similar to this: https://www.pteavstudio.com/forums/topic/24303-digitizing-slides-with-a-kodak-slide-projector/. I bought a cheap USB relay board for about $10 and made a small Python script that would trigger the relay to advance the carousel, then emulate a keypress to trigger a capture in a tethered Lightroom session (tethering is vital for keeping track of as much metadata as possible when you have that many files).

I threw a few layers of diffusion gel between the projector lamp and the slide to ensure even illumination. Make sure to also take a blank frame as a reference for vignette correction. I also experimented with using a hotshoe flash, but using the built-in projector lamp was the only way to get reliable AF, and it produced more consistent exposures in terms of metering and color temperature.

The throughput ended up being pretty good - about 1.5 s per slide when running, most of which was unattended. One issue that cropped up every 2-3 reels was that some of the cardboard mounts would jam in the carousel, which resulted in a mix of duplicate and skipped frames. YMMV, particularly if you have a newer projector.

I'm not sure this is cost-effective unless you already own a 24+ MP camera, and ideally a macro lens (though you can rent or buy/resell one fairly cheaply for this). It's also obviously a lot more work than any commercial service would be. That said, the results are spectacular, and significantly above the specc'ed/showcased performance of any service I found. With a modern (2016ish or later Sony sensor—I used a D600), you can generally out-resolve the slide film media in terms of both sharpness and dynamic range. I think slide film still wins on color depth, but not noticeably.

I would estimate that doing all 6000 slides took around 120 hours of human time. One thing that was particularly difficult and time-consuming was getting the alignment just right and keeping it there between sessions - a geared xy stage (ideally xyz) and a sturdy base or tripod on a hard surface are mandatory. You'll also may want a Giottos blower to get as much dust off the slides as possible unless they're immaculate and you're essentially doing this in a clean room.

For upwards of 20k slides, I would see if you can find a local service that's using one of the commercial SlideSnap rigs or equivalent, though it won't be cheap. I consider drum/flatbed scanning to be an unacceptable loss in quality for this application, and I would really recommend against settling for that.

Answer 2

Anti Newton glass and scan flat on the scanner bed. You'll have to demount the slides of course. Sorry, it's the only way if you want edge to edge clarity. For most folks it isn't necessary, but if you're dedicated or picky, that is the way to go.