Avoiding blurry edges with slide duplicator + macro lens

Question

I have just bought a Kaiser Digital Slide Duplicator in order to digitise 35mm slides (mostly positives), since this is something I need to do fairly often.

In general, I have seen lots of people advising the use of real macro lenses for this type of use, since they have a flat field and that should help avoid blurring around the edges; the answers in this DPReview thread, for example. I use a full-frame Canon 5DS R and already have a Canon 100mm 2.8L IS USM macro lens, and since the product page for the duplicator says full-frame sensors should be used with lenses of about 80–120mm to get full-size reproductions, I figured that should be a perfect match.

But the results so far have been… disappointing, to say the least. Horribly blurry edges. The narrower I set the aperture, the less noticeable it gets, but even at its very narrowest (32), there’s very noticeable blurring. When taking regular pictures, the lens is very sharp all the way out to the edges, so it’s not a general lens issue.

I note that the Kaiser product page actually – and rather mysteriously to me – says, “Not suited for use with macro lenses”. I can’t figure out how this statement could be true of a duplicator (how can a flat field be a disadvantage?), but then I’m dreadful at physics, so perhaps there’s a logical reason?

Here is an example of a slide shot at F11 (first) and F29 (second). You can see that F29 is better, but still not in any way usable (ignore the pink – there’s a thin sheet of plastic with pink writing on the back of the duplicator’s diffusor, and I haven’t taken it off in case I want to return the duplicator):

I tried taking off the “Detachable achromatic 10 dioptre macro lens”, just in case that was the issue, but that only, unsurprisingly, made the lens completely unable to focus, so clearly it wasn’t.

To test, I’ve tried with the only non-macro lens I have that goes to between 80 and 120mm and has a diametre I can work with (I have a 67–52mm stepdown ring and a 52–58mm step-up ring), which is a 70–200mm F4.0. Not sure if it’s the step-down ring, but the result is heavily barrelled and no sharper (at F32):

Is there something I’m missing here? What does it mean when Kaiser state their duplicate is not suited for macro lenses, when macro lenses are generally exactly what you want for duplicators? Is there something I can do to achieve slides that are actually in focus, or is this just the wrong product for my lens and body?

Answer 1

Most optics are weak around the edges, including the +10 diopter included with the slide duplicator. The purpose of the diopter is to allow the slide duplicator to work with non-macro lenses. But when the diopter is removed, the slide is too close to focus with a macro lens.

To avoid the edges when using the diopter, switch to a ~55mm lens. This is basically what happens when the slide duplicator is used with an APS-C camera. (On full frame, the edges of the frame will be "black", outside the slide boundary.)

To use the slide duplicator without the diopter, increase the distance between the macro lens and slide. Inexpensive extension tube sets have segments with 60mm threads, which can be adapted to the lens and slide duplicator using step-up/down rings. (Another option is to try an M58 helicoid, but I don't know if the thread pitch is the same as filter threads.)

If you opt to add an extension tube between macro lens and camera, the magnification ratio will be increased. You will be able to capture only portions of the slide. This may be acceptable if you are willing to stitch the pieces together.

Another type of slide duplicator has a built-in lens. I have seen them only for full-frame. When used with APS-C cameras, they will crop out the edges of the slide, but that isn't an issue for you.

Answer 2

I'm currently digitally duplicating my large old slides collection with a different, home-brew setting:

• Laptop screen as light source (showing a 100% white image).
• Slide placed some 40 mm above the screen, to de-focus the screen's pixel structure (glued together a cardboard slide holder for that purpose).
• Low ambient light (screen covered outside the scanning area).
• Place the camera so that a portrait slide nearly fills the sensor height (my lens/camera config: EOS M, 18-55 kit lens plus 10 mm extension tube between body and lens, zoomed to nominally 32mm focal length).
• Using f/11 gives enough depth of field to account for non-planar slides.
• Auto-focus works well.
• Manual exposure adjusted to nicely fill the histogram.
• Always use the slides in their natural orientation, so no rotation is necessary in post-production.
• A quick-and-dirty post processor program does portrait/landscape detection and automatic 3:2 cropping, as well as a fixed 1.5 gamma correction.

The resulting roughly 2000*3000 pixels capture my slides' resolution well enough. A few slides need manual care, but the majority turns out quite usable. Yes, I'm wasting lots of megapixels, but my slides don't need more resolution.

To my (positive) surprise, the kit zoom lens delivers a sharp image, even used outside its spec, with the extension tube. Of course it helps that with my setting, I'm staying far away from the corners.

With this setting, I'm able to do about 200 slides per hour.

Cost factor: I had to buy an extension tube set (30 EUR).

Your 100mm macro lens should be able to do the same, without distance rings.

Regarding ambient light: I found it easier to dim down the lights and have free access to the slide holder than to build a light shielding.

Answer 3

Not really an answer, but way too long for a comment.

Here's the thing about macro lenses: Most of them are optimized to give the flattest field of focus at minimum focus distance. The flat field correction used to give best performance at MFD means at infinity focus the lens will not necessarily have as flat a field of focus.

When using extension tubes with macro lenses, results will generally be better with using the macro lens focused at MFD and using only enough extension to give the increased magnification desired, rather than using a longer extension tube with the macro lens focused at infinity.

But even then, we can't expect too much from flat field correction. Even in theory, the best flat field correction can do with multiple wavelengths of light is give a field of focus that looks more like a lasagna noodle than a flat plate. And sagittal and tangential correction probably won't be shaped exactly the same, either. That's why we still need to stop down to make a two dimensional subject sharp all the way out to the edges.

---

Focus distance, by definition, is always measured from the image plane (film/sensor). Working distance is measured from the front surface of the lens.

Assuming the front of the lens is exactly half the way between the focal plane and the subject only works with a single element (theoretical) thin lens with zero thickness. For compound lenses with multiple elements, the front of the lens is not usually at 2X focal length from the imaging plane. So the working distance will not necessarily be the same as the distance from focal plane to front of lens, nor will working distance equal 2X focal length. Subject should be 4X FL from the focal plane (with optical center of lens at 2X).

1:1 is a reproduction ratio. That is, the image of subject projected onto the focal plane is the same size as the subject. For a lens focused at infinity, the optical center of the lens (the point at which a single element theoretical thin lens would need to be to focus on infinity) will be at the focal length . For 1:1 reproduction, the optical center of the lens needs top be 2X the focal length from the focal plane, and the subject needs to be 2X the focal length from the optical center of the lens. Focal plane to subject distance equals 4X the focal length of the lens.

Answer 4

Because this is a periodic / semi-frequent process for you, I recommend upgrading to better equipment that is known to produce decent results.

I like to use a Nikon PB-4 bellows with the PS-4 slide copying adapter, in conjunction with the Nikon Micro-Nikkor Auto 55mm 1:3.5 lens. Total cost for the PB-4, PS-4, and 55mm 1:3.5 lens was less than $200. The Nikon Micro-Nikkor 55mm 1:3.5 is a fantastically cheap, flat field-of-focus lens that was often used for 1:2 to 2:1 or so reproduction work, in conjunction with the bellows.

I'm not familiar with Canon bellows systems, and older Canon reproduction lenses. But luckily, older F-mount Nikon bellows and lenses can be adapted with no problem onto Canon EF and EF-S bodies. Because you're using a Canon full-frame camera, you can adapt either the entire Nikon bellows+lens setup to fit your Canon with a Nikon lens–to–Canon body adapter, or use a Canon bellows and slide copy rig and just adapt the Nikon lens to your Canon bellows. The same adapter would work.

Note that perhaps the Nikon PB-6 bellows might be a better option, giving slightly more space between the camera body (especially the handle/battery grip, and the pentaprism) and the metal bellows body. See this answer to Differences between the PB-4 and PB-6 bellows for my Nikon camera? If you go with a PB-6 bellows, make sure to get the PS-6 slide copy adapter, rather than the PS-4 slide copy adpater (which only works with the PB-4).

I recommend a bellows + reproduction lens setup because you can control the body-to-lens distance, and lens-to-slide distance, with the bellows system. When used in conjunction with a macro lens that was intended for reproduction (with a flat field-of-focus), you can get very good results corner-to-corner when digitizing slides. The added bonus is that if you upgrade your camera to a higher resolution body in the future, as long as you can adapt older Nikon F-mount lenses to the new body, the copy rig will continue to work with your newer high-resolution camera body.

Additionally, a bellows system, much like the Kaiser slide copying adapter you are using now, completely blocks all incident light coming into the lens from in front of the slide. But again, unlike the Kaiser adapter, you can completely control the magnification ratio by controlling the camera-to-lens and lens-to-slide distances, separate from only using the lens's focus control.