Camera Scan vs Film Scanner – A Detailed Comparison

I wrote a long article that compares two ways to scan slides, using two types of equipment I had:

- An ArtixScan 4000tf film scanner

- An EOS 7D with a macro lens plus a slide duplicator attachment

 

I actually elaborated the camera scan set up and many of the comparisons and findings 5 years ago (May 2013). At that time, it was only privately documented, but after several requests I now decided to write a full in-depth public article about it all, including detailed conclusions.

 

The crucial question to be answered is: Will a camera scan setup be good enough for high contrast slides?

 

With this post I simply want to share this with you.

 

Here is the article: Camera Scan vs Film Scanner

Here are some of my articles on scanning. (List recently posted elsewhere on P.net)

 

Huge Scanning Job Finished (yet again!) rescan

2015: A scanning Odyssey - Nikon Super Coolscan LS-9000 ED Nikon

CanoScan 9000F vs. CanoScan FS 4000US

Nikon Coolscan LS9000 ED, ICE, and CanoScan F4000US - Part 2 Canon vs Nikon

Thoughts on Theory and Practice of Scanning (Archival/Forensic)

 

I have tried scanning in just about every form. The little add-on tubes, the Repronar, copy stand, flatbed, and have never found any solution for archival scanning that is equal to a dedicated film scanner like the old Nikon CoolScan 9000.

That may be true, but Nikon no longer makes or services Coolscan scanners, nor the associated drivers and software. So going forward, the choice is between consumer-grade flatbed scanners, outrageously expensive Hasselblad scanners, cheap imitation scanners (digicams in mufti), or high-resolution digital camera copy systems.

 

The last alternative is clearly the best, and getting better each year, The quality of digital cameras, starting about 15 years ago. is precisely the reason film is relegated to a few photographic flat-earthers. Consequently film "scanning" is consigned to producing digital archives of existing work, which is not exactly a growing market. If you already own such a camera, you can set up an efficient film copy system for as little as $200. Slides are easy to copy accurately, and in fact you can improve on the exposure and color with little effort. Negatives are just as easy to copy, but harder to process. You can slug your way through using Photoshop, but commercial software like Silverfast HDR makes even that job easy.

The last alternative is clearly the best, and getting better each year,

 

I truly wish I could agree. I have the equipment (copy stands, macro lenses, a Universal Repronar) and have found the camera/lens combination introduces its own problems (distortion, vignette, curvature of field, ...).

The results can be reasonably good and tolerable interms of time spent if you're doing only a few images. GOTO 20,000 images and it's another matter.

 

The standard of "still able to be serviced by the OEM" is too constrictive in these days of "the time is past"

Edited December 7, 2018 by JDMvW

The quality of digital cameras, starting about 15 years ago. is precisely the reason film is relegated to a few photographic flat-earthers.

 

You don't have to insult us!

I have an LS-4000 with a roll feeder and an LS-8000 with all of the film holders, save the rotating glass holder. I could retire happily without exploring new concepts. However I have made it a personal goal to make scanning with a camera viable, even easy to set up and implement, for my own benefit and that of others who missed the scanner "boat." Medium format is somewhat more difficult because there aren't any simple ways to hold large film strips, as there are for 35 mm film.

 

I have no problems with field curvature in my macro lenses, but film curvature is another matter. Mounted slides invariably bulge in the center, unless glass mounted. My MF film has been stored in archival sleeves within hours of processing, and is dead flat. In any case, I can stop down to f/5.6 or higher, an option not available in a film scanner. That not only increases the DOF, but virtually eliminates vignetting. At 1:1 magnification, the image circle is nearly twice that at normal distances, so you're using only the center anyway. With full-time live view and focus magnification, I can focus with much greater accuracy than any reflex system or AF, at the same aperture used for taking.

 

Sorry about the "flat-earther" assignation. There's room for many interests, especially if it's a hobby. I'm from the Flat-Earth Society home base (Iowa), and I've never found them to be easily insulted, much less persuaded. They're sure they're right, but always straining to explain away the facts.

Sorry about the "flat-earther" assignation. There's room for many interests, especially if it's a hobby.

 

Apology accepted!

 

Hmm... I'm always insulting

Vincent, even though he

withholds my bananas...

http://bayouline.com/o2.gif

If you already own such a camera, you can set up an efficient film copy system for as little as $200.

 

- As much as that!?

If you shop around you can halve that easily.

Here's the kit I've been using:

 

Item 1, as you can probably read, is an Astron slide/film copier attachment. I picked it up for £10 (~ $15 US). It does need a macro lens to screw onto the front of though, and the film-holder arrangement isn't great.

 

Item 2 is an obsolete Nikon ES-E28 attachment, with the useless 28mm screw thread bored out and a 42mm camera flange fitted. It then accepts item 3, a set of cheap M42 extension tubes plus Pentax thread reversing ring.

 

The ES-E28 cost me another £10, but came complete with multi-frame filmstrip and slide holders. The camera flange came from a scrap Zenith E, bought for £1, and the extension tubes and M42 reverser cost about £5 as a bundle.

 

So, total outlay about £26 (say $40) and a couple of hours labour.

 

I was going to post comparison 'scans' from using the above on a 24 megapixel Nikon D7200 against a dedicated film scanner. However, when I dug out the film scanner I discovered the drivers weren't compatible with anything more recent than Windows 7, and the scanner was no longer supported by the maker - Pacific Image.

 

I did get it working - sort of - but it was far slower than I remember. The framing and focus were off, and the resulting part-frame scan had awful colour. So much for a fair comparison!

 

Anyway, FWIW here's a digital camera scan from an old colour negative.

I didn't even notice medallion man on his balcony when I took the picture.

Edited December 7, 2018 by rodeo_joe|1

"flat-earther"

 

The way I figure it, it's just that everything is downhill from everywhere else.

There's no substitute for a good macro lens. That's the heart of the system. I bought a 55/2.8 Micro-Nikkor for $120 and a PK-13 extension ring (for 1:1 magnification) for another $25. I already owned a Sony A7ii and a Nikon lens adapter. The slide attachment brought the cost over $200, but not by much.

 

Slide copiers disappeared for a long time, but have enjoyed a recovery in the last couple of years. The Nikon ES-1 was the first one I saw that didn't come with a cheap +10 diopter built in. It was not compatible with common film strip holders, so I used one that came with my LS-4000. Those go for about $200 on the use market, but are only as thick and wide as a cardboard mount.. The Nikon ES-2 comes with holders for both slides and film strips (up to 6 frames) - bingo!

 

The ES-1 fits a 52 mm filter ring, and provides the right working distance for a 55/2.8 AIS micro. The ES-2 is designed around a Nikon 60/2.8 Macro, with extension for the AF-D and AF-S versions. The 52 mm adapter was designed for a 40 mm Macro, and didn't have enough extension for the 55. I took care of that with a 52-62 filter adapter.

 

The whole assembly is solid enough that I don't need a tripod, nor to make adjustments between slides. I was pleased with the results using a 24 MP A7ii, but 42 MP using an A7Rii was an order of magnitude better, and definitely grain-sharp.

 

One of the gadgets you list is a reversing ring. Was that for the lens, or to make something fit to the lens? Reversing the lens doesn't help when you're near 1:1, and it's an awkward substitute for the right extension tube.

One of the gadgets you list is a reversing ring. Was that for the lens, or to make something fit to the lens?

 

- The reversing ring is needed to adapt the M42 extension tubes to a 52mm filter thread. The tubes are then screwed to my 55mm f/3.5 Micro-Nikkor + M tube, and the converted ES-E28 is screwed to the other end of the extension tubes.

 

I would have adapted a set of 52mm filter-thread tubes directly to the ES-E28, but they're not so easy to come by. Only Nikon's K tubes fit the bill I believe, and they're not cheap.

 

By trial-and-error, I discovered that extension tubes 3+4 are needed when using a DX camera for copying, and only tube 2 for use with a full-frame camera. The thinnest tube, number 1, allows 'zooming in' for a slightly cropped framing.

 

The ES-E28 also came with a 6 frame filmstrip holder and a 2 frame slideholder. The setup sounds pretty much identical to your ES-2 arrangement.

 

For lighting, I fit a speedlight to the hotshoe of the camera and point the whole thing at a white-painted wall. There's actually enough power in the camera's popup flash to do the job, but it runs the camera battery down quite quickly.

 

Focussing is done by LiveView with a small LED light pointed at the copier diffuser.

 

FWIW, I also have an f/2.8 55mm Micro-Nikkor, but found it inferior to the f/3.5 version in this application.

 

P.S. There's a Polaroid branded device on sale that appears functionally the same as an ES-2. I believe it contains a 'macro' dioptre, which I think can be removed.

Edited December 8, 2018 by rodeo_joe|1

I truly wish I could agree. I have the equipment (copy stands, macro lenses, a Universal Repronar) and have found the camera/lens combination introduces its own problems (distortion, vignette, curvature of field, ...).

 

- But those exact same complaints could be levelled at an enlarger lens. The fact that they weren't, over a period of several decades, shows that they're not an issue at all.

 

And what makes you think a scanner lens is devoid of those defects? Sure, the distortion and vignette may only be seen in one direction, but I'm pretty sure it's detectable if you really look for it.

And what makes you think a scanner lens is devoid of those defects? Sure, the distortion and vignette may only be seen in one direction, but I'm pretty sure it's detectable if you really look for it.

A point I omitted for the sake of brevity. However a scanner lens is optimized for a single distance (give or take a mm or two), FOV and aperture. Just attaching an enlarger lens, designed to fit a lens board, would require a bit of cobbling. Life is too short to re-invent every wheel.

 

I have a set of Nikon K extension tubes, which weren't expensive at the time. Like any obsolete yet desirable equipment, I'm sure scarcity and demand have driven up the price. As it turns out, they aren't needed for either the ES-1 or ES-2. For the latter, a readily available 52-62 filter adapter was sufficient.

Edited December 8, 2018 by Ed_Ingold

However a scanner lens is optimized.....

 

- Hopefully! But having seen the results from some scanners, I do wonder.

I was thinking of Nikon and Hasselblad scanners, not Plustek. In the near drum scanner category are Scitex flatbed scanners, used for large format and graphic arts. You pay handsomely for that kind of quality. By comparison, Nikon scanners were a bargain.

 

I have an older Epson with a real lens, which promised 1600 ppi, but delivers about 1300. It cost something like $1400 nearly 20 years ago, before flatbeds got dumbed down.

I was thinking of Nikon scanners, not Plustek, also Flextite and drum scanners. and Scitex scanners, used for large format and graphic arts.

I was thinking of Nikon scanners, not Plustek, also Flextite and drum scanners. and Scitex scanners, used for large format and graphic arts.

 

- I was looking more in Epson's direction.

 

I had a Plustek 7200i for a couple of days. 7200 ppi? No way! But an acceptable 3600 ppi.... if you've got several hours to spare for each roll of film you want to scan. Plus time spent tweaking the colour to something halfway-decent.

I apologize for being so anecdotal. It would be far better to make direct comparisons. Sadly, I have very few MF scans on my computer. 15 years ago, disk space was at a premium, and each 6x6 scan at 4000 ppi, 16 bit TIFF, was over 216 MP. It would be a major effort to revive my LS-8000, at least in the busiest time of year for my business. It's an important issue, and I plan to be back later with examples.

 

I have MF film handling for copying pretty well in hand. In lieu of a copy stand, I use a Novoflex focusing rail with a slide holder attachment. It's a solid, well aligned setup, but the film is not well shielded from ambient light, which must reduce the contrast. I'm half serious about digging a LF darkcloth out of the closet and draping it over the setup. Or I could do it in the dark of night (too many windows to block). Another possibility is to use my Hasselblad with a bellows attachment to do the copies. It's only 16 MP, but 4080x4080 pixels is comparable to a similar cropped from a 24 MP 35 mm format. However that would suggest. using an $11K setup (used price) to do a lesser job than a $3K scanner.

That may be true, but Nikon no longer makes or services Coolscan scanners, nor the associated drivers and software. So going forward, the choice is between consumer-grade flatbed scanners, outrageously expensive Hasselblad scanners, cheap imitation scanners (digicams in mufti), or high-resolution digital camera copy systems.

 

The last alternative is clearly the best, and getting better each year, The quality of digital cameras, starting about 15 years ago. is precisely the reason film is relegated to a few photographic flat-earthers. Consequently film "scanning" is consigned to producing digital archives of existing work, which is not exactly a growing market. If you already own such a camera, you can set up an efficient film copy system for as little as $200. Slides are easy to copy accurately, and in fact you can improve on the exposure and color with little effort. Negatives are just as easy to copy, but harder to process. You can slug your way through using Photoshop, but commercial software like Silverfast HDR makes even that job easy.

 

My CS9000 is working fine, and I'm running Nikon scan under Windows 10.

If you have questions about my setup, as shown in the article, or questions about the images I posted, just ask here.

I've been "scanning" using a Rodenstock APO-Repro lens optimized for 1:1 and a slide copying system using a dedicated film copying system for years. There is just no comparison between it and conventional scanning. Superior in every way, except dust removal, but the speed and better quality of final copy more than compensates. Just make sure to remove dust before copying. Film scanning using conventional scanners was, for me, about the best way to destroy one's interest in photography.

Not wishing to sidetrack this thread, but there may be some advantage to using an APS sized sensor for copying, rather than a full-frame one.

 

Reasons? Depth-of-field and diffraction.

 

A full-frame camera requires at least 1:1 magnification, with a consequent loss of 2 stops, or a doubling of effective aperture number. So a marked aperture of f/5.6 becomes f/11, and with the same diffraction limit as f/11. However, the depth-of-field stays at the razor thin 1:1 magnification and f/5.6 level.

 

Changing the format to APS eases the situation considerably. The reproduction ratio drops to 1:1.5 as does the aperture number multiplier. With the same marked f/5.6 aperture, the effective aperture and diffraction are at f/8.4 levels; while depth of field increases by nearly 1.5 times in real terms.

 

Of course this assumes that both sensors have the same number of pixels, which may or may not be relevant as long as diffraction isn't the limiting factor.

Not wishing to sidetrack this thread, but there may be some advantage to using an APS sized sensor for copying, rather than a full-frame one.

 

Reasons? Depth-of-field and diffraction.

I am sorry to say that this is most probably not really true.

Theoretically, given ideal lenses and ideal sensors, APS-C and full-frame have identical performance at long distances if you adjust the aperture accordingly – it all scales. This means you need to stop down the full frame lens with a factor of the focal length multiplier (just over one f-stop) to obtain identical depth of field and diffraction and resolution. At 1:1, however, things get more complicated.

 

If I enter values into the exact equations for depth of field (as derived by David Jacobson at photo.net long ago), I get:

With APS-C (your Nikon sized 24 mm sensor width, i.e multiplier 1.5), using your suggested f/5.6 aperture, 36 mm objekt width, and assuming an acceptable circle of confusion that would correspond to 2 pixels on a hypothetical 20 Mp camera, I get a depth of field of 0.37 mm. If I do the same thing on a full frame camera, using f/8, I get depth of field of 0.42 mm! Quite surprising that it is microscopically larger, but that is what the equations tell me (assuming a pupil magnification of 1). Focal length does surprisingly not matter in the first 5 decimals of the DoF.

 

At longer distances, above 220 mm object width, the full frame gets shorter DoF, and at long distances you need to stop down to f/8.5 to get the same DoF (which is expected).

 

Diffraction, however, seems to be more complicated, since I think the formulas I have are only valid at long distances. I have a feeling that full frame may be at a slight disadvantage here for 1:1, about 20%, but I cannot prove it.

 

All-in-all, however, there is basically no significant difference between an APS-C and a full frame as regards DoF and diffraction, provided you stop down the full frame about one f-stop.

 

I personally think f/5.6 is too wide on APS-C unless the slide is glass mounted – there is substantial risk that the frame edges become unsharp. I suggest f/8, which means f/11 on full-frame (just over one half mm in DoF).

 

There are still a few advantages with APS-C: Since 1:1 on APS-C is a much smaller object than on a full frame, I can easily do a factor 1.6 crop of a 36 mm slide with the macro lens I have! That's nice! And the whole setup is relatively small and stable, and the lens is considerably cheaper and lighter than the corresponding quality for full frame.

Well of course you get more depth-of-field at f/8 than you do at f/5.6.

 

I was comparing like-for-like, not apples to oranges.

 

I'm not familiar with Mr. Jacobson's formulae for depth-of-field, only the ones I've formulated myself, which take careful account of physical aperture size and tangential half-angle. The 'accepted' circle-of-confusion is, of course, fairly useless for this purpose, but it scales to the pixel level quite readily - simple geometry of similar triangles.

 

However you cut it, a smaller magnification at a fixed aperture and with the same focal length must give a greater depth-of-field. Even allowing for the higher magnification needed to get the smaller image to the same viewing size. The depth-of-field increase is approximately inversely proportional to the difference in sensor size. Even more so if the copying lens focal length is scaled with sensor size.

 

It's diffraction that may or may not make any difference. The effective f-number is multiplied by 1+m, as is the Airy disc diameter, doubling the marked aperture number for 1:1 reproduction and multiplying it by 1.667 for a 1:1.5 APS sensor. So in this case the larger sensor wins out - slightly.

 

I can easily see the difference in sharpness due to diffraction when stopping the copying lens down from f/5.6 to f/8 (marked aperture) on a APS camera. However, with an APS sensor the f/5.6 aperture is fully useable, with enough depth-of-field to take care of some degree of bowing in the film.

 

There are advantages and disadvantages to both full-frame and APS sensors in this application, but on balance I suspect, for a given megapixel count, that the advantage may go to the smaller format.

 

I shall have to check the theory empirically, but comparisons so far seem to show that a full-frame sensor doesn't show any real advantage over APS, and that using a wider aperture and shorter focal length copying lens on APS may well result in improved copies.

 

I've also shown to my own satisfaction that increasing the copying 'resolution' beyond 4000 ppi doesn't get any more detail from the 'average' negative. Well, actually from a carefully focussed and exposed negative shot on T-max 100.

Not wishing to sidetrack this thread, but there may be some advantage to using an APS sized sensor for copying, rather than a full-frame one.

 

Reasons? Depth-of-field and diffraction.

 

A full-frame camera requires at least 1:1 magnification, with a consequent loss of 2 stops, or a doubling of effective aperture number. So a marked aperture of f/5.6 becomes f/11, and with the same diffraction limit as f/11. However, the depth-of-field stays at the razor thin 1:1 magnification and f/5.6 level.

 

Changing the format to APS eases the situation considerably. The reproduction ratio drops to 1:1.5 as does the aperture number multiplier. With the same marked f/5.6 aperture, the effective aperture and diffraction are at f/8.4 levels; while depth of field increases by nearly 1.5 times in real terms.

 

Of course this assumes that both sensors have the same number of pixels, which may or may not be relevant as long as diffraction isn't the limiting factor.

 

And yet many people engage in macro photography with full frame cameras. Are they all misguided? Would they be better off using the cameras in their phones?