DIY DSLR Scanning Rig Design

[Jesse3Names]

Well, as I have decided to purchase a Wanderlust Travelwide (whenever they ready the next batch) to shoot 4x5" film, I need to be able to affordably scan the negatives. The problem is, I'm pretty much broke. That's what accepting your fate as a Master's student for a few years does to you. So I can't afford to pay for each scan of each negative to make sizable enlargements or for a proper scanner, e.g. the Epson V700. This applies to my 35mm film and 120 formats if I rig up a medium format back for the Travelwide, as I'm considering.

After taking a look at how some people are doing this, I knew I needed to use my Canon 50mm f/1.8 II lens - it's cheap and better yet, I already own it. With the appropriate macro extension tubes (trial-and-error on tube lengths to get close to 1:1 magnification with my full-frame 6D body) I should be able to take high-quality RAW samples of the negatives and stitch them with PTGui Pro (amazing panoramic software I already own, as well). I know I need to control lighting source and color temperature and correct for lens distortion in Photoshop with a lens profile. To control lighting, I will probably drape a black cloth over the rig and to control color temperature, I will use the 4x5" 5000 K color balanced bulb and calibrate carefully with a gray card shot/setting WB temp. to 5000 K in Magic Lantern firmware, whichever gives better results. To prevent Newtonian rings from appearing, I plan to use a BetterScanning.com anti-Newtonian ring glass plate - they sell 8x10" sizes so I'm sure I could request a 4x5" cut of the glass.

The basic idea of my rig, as pictured below, is to have a platform holding a 52mm (male) to 82mm (female) filter adapter upside down. The 52mm male end of the filter adapter will screw into the female filter threads of the 50/1.8 lens at ~1:1 magnification. The platform will fix the distance between the lens and the film plane on the light box, be on omnidirectional ball casters that won't bind up like traditional furniture caster, and should allow for easy framing of the negative sample in Live View. I have access to a desktop CNC mill to allow for me to recess the 82mm female end of the filter adapter into the lens board (likely will be made of 3/8" thick HDPE I have for another project) so the camera will screw in and be stable. I apologize for the lack of scale in the drawing - it was a very quick sketch-up to get my concept down on paper to improve upon later. The wood frame supporting the lens board and subsequently the camera need only be as wide as the short dimension of the light table plus 2*(1/2 the short dimension), in this case that's only 8" long over the 4" wide light table. That's just to be able to easily sample all the way to the edge of a full-sized 4x5" negative. Do any of you scanning veterans have any suggestions to improve image quality/calibration/rig design of this method? I have experience in panoramic photography, but this is nothing more than applying a lens correction profile, exporting to 16-bit TIFF files, and stitching in PTGui Pro. I understand scanning is another kind of art, so I'm trying to get at the max resolution of film without introducing any significant new technical challenges into the process... like learning the intricacies of a scanner and film holders.

 

[KM-25]

It's an interesting concept if not a bit rough and by the sounds of it you should be happy with the results. That bieng said, there are some introduced bugaboos that are specific to how you are going about it.

Hopefully what I am currently working on to release in the near future as an actual product will reveal what those are. It should basically permanently quiet the constant complaints that there needs to be scanners made, it's pretty damn amazing, especially the associated software...

 

[Jesse3Names]

It's an interesting concept if not a bit rough and by the sounds of it you should be happy with the results. That bieng said, there are some introduced bugaboos that are specific to how you are going about it.

Hopefully what I am currently working on to release in the near future as an actual product will reveal what those are. It should basically permanently quiet the constant complaints that there needs to be scanners made, it's pretty damn amazing, especially the associated software...

Yeah. My idea wasn't to make it compatible with all different camera & lens combinations, but instead just make it work for my application. Simple, but does the job. So, this product you're soon to release - need a beta tester? Please keep me in the loop, I'm not in a rush to get this built, so if you release something significantly better I may end up going with that for a fraction of the cost of a scanner, I would hope.

 

[Dwig]

I would be prone to making a few changes:

1. Use a 50-60mm macro lens instead of a 50mm f/1.8. The fast 50 will not yield the best IQ when used a 1:1. it will also have more spherical distortion than a good macro and removing this distortion in the stitching process will lower IQ somewhat. If budget prevents using a good macro then plan on cropping the each image somewhat (e.g. use an APS-c or m4/3 body with a FF lens) before stitching.

2. Rather than a system using free movement, I'd opt for a fixed 2 axis arrangement. One set of rollers running in a track for the X axis and a slot/trench/wide groove for the square lens board to move along the Y axis. The current design doesn't prevent the camera from being rotated. A slight rotated image can be aligned when stitched, but at a modest cost in IQ.

 

[Jesse3Names]

I would be prone to making a few changes:

1. Use a 50-60mm macro lens instead of a 50mm f/1.8. The fast 50 will not yield the best IQ when used a 1:1. it will also have more spherical distortion than a good macro and removing this distortion in the stitching process will lower IQ somewhat. If budget prevents using a good macro then plan on cropping the each image somewhat (e.g. use an APS-c or m4/3 body with a FF lens) before stitching.

2. Rather than a system using free movement, I'd opt for a fixed 2 axis arrangement. One set of rollers running in a track for the X axis and a slot/trench/wide groove for the square lens board to move along the Y axis. The current design doesn't prevent the camera from being rotated. A slight rotated image can be aligned when stitched, but at a modest cost in IQ.

In my first sketch of the rig, I had a complicated, near scaffolding-like, four-posted, anchored tower with tracks for 2 directions of guided movement. But I quickly trashed that idea because I didn't like the rotational forces it would have put on the tower when I mounted the DSLR. When I decided to mount the camera using the filter threads instead of the tripod screw (removing any rotational forces that would shake/reposition the camera), I totally forgot about rotation! Agh, rookie mistake - I've done enough panoramas to know how much one little tweak to an image or two can ruin an entire panorama, or at least make it near-impossible to stitch properly.

So in v2 of my DIY rig, which I quickly have drawn up and posted below, a lens board with the same 52mm male to 77mm female filter ring sits on cylindrical rails with 90 degree-separate wheels. The boards the rails screw into on each end are ultimately on the same type of rails that are fixed together around the light table in a box structure to keep things in 2 dimensions. If I didn't attach the lower rails to each other, it'd be extremely difficult to set up each time. Plus this isn't exactly a large rig, so I can spare some extra materials for ease of use.

If I use a 1:1 macro lens, I can image a 35mm slide to 20 MP (the resolution of my full-frame DSLR) and a 4x5" negative "scan" would result in 101 MP. I like the Canon 65mm f/2.8 1-5x macro lens, but do you know if the lens' magnification ring would retract under the downward weight of the camera as it's mounted on the filter threads and the DSLR body is unsupported otherwise?

 

[lukitas]

If I get this right with my decrepit neurones, It would be easier to get the precision needed if your camera were fixed, and you move the negative holder on a sled. The camera is a lot heavier than the negative, so it should be easier to control the movement of the negative, than to keep the camera within bounds while moving. Also, the area of even light can be smaller : you only need to have the area your camera can see to be evenly lit.

When camera and light source are lined up, any irregularity in the light source can be compensated for by taking a picture of the raw light source, and using that as a mask to even out exposure.

The hardest job will be to get the negative to move in a perfectly horizontal plane. At 1:1, thousands of millimetres make a difference : you could get the dust on the back of the negative perfectly sharp, whereas the emulsion is just out of focus (if you wanted to suppress grain, this would not necessarily be a bad thing). Getting your sled to run true will be the most difficult part of this project. You don't want any dust on you rollers or your rails.

I whole-heartedly second the suggestion to get a serious macro lens. For the moment, and for the lack of the wherewithal to acquire a nice micro Nikkor, I am using the kit zoom on a D3100, with a macro ring, and the corners aren't quite as sharp as the center. The Nikkor 50 f1,4 is even worse. A macro lens is corrected for a flat field at close distances : if your corners are better, you need less overlap, less shots to stitch together. You might get better results if you turn your lens front to back, but then you would have a fixed focusing distance, turning the barrel will not move the optics in this set-up. You would have to calculate the exact distance of extension needed for 1:1 magnification.

I wish you good luck, and I hope to see how you get there. I might steal some ideas off you. There's a 20-year old stack of 6x6's waiting to be scanned.

 

[TXForester]

For a macro lens, you could use the Olymus OM mount 50mm f/3.5 macro. I use one on a Canon dslr and have used with a cheap 35mm slide copier. My results were decent considering I could not flatten the slide, control white balance etc. Ought to do well with your rig design. They can be had for under $100 and another $15 for a focus confirmation mount adapter.

 

[aizan]

we salute your badassery, Jesse3Names and KM-25!

 

[Jesse3Names]

If I get this right with my decrepit neurones, It would be easier to get the precision needed if your camera were fixed, and you move the negative holder on a sled. The camera is a lot heavier than the negative, so it should be easier to control the movement of the negative, than to keep the camera within bounds while moving. Also, the area of even light can be smaller : you only need to have the area your camera can see to be evenly lit.

When camera and light source are lined up, any irregularity in the light source can be compensated for by taking a picture of the raw light source, and using that as a mask to even out exposure.

The hardest job will be to get the negative to move in a perfectly horizontal plane. At 1:1, thousands of millimetres make a difference : you could get the dust on the back of the negative perfectly sharp, whereas the emulsion is just out of focus (if you wanted to suppress grain, this would not necessarily be a bad thing). Getting your sled to run true will be the most difficult part of this project. You don't want any dust on you rollers or your rails.

I whole-heartedly second the suggestion to get a serious macro lens. For the moment, and for the lack of the wherewithal to acquire a nice micro Nikkor, I am using the kit zoom on a D3100, with a macro ring, and the corners aren't quite as sharp as the center. The Nikkor 50 f1,4 is even worse. A macro lens is corrected for a flat field at close distances : if your corners are better, you need less overlap, less shots to stitch together. You might get better results if you turn your lens front to back, but then you would have a fixed focusing distance, turning the barrel will not move the optics in this set-up. You would have to calculate the exact distance of extension needed for 1:1 magnification.

I wish you good luck, and I hope to see how you get there. I might steal some ideas off you. There's a 20-year old stack of 6x6's waiting to be scanned.

I'll keep working at it. Can I use a Canon Extender EF 2x III to keep the same distortion? Or is it really just better to use a dedicated lens capable of higher magnification without adapters? I don't really want to spend $800-1000 on a Canon 65mm f/2.8 1-5x lens, but to scan my 35mm film also, 1:2 is not enough for the prints I want to make of those. That would result in a 10 MP image (my 6D produces 20.2 MP shots so half that is the result of 1:2). I just want one lens that can do it all without paying a lot of money - I'm not an unreasonable person... haha.

 

[lukitas]

I'll keep working at it. Can I use a Canon Extender EF 2x III to keep the same distortion? Or is it really just better to use a dedicated lens capable of higher magnification without adapters? I don't really want to spend $800-1000 on a Canon 65mm f/2.8 1-5x lens, but to scan my 35mm film also, 1:2 is not enough for the prints I want to make of those. That would result in a 10 MP image (my 6D produces 20.2 MP shots so half that is the result of 1:2). I just want one lens that can do it all without paying a lot of money - I'm not an unreasonable person... haha.

I don't know much on the canon lens line-up; older nikon macros can be had at reasonable prices, I'd be surprised if that wasn't the case for canon. The use of an extension ring or bellows allows for mismatching lenses to cameras.
Except in the case of dedicated extension rings, electronic functions such as autofocus and diaphragm control are disabled anyhow.

A teacher I had taught that a reversed lens (mounting a 50 mm front to back) gives better results at the macro scale, because the ratio from large to small is inverted. I'm a little sceptical about this, because, if a lens shows serious aberrations when pointing the front element toward a close-focused subject, I think it would probably reproduce the self-same aberrations when pointing towards the negative. It could work when the front element is further from the negative than the back element is from the subject.

Maybe it only works for reproducing negatives taken with a certain lens, because that would invert the aberrations in the lateral plane. Not convincing, as aberrations that affect the vertical plane would be compounded.

Ah the joys of large reproduction ratio's. Long time ago, I used to work with a repro camera, with a lightbox in the base for film, weighed a ton. I don't remember it's limits, but it must have been 25 to 500 percent or something of that order. Enormous lens. I think it was 250 mm, and there was a second lens, a bit wider. I'd love to get my hands on one of those machines again.
I'd never thought of trying to print negatives on the repro.

 

[Jesse3Names]

I don't know much on the canon lens line-up; older nikon macros can be had at reasonable prices, I'd be surprised if that wasn't the case for canon. The use of an extension ring or bellows allows for mismatching lenses to cameras.
Except in the case of dedicated extension rings, electronic functions such as autofocus and diaphragm control are disabled anyhow.

A teacher I had taught that a reversed lens (mounting a 50 mm front to back) gives better results at the macro scale, because the ratio from large to small is inverted. I'm a little sceptical about this, because, if a lens shows serious aberrations when pointing the front element toward a close-focused subject, I think it would probably reproduce the self-same aberrations when pointing towards the negative. It could work when the front element is further from the negative than the back element is from the subject.

Maybe it only works for reproducing negatives taken with a certain lens, because that would invert the aberrations in the lateral plane. Not convincing, as aberrations that affect the vertical plane would be compounded.

Ah the joys of large reproduction ratio's. Long time ago, I used to work with a repro camera, with a lightbox in the base for film, weighed a ton. I don't remember it's limits, but it must have been 25 to 500 percent or something of that order. Enormous lens. I think it was 250 mm, and there was a second lens, a bit wider. I'd love to get my hands on one of those machines again.
I'd never thought of trying to print negatives on the repro.

I've done quite a bit of research on designing the 2-D guide system for the light box since the last post. I have rediscovered Igus. They're a Danish company, I believe, but they make linear guide systems. I have a DSLR slider from a few years ago from them and I remember it worked well, but I've stopped using it. I have four 10mm guides from that, but I have found a supported (has its own flat base) round rail at a minimum of 16mm diameter. They make 16mm guides now, too, so I will probably end up going with a setup from those parts. The guides are nice because they wrap almost completely around the rail instead of relying on minimal contact points of wheels on a round rail, which could (if not aligned perfectly) introduced lateral forces and ever-so-slightly twist the light box relative to the camera sensor orientation. I looked at wheels on a square rail, but honestly, this setup looks far more simple and easy. I have an alternative by parts available from Openbuilds Part Store online if the Igus parts end up being far too expensive for what they are in total. The Openbuilds parts are true wheels, but they have formed bars and plates that would allow the wheels to move in a perfectly matching groove, so in the case it saves a lot of money, I could do with that alternative.

1:1 is definitely plenty for 4x5, but I would love to be able to scan 35mm film as well, as I've said. I would only be able to get ~20 MP scans of 35mm film since my full-frame DSLR has that many pixels. Fortunately I'd be able to push 100 MP on a 4x5 scan, which would be plenty large for any prints I plan to make. So if I can't get a lens capable of more than 1:1 (I'd take 2:1 in a heartbeat because that'd give me ~40 MP 35mm scans!) that also has a lens correction profile in Adobe Camera Raw, I'll just go with a 1:1 lens with a profile. I've used reverse-mounted 50mm f/1.8 lens, but obviously my concern is lens distortion so for this application it doesn't work. It does give you around 1:1 magnification. I calculated it for a research application on my Master's work I tried it out on.

 

[Phil_F_NM]

You can use a 50mm f/3.5 Micro-Nikkor (or more realistically the 105mm) on your Canon digital with an adapter. Cheap, simple and as a manual focus lens, you don't have to worry about the camera messing up your shot and hunting through the negative for infinity.

I started doing this same project but for scanning exclusively 6x6 negatives. Eventually, I capitulated and bought a dedicated Nikon LS8000 scanner that I found for a great price (from a fellow RFF member.)

Basically, you don't need the ANR glass since 4x5 negs have such a rigid base that already keeps them parallel to the lens standard just hanging from the film holder. Adding ANR glass is an unnecessary step in my opinion. Also something that can introduce reflections and possible gremlins.

Just create a negative stage with a neg carrier from a LF enlarger, make a tube to hold the negative far enough out that you get corner to corner with the macro lens (with 4x5 and a 50mm lens, it's going to stand off the camera at least a foot and a half) then fab up a mount to attach it to the front of your lens. Then you're good to go.

Phil Forrest

 

[Jesse3Names]

You can use a 50mm f/3.5 Micro-Nikkor (or more realistically the 105mm) on your Canon digital with an adapter. Cheap, simple and as a manual focus lens, you don't have to worry about the camera messing up your shot and hunting through the negative for infinity.

I started doing this same project but for scanning exclusively 6x6 negatives. Eventually, I capitulated and bought a dedicated Nikon LS8000 scanner that I found for a great price (from a fellow RFF member.)

Basically, you don't need the ANR glass since 4x5 negs have such a rigid base that already keeps them parallel to the lens standard just hanging from the film holder. Adding ANR glass is an unnecessary step in my opinion. Also something that can introduce reflections and possible gremlins.

Just create a negative stage with a neg carrier from a LF enlarger, make a tube to hold the negative far enough out that you get corner to corner with the macro lens (with 4x5 and a 50mm lens, it's going to stand off the camera at least a foot and a half) then fab up a mount to attach it to the front of your lens. Then you're good to go.

Phil Forrest

Phil! We had talked a year or so ago about your Voigtlander 21mm lens I was thinking of purchasing for my Nikon S3 RF. How've you been?

You must mean the Nikon 55mm f/3.5 Micro-Nikkor, not 50mm. How big of an issue was lens distortion when you imaged your 6x6 slides? I plan to take multiple photos of the negative (if it's any bigger than a 35mm strip since that's the size of my sensor with a 1:1 setup) - not sure if you understand. You said corner-to-corner with a 4x5 negative on a digital camera with an extension tube and I interpreted that as filling my Canon sensor with the 4x5 image in a single photo. I'm thinking of saving some money and renting a Canon 65mm f/2.5 1-5x macro lens from something like Borrow Lenses. that way I would have access to its lens correction profile and up to 5x magnification! I could just scan negatives in batches... or maybe very politely ask a local camera shop if I could bring my rig in and use a shelf model in-store! I may have to call in a few favors for that one and beg ever-so-heavily.

I didn't know of large format negative carriers! Thanks. That will work perfectly and they're much less expensive than ANR glass. Do Newton's rings appear because of the shape of the lens elements, a not perfectly flat film strip, or a combination of the two? After looking, I plan to incorporate a macro rail focus rack that can hold the weight vertically for extended time... or at least build something that does the job with the weight.

 

[Jesse3Names]

After quite a bit of thinking, I have decided to redesign the light box. I plan to buy a single, long LED strip and cut it into sections between units of LEDs, lay them next to each other on the base, and re-wire the units to each other. Basically just extending the space between the units to create a matrix of LEDs on the baseboard. This way it'll be an even distribution of light instead of attempting to diffuse side-lighting. Also, I can use a 12V power supply and won't have to modify the connector that comes with the LED strip itself.