DSLR Scanner: Camera Supports and Positioning

[Peter J. De Smidt]

Hi Joseph,

These are some good things to think about. Having spent a little time messing with some bearings today at our farm supply store, I'm leaning towards agreeing with you that they're unnecessary for our project.

Regarding the goals:
1) achieving evenness of illumination,
2) consistency of focus,
3) finding a lens with minimal distortion and curvature,
4) simplifying and improving stitching,
5) getting rid of dust.

In my case, the biggest pressing issue is 4). Some of the software available does a good job with structured mosaics captured with something like a gigapan. My hope is that getting the sampled areas to be very regular in position will enable the use of the structured settings of the software to eliminate manually adding control points, which for me is just as tedious as spotting. I certainly might be wrong about that. The gigamacro project seems to work pretty well, though, and a motorized Z-axis would allow focus stacking, if needed.

Regarding minimizing dust, I don't have any idea how to do that other than to use standard cleaning techniques. I have a hepa filter running in my scanner area, and I used forced air and a dslr sensor cleaning brush to clean the film.

[rdenney]

Focus consistency requires a flat negative, and motion in the negative stage that maintains a precise distance from the camera. That precision needs to be good enough so that the camera does not have to be refocused, which will introduce some changes in magnification, especially since the motion error will probably accompany an axial error.

And the film really needs to be at right angle to the lens axis, and parallel to the sensor. Error there will cause geometric distortion, and likely focus variation.

It's not that hard to achieve good precision in these dimensions. But it is really hard to achieve if we don't set out to.

The edges of the sensor should be parallel to the motion--that avoids having to rotate each image or deal with stair steps in the tiles.

It's easier, though, to correct lens distortion and falloff using software, as has been discussed at some length. The better the lens and light source at minimizing these, the less work the software will have to do. But there may always be some software correction required--distortion less than a pixel eliminates the need for blending, and blending probably does more damage than correcting distortion. We've discussed that also, and I'm persuaded.

Planes sliding against each other will trap dust and wear or scratch.

I built a vertical film stage so that I could avoid glass and not worry about sag. It also lets me look through the camera with the apparatus at work-surface height. The camera's mount is also loaded normally rather than being cantilevered.

Aligning the slides I used was not particularly difficult, as it turned out. And it's one less thing to worry about.

Using lead screws may not be precise enough to avoid control points (either manually placed or found by the software), I don't think, at least without using a fine stepper motor. The Velmex slide that Peter and I used for focusing has 40 threads/inch. That's 10.6 microns/degree of rotation. My 5D has sensels just a bit smaller, so that thread and a degree wheel on the handle could provide precise positioning of the film stage. But it would be inconvenient in practice without being motorized--that's a lot of turning. And those slides are expensive, especially in long lengths.

Just some points to consider as a reminder.

Rick "hoping for some testing this weekend" Denney

[jb7]

Good luck with your testing-

You appear to have criticisms of my design, though you continue to neglect to address them directly-

Just quickly-

Do you think there is a problem with cantilevering the camera? The camera isn't subjected to any live loads, and the structure is stiff enough to allow focus to be maintained through a complete scanning cycle, at the very least. In testing, I did use a right angle finder return the viewfinder to vertical, but later found that operating the camera via the computer was better than any other method.

You mention the problem with dust and scratching of the glass-
In my testing, the dust in contact with the transparency is the dust that is the problem,
dust on the bottom of the 6mm thick glass isn't a problem at all-
similarly with any scratching that might be induced by sliding a sheet of glass over a sheet of cartridge paper.
I think you'd need to try very hard to scratch the glass, or maybe sprinkle some filings around-

Regarding control points, the worst of them can be deleted, providing you have better ones. The average control point distance on my scan was 1.69 pixels, less than the resolution of the lens, and undetectable at the scales we're dealing with- 1.69 pixels from an image that's 21,000 pixels high. I'm not sure that it's going to be possible to be pixel perfect in an optical system that requires more than 2 pixels to resolve a pair of lines.

As I said, good luck with the testing-
I have no doubt you'll be able to capture higher resolution in fewer pixels, given the dedicated macro optics you've chosen-

I've prepared some full resolution samples from my tests, will post them later this evening-

[Peter J. De Smidt]

It's not that hard to achieve good precision in these dimensions. But it is really hard to achieve if we don't set out to.

I'm doing my best to do so.

Planes sliding against each other will trap dust and wear or scratch.

My carrier consists of a piece of 6x6" P99 acrylic for the bottom, and a piece of 1/4" thick ABS for the top. Under the acrylic are four feet made of uhmw polyethylene. These feet slide on a thick glass plate. The carrier isn't heavy. I doubt that wear will be a big issue.

I built a vertical film stage so that I could avoid glass and not worry about sag. It also lets me look through the camera with the apparatus at work-surface height. The camera's mount is also loaded normally rather than being cantilevered.

I really don't see the cantilevered situation as being a problem. With my very first test, I set a gitzo tripod over a light box. To the center column, I attached a super clamp. To the super clamp, I attached a 410 head. To the head, I attached a PB-4 bellows. Now that setup was cantilevered. But the results were still very close to my Cezanne, and that was using continuous light. With my current setup, I can stand on the negative stage and the camera support structure, basically anything but on the velmex carriage or camera. I doubt that the very small cantilever situation is a problem, and I doubt it is with Joseph's design.

Using lead screws may not be precise enough to avoid control points (either manually placed or found by the software), I don't think, at least without using a fine stepper motor. The Velmex slide that Peter and I used for focusing has 40 threads/inch.

My Velmex has 20 threads/inch, although a 40 threads/inch screw is available. Regarding control points, the only way to find out would be to try it. I doubt the gigamacro uses manually added control points, and they use lead screws for positioning. Yes, I plan on using a stepper motor with 200 whole steps in a revolution. Microstepping will also be an option.

[rdenney]

Okay, some outcomes:

1.) In the Omega holder, the film has about 0.3mm of curl. I put that in the dammit category. But I need to calculate depth of field before becoming too disappointed, and test further.

2.) The slide projector as a light source is unacceptable. In any position, the camera images the projection bulb, which in no position fills the frame. I can still use it, but diffusion is required. A 45-degree reflector ala what Peter has done makes a good diffuser, and I was able to keep going with just propping up a piece of white paper. A piece of diffusion glass inserted as a slide might solve that problem.

3.) I can't tell yet whether the Canon 50mm macro lens at 1:1 is sharpest at f/5.6, f/8, or f/11. But at least that doesn't seem to vary across the frame--the field seems quite flat.

Yes, I know I hit three threads with this list.

Gotta go play a gig--more testing later.

Rick "whose brain isn't wrapping around the slide projector issue at the moment" Denney

[Peter J. De Smidt]

Something like this:

is useful to past on the knob of a Velmex slide. Idea borrowed from Charles Krebs.

[Nathan Potter]

Okay, some outcomes:

1.) In the Omega holder, the film has about 0.3mm of curl. I put that in the dammit category. But I need to calculate depth of field before becoming too disappointed, and test further.

2.) The slide projector as a light source is unacceptable. In any position, the camera images the projection bulb, which in no position fills the frame. I can still use it, but diffusion is required. A 45-degree reflector ala what Peter has done makes a good diffuser, and I was able to keep going with just propping up a piece of white paper. A piece of diffusion glass inserted as a slide might solve that problem.

3.) I can't tell yet whether the Canon 50mm macro lens at 1:1 is sharpest at f/5.6, f/8, or f/11. But at least that doesn't seem to vary across the frame--the field seems quite flat.

Rick "whose brain isn't wrapping around the slide projector issue at the moment" Denney

Rick, .3mm (13 mils) is pretty good. Depth of focus, Df = 2CN so at N = f/5.6, Circle of Confusion C at 10um (.010 mm.), you have a Df of 0.11mm (about 4 mils). I assume you would choose a COC roughly about the size of the DSLR pixel dimension in order to make use of the inherent resolution of that DSLR. But it is a bit of a task to keep the focus plane centered around +/- 0.15 mm. Of course the Df scales linearly with N so going to smaller aperture N gains you Df leeway at the cost of lens resolution loss. This will be tricky if we want to approach real drum scanning capability in the 5 to 10 um pixel capture, at 1:1, referenced to the film surface. Any higher magnification exacerbates the Depth of Focus problem. But maybe full frame capture at 1:2 is attractive for most large format film. That would yield a 48 X72 mm frame and increase the pixel capture size by a factor of 2 referenced to the film.

Using the slide projector would be tricky since the dim imaging of the bulb filament can be a problem. This possibly due to the presence of a planar lens surface within the projector optical path, or the reflector focal length falling at a critical location forming an aerial image that can be imaged with the macro lens.

As I think more carefully about this the use of a flash source becomes even more attractive. The high speed eliminates vibration, and the 6000K color temp. is friendly to the DSLR. Peter is on the right track here, unless repeatability of intensity is a problem. Also I think there is a way to focus the flash to a near point using condenser lenses to achieve a collimated uniform source that may be highly beneficial in reducing scattered light and achieve a much better Dmax performance. The Dmax performance would be a key attribute to the DSLR approach so may be realizable.

Nate Potter, Austin TX.

[Peter J. De Smidt]

Here's the fairly current prototype:



Here's a Google Sketchup model of a potential linear positing system:http://dl.dropbox.com/u/3595413/sled2c.skp Please be kind, I just started using cad today. Everything is close to the proper scale.
The motors would be: http://www.probotix.com/index.php?vi...product_id=152
The linear rails and bearing blocks would be: http://www.ebay.com/itm/160773515442...=p5197.c0.m619

[jb7]

Nice model Peter-
one suggestion, you might save some space by positioning the threaded screw element (that sticks out the back) beside one of the linear rails-
it might even be possible to lose one of your linear rails on each assembly too.

Also, a while ago, I bought some of these trapezoidal rods, complete with a matching tap- for another project I haven't started yet-

Anyway, these are in the UK, but it might still be useful- being able to tap your own lead nuts might provide some sort of saving-
check out his other items too, but it might be cheaper for you to source in the US-

http://www.ebay.ie/itm/x3pcs-TR10x2D...#ht_500wt_1374

[Peter J. De Smidt]

Joseph,

Thank you for the suggestions. I had looked for an inexpensive ACME tap recently, but I wasn't able to find one.

The lead screw positioning and bearing rail suggestions are certainly worth looking into. Once I have some of the rails and blocks, it should be fairly simple to do some experimenting.