PDA

View Full Version : Making a scanner with a DSLR



Pages : [1] 2 3

Frank Petronio
22-Dec-2011, 13:30
So instead of belly-aching over how crummy flatbeds are and how old drum scanners are a pain to deal with, why don't all the engineers here tackle creating a new, modern drum scanner that's as open-sourced, off-the-shelf, and future-proofed as possible?

A lot of those old drum scanners were pre-PC, they must not be that complicated... Other than the drums, what else needs to be fabricated rather than bought? Isn't most of the value in the engineering, not the hardware? Heck I can buy $50 lasers at Home Depot.

What if 1,000 serious film photographer put down $1,000 deposits on a $2,000 price-point, Heathkit-style drum scanner? Wouldn't a million dollars of pure R&D money be more than most of the big corporations spent on their classic drum scanners back in the day?

We could get those argumentative lawyers that keep pestering the mods to set up a clean not-for-profit organization to administrate the project. Make it open source, everyone contributes... Ending up with a Volkscanner for the masses, able to handle 8x10 and modern software and interfaces (just have a network or even wifi interface)... Best $2000 ever spent - you know there are easily more than 1,000 people who would jump at this.

So why not?

Ed Kelsey
22-Dec-2011, 13:34
Man that must be some killer weed you're smoking.

domaz
22-Dec-2011, 14:13
Well it seems older drum scanner used photomultiplier tubes. I'm betting you can't get those anymore. Newer ones use CCD arrays which are not easy for amateurs to deal with. It seems more practical to put a whole bunch of APS-C sensors on a device that can position them. You take one picture with the whole bunch of sensors, shift the sensors (because there will be borders between the sensors), take the picture again and stitch the result. If this could be done with an 8x10 cameras I'm sure the resulting digital image would be in the Gigapixel range easily.

Andrea Gazzoni
22-Dec-2011, 14:26
it seems pmt tubes are still produced

http://sales.hamamatsu.com/en/products/electron-tube-division/detectors/photomultiplier-tubes.php

bigdog
22-Dec-2011, 14:43
Serious question: Is a drum scanner inherently better than a flat bed, given all (most) other factors being equal? Or, is it a matter that the old high end drum scanners were better than the present consumer flatbeds? I mean, if the Epson V750 "Pro" msrp's at $850, and you're willing to spend $2000, what could be done with a flat bed at the $2000 price point (that only needs to go up to 8x10 - I realize there are larger flat beds.)

Brian C. Miller
22-Dec-2011, 15:20
What if 1,000 serious film photographer put down $1,000 deposits on a $2,000 price-point, Heathkit-style drum scanner?

*shudder*

I used to work for a manufacturer (yes, we had our own pick-and-place machines) and my boss designed part of the Intel 80386, and the company president wrote the book on the Intel i860, literally. I can see $2,000 in just parts alone, and the result would look ugly, to put it mildly.

Hurdles: Focusing, stability, and speed. These are mechanical things. The "easy" stuff, like interface, is very simple: network connection. A 1-Gb network connection is easy and will be adaptable for at least a decade or two into the future.

There has to be a lens assembly commercially available which can be actuated by stepper motors. The mechanics for the drum would probably have to be custom machined.

And what about when it breaks down? Then you might have a $2000 hunkajunk and just hoping that somebody has some free time to help you out.

$2000 can buy a really good flatbed scanner with decent film scanning performance.

Frank Petronio
22-Dec-2011, 15:29
Yeah a drum scan will have less noise and be "clearer" with a little extra range. Not that you can't work with Epson scans - I do. There is always going to be a small market for people who want the best, but it will always be there too. An even bigger market is for the photographers who is willing to pay 2-3x what a good Epson costs.

How is not something that could be manufactured in a clean garage?

Sevo
22-Dec-2011, 16:15
Having operated them in the past, drum scanners were bitchy, painfully slow, required disgusting amounts of service and quite extraordinary amounts of consumables (Mylar sheets, fluid, cleaner, cleaning wipes and the odd drum and lamp ever so often). Nobody doing a run-of-the-mill scan service can still do that profitably on a drum scanner, and the few individuals still making money out of them have tweaked their scanners to perfection, and probably would not be inclined to switch to a new scanner built to meet a much lower price point...

Currently the market is that saturated with old pro scanners that there is no profit in any new high end scanner development. Whenever there will be enough demand to start high end scanner building again, it will be in low numbers, and will probably take a entirely different direction - more like a 2D Imacon or those micro-movement enhanced medium format backs.

A high quality macro lens, FF or APS-C sensor and a piezo base for micro stepping the sensor to multisample at a finer than raster pitch should today be capable of delivering above 100MP scans at pro digital camera quality, using current off-the-shelf components rather than anything proprietary or not made any more - which is essential if the thing must be affordable enough for enthusiasts and archivars rather than the advertising industry, and has to be made by the ten rather than the thousand.

Frank Petronio
22-Dec-2011, 16:18
A high quality macro lens, FF or APS-C sensor and a piezo base for micro stepping the sensor to multisample at a finer than raster pitch should today be capable of delivering above 100MP scans at pro digital camera quality, using current off-the-shelf components rather than anything proprietary or not made any more.

Most useful comment. So is anyone doing this or working on it yet? Or where do you buy it and how much does it cost?

Isn't the drum scanner always going to have an advantage in dynamic range, unless they start to develop camera sensors with a greater-than-current range? Are there any roadblocks to getting a higher range than film out of a digital sensor? Or can you do a high and low exposure and combine them in HDR?

Mike Anderson
22-Dec-2011, 16:27
Well it seems older drum scanner used photomultiplier tubes. I'm betting you can't get those anymore. Newer ones use CCD arrays which are not easy for amateurs to deal with. It seems more practical to put a whole bunch of APS-C sensors on a device that can position them. You take one picture with the whole bunch of sensors, shift the sensors (because there will be borders between the sensors), take the picture again and stitch the result. If this could be done with an 8x10 cameras I'm sure the resulting digital image would be in the Gigapixel range easily.

Yeah I'd throw a bunch of cheap sensors and a bunch of cheap computing power at the problem. I'd make everything static, just greatly overlap each sensor's image field and have the computer correct for lens inaccuracies, film curvature and average out the noise.

...Mike

Leigh
22-Dec-2011, 16:34
I could contribute significant sweat equity in the form of design work, both mechanical and electronic. I certainly have the credentials for same.

I can also do the prototype, since I have a full machine shop and an electronic development lab.

This is actually quite easy and simple. Only the film and its support drum move. Everything else is in a fixed position.

- Leigh

Jim Michael
22-Dec-2011, 16:37
Essentially macro gigapans, probably with different stitching algorithm. The hardware is out there already.

domaz
22-Dec-2011, 16:54
Yeah I'd throw a bunch of cheap sensors and a bunch of cheap computing power at the problem. I'd make everything static, just greatly overlap each sensor's image field and have the computer correct for lens inaccuracies, film curvature and average out the noise.

...Mike

I was thinking about this at the point of capture- a replacement for film. But the only way to overlap each sensor's image field is to put a lens on each sensor. Can any kind of lens focus on the image being put out by the Large Format lens in front of it?
Diagram to make things clearer:
SM--L

S=array of sensors
M=array of micro lens for each sensor
L=LF lens

Is there any such micro lens(M) that could focus the image from the LF lens(L)?

Ivan J. Eberle
22-Dec-2011, 16:57
For LF scanners don't have to be much better than a really good flatbed or an average PMT scanner to please the majority. It's with the smaller formats (135, 645, 2-1/4) where the 3 to 6 micron apertures of the fine-art-purposed PMTs really shine. I see an Aztekbusiness-model opportunity for others who might adapt and repurpose some of the other makes of hardware that are often practically being given away now. But it would be hari Kari to start from scratch with all the idled prepress gear flooding the warehouses.

Frank Petronio
22-Dec-2011, 17:18
OK I am only going to show more ignorance here, but aren't the flatbed scanners we've already got doing this same thing? Doesn't a Creo move it's sensor all over the place in the X and Y axises? So it is simply a matter of using a larger chip from a professional digital camera and an overall tighter, more precise and tuned mechanism?

So the scanner improvements would be kind of like the difference that chip size makes in quality, such as in going from a $100 point and shoot to the latest medium format digital back?

Mike Anderson
22-Dec-2011, 17:40
OK I am only going to show more ignorance here, but aren't the flatbed scanners we've already got doing this same thing? Doesn't a Creo move it's sensor all over the place in the X and Y axises? So it is simply a matter of using a larger chip from a professional digital camera and an overall tighter, more precise and tuned mechanism?

So the scanner improvements would be kind of like the difference that chip size makes in quality, such as in going from a $100 point and shoot to the latest medium format digital back?

Is there much difference in quality between a teeny cheap sensor and a big expensive sensor if they're both working at their preferred ISO?

(I'm ignorant too here.)

...Mike

Frank Petronio
22-Dec-2011, 18:37
Is there much difference in quality between a teeny cheap sensor and a big expensive sensor if they're both working at their preferred ISO?

(I'm ignorant too here.)

...Mike

Really, just make more, smaller passes and make it slower and cheaper.

Jim Michael
22-Dec-2011, 19:15
This (http://gigamacro.com/gigapixel-macro-imaging-system/) is sort of what I had in mind.

Frank Petronio
22-Dec-2011, 19:28
Wow that is cool thanks

But still, in absolute terms, a good drum scanner will capture a longer tonal range with less noise, even though something like the GIGAmacro could possibly beat it on resolution. Or?

Mark Stahlke
22-Dec-2011, 19:29
I think Frank is on to something and not just on something.

Why are there no mid-range film scanners? Film scanners jump from the Epson v700/750, a high end consumer grade scanner, that goes for around $700 up to something like the Flextight scanners that sell for upwards of $10,000. There is nothing in between. Why not?

Surely, someone could build a scanner similar to the v700 with some upgrades to all the critical pieces. A better sensor, better optics, precision machined gears, high precision stepping motors, and a really good film holder. I think such a scanner could come in at a price point around $2,000 to $3,000. I know the market would be limited but I still think there is some money to be made there.

Frank Petronio
22-Dec-2011, 19:46
I think Frank is on to something and not just on something.

Why are there no mid-range film scanners? Film scanners jump from the Epson v700/750, a high end consumer grade scanner, that goes for around $700 up to something like the Flextight scanners that sell for upwards of $10,000. There is nothing in between. Why not?

Surely, someone could build a scanner similar to the v700 with some upgrades to all the critical pieces. A better sensor, better optics, precision machined gears, high precision stepping motors, and a really good film holder. I think such a scanner could come in at a price point around $2,000 to $3,000. I know the market would be limited but I still think there is some money to be made there.

Seems like the institutional and government offices alone would be good customers, not to mention photographers. Build it professional quality, even like those old Agfa Arcuses.

Peter De Smidt
22-Dec-2011, 20:51
A good drum scanner has greater dynamic range than a ccd scanner, but it doesn't have greater range than an HDR style multi-exposure ccd system, since the latter can have as big a dynamic range as you need. All you need to do is to take more exposures at different levels. A variable light source would also be good, but you'd have to be careful of spectral shifts. Perhaps neutral density filters would work.

The big problem is alignment, and the precise movement of the film. (I think it's better to move the film holder than the sensor.) The level of alignment is what it would take to make a huge print under an enlarger. The film must stay perfectly flat and parallel to the ccd sensor and lens.

Some type of x-y position system would be needed for the negative carrier, perhaps with two calibrated lead-screws.

So, get, build or modify a light box with a very even source, one that's quite a bit better than the industry standard type boxes. (I don't know how repeatable flash equipment is, but if it is repeatable enough, it might make a good light source, as you can use a very short duration of flash, which should help with vibration. You'd still need a constant source for focusing.) Make a tray that goes on top, it doesn't have to touch the light box, which will hold a film holder that can be moved in a repeatable, grid-like fashion. A clam shell holder with p95 acrylic would be ideal. You'd then need a mount for the camera. Since it wouldn't have to move much, it should be a multi-column mount for stability.

A macro bellows, such as a Nikon PB-4, along with a high quality enlarging lens would be a good choice, probably a 50mm, as they're capable of the highest resolution.

The carrier stage would have to be masked very well, and the light would half to be blocked from the front of the lens to the negative.

Operate the camera remotely through live view, although focus would be manual.

The whole system would have to be very sturdy and well-isolated from vibration.

Nathan Potter
22-Dec-2011, 20:57
Looks like the GigaMacro is already available. One would only need a backlight to scan transparent film media. In the $15 K range that's not too bad. But at the advertised resolution there are a bunch of very precision mechanical parts especially for X and Y motion that are costly. Essentially this is a step and repeat camera where the resolution and contrast will be limited by the macro lens. Scan time will be limited by the field of view of the macro lens and stitching time can be added to the physical scan time.

A higher magnification lens heading toward a microscope objective will provide resolution down to a few um but a small FOV, so many more images are required. But higher magnification requires tighter focus control, or auto focus, so more precision. Mechanical precision is expensive.

What's less expensive is signal processing. Thus it may be possible to trade the mechanical precision in X and Y motion for pattern recognition and image reconstruction algorithms; meaning the scan can be sloppy if the images overlap.

Lots of possibilities beyond the GigaMacro but using a high quality CMOS sensor with a magnifying macro lens would make sense and ultimately provide a less expensive machine than using a drum.

Nate Potter, Austin TX.

Daniel Stone
22-Dec-2011, 21:06
Problem here is(IMO):

We're not looking to design a line of cars, with a cheapo/intro model, a medium($40k/year) model, and a heart surgeon's model w/ all the bells and whistles ;).

The high-end scanners like the Azteks, Heidelbergs, Creo/Kodak's were designed for PROFESSIONAL use. That means MORE MONEY SPENT. The latest generations of drum scanners from Aztek, Heidelberg, Screen, and the latest hi-end flatbeds like the Kodak Eversmart Supreme II's, no expense was spared to put the BEST tech into them, and for them to perform flawlessly(well, when used properly). Just like the top-tier Ferrari's, they need a servicing every now and then. Just part of doing business. Bulbs and belts need to be changed, lead screws need to be lubed, and sometimes, a full goings-through needs to be performed to make sure no bugs pop out when not wanted. Of course many people neglect their "normal maintenance cycle", thinking "That won't happen to me". Well, it happens.

If you want the best tech possible, and I'm assuming 16bit TIFF file output, then you'll need to pay for it. PMT's aren't cheap, and small companies like Aztek(I was there yesterday actually, dropped in since I was in the neighborhood, they're 3 people total I saw in the office, YES... 3 people). They need to be paid, and keep enough to keep the lights on.

The problem I see here is two-fold: You have your lab owners/scanner operators(Lenny for instance), and you have your "amateurs"(some who might know just as much as the "pros", if not more, but might only be scanning for themselves, not as a service for profit). That's the KEY word: PROFIT. Its called keeping the lights on. The Aztek Premier will clock in around $50k after the scanner, mounting station, a drum or two, Pro software($1500), computer to drive the scanner, a shipping crate to hold it during transit, and $1k to ship it to you across the country/world, etc... Yes, its expensive. But you can buy an 80mp digital kit, outfit it with 3-5 "top tier" lenses, and still be around the same amount of dough in the end. If you're scanning/shooting for yourself, then its your call. But in the end, if you want to deliver the goods to paying clients who are willing to spend $100+ on a scan, then you'll probably want the best machine w/ the best support possible. In the USA, IMO, that's Aztek. They've been sooooo helpful in aiding me in my acquisition of my drum scanner, and answering technical emails/phone calls w/o asking for $$$ in return. They know I'm serious about my craft, and don't have money to burn, but I don't bullshit with their time either. They're professionals(IMO), and professionals SHOULD be paid.

If I was going to introduce a drum scanner, I'd want to make sure you can mount up to 11x14 film on it. And I WOULD NOT be doing it for less than $40k. If I hadn't already gone with a drum scanner, I'd be going with a Kodak Eversmart Supreme II flatbed(and still fluid mounting almost everything on it). IMO, its more flexible, and the minor difference in file quality(IMO) isn't worth the headache and risk of damaging someone's film on a fast-spinning drum scanner(even though I know how to properly mount film).

Oh, what about overhead on these "1000" scanners, there's gotta be tech support on these, someone to talk to when something goes wrong. That costs money, its called overhead.

I think that drum scanner manufacturers INTENTIONALLY priced themselves high, just so the "small fish" photographers w/ the "I need the best" mentality REALLY think twice before ordering something like these machines. There's more involved than most people really think...At least I know there is, I'm learning a buttload, and I'm not technical(but I'm learning I need to be :D)

just my $.02

-Dan

Peter De Smidt
22-Dec-2011, 21:57
This is a neat challenge. When my Cezanne dies, I'll probably pursue it.

Sure, we have many disadvantages compared to a big company, but we also have many advantages. The software available now for raw processing is much better than any scanner software I've used. We can replace our sensor, i.e. the camera, for a modest amount, and they have improved dramatically. Lots of the macro bellows and so on are available for little money, and enlarger lenses are close to free. In addition, we don't have to use materials that are compact or easy to move. For instance, we can use a 1/2 inch (or thicker) piece of glass as a base to slide the negative holder on, and many of us already have glass negative carriers for our enlargers. It's automating the whole thing that would be the really big challenge, but it doesn't really have to be that automated. If it must be, there are DIY CNC machines that have to be accurate down to a 1/1000 of an inch, and a technically inclined person could probably modify such a device.

In addition, commercial scanners have to be so precise because they only read a couple lines at a time, and the sampling has to be incredibly precise so that all of the sampled lines can be put together. With a dslr, though, you take reading of a much wider area, and as long as each sample area is overlapped enough, it's no problem for software to put the pieces together.

Jim Michael
23-Dec-2011, 04:53
This is something I've been looking into for some time hence the reference to the GigaMacro. There is an experiment that can be performed to validate the approach which is to construct a pin grid system that allows systematic movement of a stage holding a target below a macro lens. No stepper motors involved, just shoot and manually move to next position.

Re focus, there is a technique used in photomacrography that might be of use called stacking, where the stage is moved along the lens axis to bring various points in focus. Not sure the utility in this context, but just brainstorming.

Re lenses, if you need to get down below 1:1 it might be worthwhile looking at microscope objectives. I've been seeing some pretty nice looking stuff from those. There are a couple of configurations, such as objective on bellows and infinity objective on a lens.

When is it more advantageous to move the imager vs. moving the stage?

Peter De Smidt
23-Dec-2011, 07:43
<snip> There is an experiment that can be performed to validate the approach which is to construct a pin grid system that allows systematic movement of a stage holding a target below a macro lens. No stepper motors involved, just shoot and manually move to next position.<snip>

That is what I have in mind, as well.



Re lenses, if you need to get down below 1:1 it might be worthwhile looking at microscope objectives. I've been seeing some pretty nice looking stuff from those. There are a couple of configurations, such as objective on bellows and infinity objective on a lens.


That would certainly be worth looking into, especially for smaller than LF film. For LF, something on the order of 1:1, i.e. image size of one sampled area equals image size of sensor, should give pretty high quality. With my digital camera, an older model, that should allow a 10x enlargement or so, assuming the negative is sharp enough. That's bigger than I'd ever want to print. 1:1 would be well within the capabilities of an enlarging lens on a bellows.




When is it more advantageous to move the imager vs. moving the stage?

Well, a camera/bellows/lens combo is quite heavy, and it would be non-trivial to be able to move it precisely, perfectly parallel to the negative, and without inducing vibration. With the negative stage, on the other hand, you could use some thick textured acrylic (P95) to make a clam shell holder. You could put a small foot in each corner of teflon (or similar) tape, and you could slide it on a very heavy sheet of glass.

Here's picture of coraline algae growing on the glass of our aquarium. The colony was about 1/4th" in diameter. I used my d200, a Pb-4 macro bellows, and an enlarging lens. I simply eye-balled the alignment, and my D200 doesn't have live view.
http://i955.photobucket.com/albums/ae37/peterdesmidt/Coraline.jpg

Mike Anderson
23-Dec-2011, 10:24
If we just make a mechanical product that accepts a wide range of digital cameras it would allow you to choose your price/quality point by using a $500 camera or an $8000 camera. And as digital camera technology continues to scale so would the system, the mechanical part we build would not obsolesce.

Precision adjustment would be needed to set the height and level the plane of the camera. X Y placement needn't be precise as long as there's overlap because the stitching software can easily deal with the slop.

...Mike

Nathan Potter
23-Dec-2011, 10:55
Yes! You guys have the idea both mechanically and optically. This can already be demonstrated using a light table and digital camera with a good macro lens of low distortion. Use conventional stitching software to reconstruct the image. The key here is that this is not live image capture - the subject is stationary.

The next task is to build a prototype that more or less automatically does the same thing, faster and more precisely. The trick in all this is to use low cost OEM parts on a solid chassis in order to keep the ultimate cost down. The engineering is relatively easy - the market analysis not quite so.

Nate Potter, Austin TX.

jp
23-Dec-2011, 11:08
The usage problem I see with a stitching scanner is that there is no way to preview to set exposure/levels without actually doing the scan. If it had a high speed mode, it could do a quicky scan for setting exposure. Remember DSLRs are pretty sensitive to exposure accuracy.

It's a constructive comment because I like the idea.

I would agree that presently available stitching software can take care of alignment imprecision, except for focus of course. I would think a cnc router table would be a good start. Moving the film instead of the camera would provide an easier to build light source, as a smaller even diffuse light is easier to build that a very large (11x14+) diffuse source.

Putting the whole thing in a cabinet that is dust proof and light proof is a good idea. Not everyone has a clean pristine lab to set such a system in. To dampen vibration, electromagnets could temporarily attach the movable stage to it's frame during exposure, and release it's grip during positioning.

The stage itself could be a filmholder made of white diffuse plastic, or at least the center part of the it. If the film moves a little during positioning, the stitching software can take care of it.

Frank Petronio
23-Dec-2011, 11:16
And what about dynamic range? Won't it only be as good as a digital camera?

Mark Stahlke
23-Dec-2011, 11:23
And what about dynamic range? Won't it only be as good as a digital camera?
Yes, but HDR techniques can ameliorate that.

The DSLR based "scanner" is starting to sound like a glorified copy stand.

Peter De Smidt
23-Dec-2011, 11:35
If we just make a mechanical product that accepts a wide range of digital cameras it would allow you to choose your price/quality point by using a $500 camera or an $8000 camera. And as digital camera technology continues to scale so would the system, the mechanical part we build would not obsolesce.

Right! And our sensor system would be current and servicable for a reasonable amount, unlike pro-scanner repair costs.


Precision adjustment would be needed to set the height and level the plane of the camera. X Y placement needn't be precise as long as there's overlap because the stitching software can easily deal with the slop.

Yep. Good points.


<snip> The key here is that this is not live image capture - the subject is stationary.<snip>

Good point.


The usage problem I see with a stitching scanner is that there is no way to preview to set exposure/levels without actually doing the scan. If it had a high speed mode, it could do a quicky scan for setting exposure. Remember DSLRs are pretty sensitive to exposure accuracy.

That’s true, although one could use the light table to figure out the brightest area of the negative. Take a test frame of the brightest area. Adjust exposure. Now take a picture of the darkest area. Is the dynamic range enough? If yes, proceed with the “scanning”. If no, take another frame with more exposure. Let’s just say that one stop more exposure would be enough. Now take two pictures at each sample site, using auto bracketing to give the correct exposures. Before stitching, use HDR software to blend the exposures. Stitch the blended exposures to get your file. While I haven’t tried this, I expect exposure fusion would work much better than traditional HDR tone-mapping.


The DSLR based "scanner" is starting to sound like a glorified copy stand.

I agree. Getting fancier, though, would be very expensive.

If we work on this, we could come up with a DIY plan, including a parts list, workflow tips…

Ed Kelsey
23-Dec-2011, 11:43
Why not build your own car? This is really a ridiculous idea if you ask me.

Frank Petronio
23-Dec-2011, 11:49
Why not build your own car? This is really a ridiculous idea if you ask me.

Ed, do you wake up ready to shit on everything all day long or do you just squat and pucker up for this forum?

jb7
23-Dec-2011, 11:52
As has been mentioned, a high degree of accuracy is not needed, the stitching software can take care of that. Perhaps the RepRap 3d printer platform (or something similar) could be modified to provide the moving baseplate? It's a very cheap system, proven technology, both hardware and software, and not difficult to imagine how it might be adapted to accept a light box and camera mount- plus, it's all open source-


http://reprap.org/wiki/RepRap

E. von Hoegh
23-Dec-2011, 11:54
Why not build your own car? This is really a ridiculous idea if you ask me.

I haven't chimed in on this thread because I know dick about scanners.
But I do know that very often, a small and dedicated group can come up with a product that blows the doors off the commercial offerings.

Why not let them have fun?:)

Mike Anderson
23-Dec-2011, 12:03
The usage problem I see with a stitching scanner is that there is no way to preview to set exposure/levels without actually doing the scan. If it had a high speed mode, it could do a quicky scan for setting exposure. Remember DSLRs are pretty sensitive to exposure accuracy.

Making each region-capture a 5 shot bracket wouldn't take much longer than a 1 shot. An HDR approach might be the more popular workflow anyway in which case setting exposure needn't be precise.

...Mike

Peter De Smidt
23-Dec-2011, 12:13
Perhaps the RepRap 3d printer platform (or something similar) could be modified to provide the moving baseplate?

http://reprap.org/wiki/RepRap

It certainly should be possible.

We should probably do some tests just to see what the resolving power and contrast performance of the optical/sensor system would be, not doing any stitching to start. If we can get the quality we want on one frame, then we could move on. If we can't get the quality we need on one frame, then there would be no point in automating the process. We should probably aim to at least achieve the quality achieved by one of the pro-scanners. I have a Cezanne. Supposedly it's resolution is very good, around 6000 line pairs per inch, but it's dynamic range can't match a good drum scanner. I also have a PB-4 bellows and a bunch of enlarging lenses. And so I can compare (and post) comparisons. The main issue would be that my D200 doesn't allow live view, and so very precise macro focusing is a pain.


Making each region-capture a 5 shot bracket wouldn't take much longer than a 1 shot. An HDR approach might be the more popular workflow anyway in which case setting exposure needn't be precise.

Great point, Mike.

Peter De Smidt
23-Dec-2011, 12:19
I was just given a Nikon SB-28 Speedlight. It goes down to 1/64th power. I'm considering making a light source out of white styrofoam and difused plexi for this experiment. Does anyone know how repeatable the intensity of these flashes are? I guess I'll find out. Maybe I'll trying using one of the mixing boxes from my De Vere 504 for the light box.

Jim Michael
23-Dec-2011, 12:28
For cinematography they are using a log interpretation of exposure to achieve a DR on the order of 14 stops (there's a whitepaper on Sony's site with the math). Perhaps something like that could be employed to take advantage of the HDR results.

Jim Michael
23-Dec-2011, 12:30
Peter, I'd guess the exposures would be pretty consistent if the flash is in a fully manual mode and the flash is allowed to fully recycle.

Mike Anderson
23-Dec-2011, 12:52
What if there were 2 parts, a bridge to hold the camera and a light box. They would both placed on a flat surface like a piece of glass (not included).

The light box would have wheels and motors within, it would scoot itself around the surface like one of those robot vacuum cleaners. There would be no external drive rods and assemblies.

The downside of this is lack of precision in locating the light box. We agree that the precise X Y location isn't important.

But what about rotational/alignment precision? If the robot light box doesn't stay aligned precisely North-South-East-West, will the software be able to correct for this when stitching without degrading the image?

(To minimize the need for mechanical precision and put the burden on computation would be nice.)

...Mike

jp
23-Dec-2011, 13:08
Why not build your own car? This is really a ridiculous idea if you ask me.

Thousands of people build their own cars every year, to get something better/different than what a manufacturer offers. Superperformance is the leader in this. A fully crowdsourced car is the new LM Rally Fighter, where you go to the factory and help build your own car. http://www.rallyfighter.com/buy-a-rally-fighter. Factory Five and Solidworks had a crowdsourced design project as well for a less expensive car. http://grassrootsmotorsports.com/design/

When international conglomerates don't build what is needed, slightly crazy people find a way.

E. von Hoegh
23-Dec-2011, 14:06
Thousands of people build their own cars every year, to get something better/different than what a manufacturer offers. Superperformance is the leader in this. A fully crowdsourced car is the new LM Rally Fighter, where you go to the factory and help build your own car. http://www.rallyfighter.com/buy-a-rally-fighter. Factory Five and Solidworks had a crowdsourced design project as well for a less expensive car. http://grassrootsmotorsports.com/design/

When international conglomerates don't build what is needed, slightly crazy people find a way.

Nothing new about that....http://www.flickr.com/photos/michaelwardphotos/3380579654/

Peter De Smidt
24-Dec-2011, 04:42
Let's talk about goals. Here are some suggested ones:
1) High Quality, with the resolution of a pro flatbed but greater dynamic range.
2) Low price.
3) Ease of construction.
4) Ease of use.
5) Scalability.

The targeted user would be people like us: Film using photographers who can't afford or justify buying a good drum scanner.

I'll build a prototype using a PB-4 bellows, 80mm Rodagon, Portatrace light box, Nikon D200, because that's what I have.

It will be a table top device. There will be a camera support structure made out of wood, basically a thick beam supported by columns on both ends, like an upside down U. The PB-4 macro rail will be bolted to the cross beam. The light box will sit on the table underneath the support structure. Above the light box, there will be a thick piece of glass. I'll build a frame for the glass, such that three bolts can be used to adjust it's alignment. [The bellows will be aligned by pivoting it on it's mounting screw, and then the the glass table will adjusted to be parallel to the front of the lens. I'll use a Screen clam shell negative holder, which will be set on the glass, and various opaque masks will be used.

Some of the quality bottle necks will be:
A) The precision of the bellows, i.e. is the front perfectly square to the back.
B) The lens. Unfortunately, I no longer have a quality 50mm enlarging lens, at least I don't with a 39mm thread. I do have a bunch of other medium format and large format enlarging lenses, but I expect that the best performance would be with a 50mm f2.8 6 or 7 element design. My 80mm Rodagon, a fixed aperture lens, was designed, I think, for best performance at 1:1, which should be pretty close to what we need for LF scanning.
c) My ability to focus. Unfortunately, my D200 doesn't have live view, especially the ability to use live view while tethered to a computer. This would be a good use for D90 and D300s, as they allow live view.
d) The light source is visibly uneven, but we only need to use a small section of it. It should be good enough for a proof-of-concept.

sully75
24-Dec-2011, 05:06
Do it!

Just wondering...what bellows are available for Canons?

Jim Michael
24-Dec-2011, 07:41
There are two types of Canon bellows, with and without circuitry for the aperture. The former is pretty spendy and available from B&H. Latter is less than $100 and can be found on Amazon. I'll be testing a macro lens so I don't need bellows up to 1:1 and have extension tubes for slightly over that. If you do go with bellows you can pick up an old Micro Nikkor and use with an adapter, or use a microscope objective as previously described. q.v. photomacrography.net for good info.

At 1:1, using a 30% overlap, with a sensor in a Canon 7D, I'd estimate it would take about 80 exposures to cover a single 4x5. That seems like oversampling a tad. Raw is 14 bit. I haven't tested my macro lens yet to see what resolution to expect.

Peter De Smidt
24-Dec-2011, 08:48
Oh boy, 80? That would almost be as tedious as making a profile with a Spyder3Print! Hopefully, the needed amount of sampling sites will be lower.

Macro lenses are also a good idea. I do have a couple of Nikon ones, a 55mm and a105mm, and I can test those as well.

Nathan Potter
24-Dec-2011, 09:56
Transferring a portion of a transparency to a sensor through a macro lens has been done by many people including me. It is a sort of academic exercise now in that it can be done at high resolution with a top quality lens. The lens is the limiting factor in the detail that gets resolved. The choice is in the degree of magnification that one chooses. At 1 to 1 and smaller the resolution is set by the sensor. At higher magnification the resolution limit is set by the resolving power of the lens. As magnification is increased of course the field of view is decreased so more frames are required to cover the film. Even the best macro lenses will have difficulty in achieving an airy disc image as small as the 3 um or so that a good Aztec is capable of.

Of considerable interest is the type of back illumination that one might choose. That can be all the way from a diffused source to a highly collimated source. It only needs to cover the field of view of the objective if it moves in synchrony with the lens (as someone suggested above). A collimated backlight has the advantage of providing much less light scattering so higher contrast image transfer and more grain imaging than a diffuse source. An adjustable light source might be quite handy.

Dynamic range is limited by that of the sensor regardless of what is on film, until the film range falls below that of the sensor. For large film dynamic ranges one could employ digital HDR techniques already mentioned.

Nate Potter, Austin TX.

sully75
24-Dec-2011, 10:41
I know that negative film can capture more real outside dynamic range than digital can (at this moment, at least) but does that mean that digital can't then capture that dynamic range from film? I sorta thought film compressed it into something smaller that digital would be able to capture without resorting to HDR.

sully75
24-Dec-2011, 10:42
I wonder if this lens on a canon body would be useable...

http://www.usa.canon.com/cusa/consumer/products/cameras/ef_lens_lineup/mp_e_65mm_f_2_8_1_5x_macro_photo

Peter De Smidt
24-Dec-2011, 10:48
I've been following the various threads on the subject over the years. The concept is certainly not new. What would help, though, is coming up with an inexpensive and easy to build system. I agree that we probably won't reach the ultimate resolution of the very top drum scanners, but for medium and larger format, do people really need that high of resolution scans? I scan my 4x5 negs on my Cezanne at 2400 spi, and honestly that's overkill. My biggest printer is a 24" ....

Jim Michael
24-Dec-2011, 12:49
Interesting points re the collimated light source Nate and clearly demonstrated to me printing scratched negatives on my D2 with the condensers vs. the Aristo head. The light source and camera can remain static while the stage is moved around between them.

Peter De Smidt
24-Dec-2011, 13:17
One nice thing is that it should be easy to play with different light sources. I expect a diffuse source will minimize grain, whereas the collimated one should be a bit sharper, but maybe with larger grain.

We probably want to maximize info gathered by the sensor, which means a good histogram with red, green and blue. As such, I hope to use a 5K very high CRI light source, possibly using a Solux bulb, or perhaps flash, although bracketing with the later would be a pain.

Just an idea, but it would be fun to play with scanning color negative film with a filter over the light source to counter-act the orange mask.

Peter De Smidt
24-Dec-2011, 13:30
See: http://www.arrickrobotics.com/xy.html

and:

http://www.instructables.com/id/Low-Cost-Hobby-Servo-XY-Table/

Or even better:

http://www.instructables.com/id/Internet-Arduino-Controlled-T-Slot-XY-Table/

Sevo
24-Dec-2011, 16:28
http://www.instructables.com/id/Internet-Arduino-Controlled-T-Slot-XY-Table/

Cool, but even the least of them would be overkill, if you join/stitch by software rather than relying on the mechanics. As long as you scan at high enough resolution and with enough overlap between frames to stitch by grain pattern (which is highly accurate as grain is random and - other than image content - extremely unlikely to have any pattern repetition within one negative), there is no precision requirement for the camera movement in the 2d plane. I.e. you might as well use a table only capable of moving in 1cm steps...

jb7
24-Dec-2011, 16:47
Automated stitching through the use of software is the obvious way to approach this problem- just as automated positioning of the original might be desirable, if upwards of a hundred exposures per scan have to be made- assuming an 8x10, for example.

Of course, you could make each exposure after repositioning the original manually, I suppose...

Peter De Smidt
24-Dec-2011, 18:06
Cool, but even the least of them would be overkill, if you join/stitch by software rather than relying on the mechanics. As long as you scan at high enough resolution and with enough overlap between frames to stitch by grain pattern (which is highly accurate as grain is random and - other than image content - extremely unlikely to have any pattern repetition within one negative), there is no precision requirement for the camera movement in the 2d plane. I.e. you might as well use a table only capable of moving in 1cm steps...

I agree. In particular, the x-y position system could be smaller, and it doesn't need to be all that fast. Variable step size might be nice, though, as it could then handle mid and high resolution. It's someplace to start. I like the t-track extrusions, as many people won't have a full workshop available. One potential problem would be how to link control of the positioning table with control of the camera.

I wouldn't mind manually positioning the negative stage for, say, 12 exposures; but 80 would be out of the question. If I want that kind of pain, I'd rather start hand pouring my emulsions again.

Jim Michael
24-Dec-2011, 18:42
There needs to be an orderliness to the process. If HDR approach is used I think you may need to merge the mosaic at each exposure level and then combine those layers, else it may lead to an uneven density in each tile, like you might get from shooting a pano with the light constantly changing. How good is HDR software at handling (combining) layers resulting from stitched panoramas?

Peter De Smidt
24-Dec-2011, 19:49
That doesn't work that well, as the stitches are unlikely to match exactly. While you could write a script to have Photoshop align the layers, my guess is that it's be better to do the HDR first, and then do the stitching. That's what's done with many HDR panoramas. All of this is assuming that we need HDR. With bw, it might not be needed unless one let density get out of control.

Will Frostmill
24-Dec-2011, 20:17
re: HDR
Would HDR really be necessary? Negative film has a really shallow curve, doesn't it, with that huge dynamic range spread very thin? I seem to remember looking at the curves for TMAX in Leslie Strobel's book and on Kodak's info sheets, and thinking that a huge range was captured in a really modest curve.

I could see slide film being a problem, but for the fact that most slide film did not have a dynamic range of greater than 6 or 7 stops, and most DSLRs can do 9 to 13 stops. (Admittedly the way dynamic range is defined on film vs on sensors is different, but...)

Peter De Smidt
24-Dec-2011, 21:23
The question with bw film will be the noise characteristics in the shadows. I expect it'll be fine with one exposure with standard contrast film, but if the shadows get too noisy, multiple exposures could help. Color films, though, would probably benefit from multiple exposures.

sully75
2-Jan-2012, 20:14
bump. get on this people. I'm not smart enough to help.

Frank Petronio
2-Jan-2012, 20:16
Yeah is it done yet? I got $300 in PayPal for it. WTF, getrdone!!!

Ben Syverson
2-Jan-2012, 21:01
It's funny, Justin and I were just talking about developing something like this as a Wanderlust Cameras project.

It would be 4x5 at first, due to leverage on the rails—an 8x10 version would need to be much larger and heavier.

It would be fully manual. If you've ever used the LCD and button interface on the GigaPan, you'll know why... It's just too fiddly, and 4x5 wouldn't require enough frames to warrant automation. And if you want to do HDR (only necessary for slide film) or supersampling to reduce noise, you don't have to reprogram anything.

In general terms, it would be a specialized XYZ macro rail system with detents, manufactured from an overkill of solid aluminum. You bring the light source (maybe we'd make one), camera and lens. So in five years, you can upgrade the camera for even better resolution.

It's crazy that people are paying $4000 for Coolscans. Anyone could contract a system like I've described above, and retail it for under $250. If no one does it in the next year or two, Justin and I will do it...

Brian C. Miller
2-Jan-2012, 22:17
OK, let me see if I have this straight:

1. Motorized X-Y table over light table
2. Camera mount
3. Computer control of X-Y and camera

There isn't anything here that's complex. The user supplies the light table according to the need. The XY table goes over the light table at Z height. The computer controls camera position, focus, exposure and shutter. A big black box covers it to keep out light and dust.

Done.
4x5, 8x10, 11x14, 16x20, 20x24, freakin' huge whatever. Same basic table hardware. An Arduino controller ($60) with USB can control stepper motors which run the table. You don't need high precision because the camera just needs to be put in an overlapping spot to make the image. Then move it to the next segment.

Am I missing something?

Frank Petronio
2-Jan-2012, 22:37
Create a nice app for it, give it a snazzy name, package it up, hit a $500 price point?

sully75
2-Jan-2012, 22:55
Anyway to digitally manage dust?

Is this going to be too crude to do 35mm?

Peter De Smidt
2-Jan-2012, 23:11
Am I missing something?

Nope, you've got it.

The big questions are:
1) Can such a system provide the needed quality? (It's an awful amount of work if we can't do better than a V700.)
2) Can stitching software operate successfully at the grain level? With normal sized stitches, there tends to be much more prominent image elements.

I just acquired some 1/2" thick glass plates, and I can use a Screen clamshell holder to hold the negative, but I don't have a live view capable camera, and so focusing is going to be very tricky. If someone has a D300 (or similar) that they could lend me for a week or two, it would really help.

Brian C. Miller
2-Jan-2012, 23:16
Oh, yeah, and servos on the camera mount for auto alignment. Don't want to forget that one!

So what you really want is a drum scanner, but you'll settle for a computer-controlled copy table adapted to a digital camera. Not at all difficult.

The Open Camera Controller (http://www.hdrlabs.com/occ/index.html) can be a start on how to control a DSLR. The Arduino page has notes on steppter motor interface. The Gimp has a batch mode and ScriptFu that can stitch the images together and do some other stuff if you like. The rest is commodity hardware.

Hmmm, I only have a point & shoot, and I'm not upgrading just yet.

Brian C. Miller
2-Jan-2012, 23:27
The big questions are:
1) Can such a system provide the needed quality? (It's an awful amount of work if we can't do better than a V700.)

It can provide the quality according to the DSLR and macro lens used with it, nothing more than that. This is just an automated copy table. It can't do more than that. If the camera can focus down to the grain, then fine. Tim Parkin could shed some light on that, as he has a DSLR on a microscope. I'm guessing that a DSLR with a 1:1 macro lens won't be able to resolve the grain, but I have no personal clue.

If somebody here has a DSLR with a 1:1 macro lens, it would be easy to get some actual results.


2) Can stitching software operate successfully at the grain level? With normal sized stitches, there tends to be much more prominent image elements.

No idea. I'm guessing that it could.

Frank Petronio
2-Jan-2012, 23:54
Right now Nikon is reading this thread, quick get out the NDAs.

Eh no. Photographic companies listening to customers to make products? Hahahaha how old-fashioned.

Still... a 0.063% royalty would be nice.

Peter De Smidt
3-Jan-2012, 00:19
Well, sure I have a number of 1:1 capable macro lenses, and an enlarging lens on a macro bellows can do much more than that. But as I've mentioned, focusing at that level with my camera is a huge pain...And I already have a nice scanner, and a violin to finish making...

tlitody
3-Jan-2012, 06:10
will it work on an IPhone?

Brian C. Miller
3-Jan-2012, 11:50
Yes, it will work on an iPhone. DSLR cameras can rest comfortably on top of an iPhone and still function. Same goes for LF cameras, too. Caterpillar 797F trucks can also work on an iPhone, but after they do, the iPhone won't work!

rdenney
3-Jan-2012, 12:06
I could see slide film being a problem, but for the fact that most slide film did not have a dynamic range of greater than 6 or 7 stops, and most DSLRs can do 9 to 13 stops. (Admittedly the way dynamic range is defined on film vs on sensors is different, but...)

Slide film may only see 6 or 8 stops of subject brightness range, but it actually spreads them out to a wide range of densities, from clear film to nearly opaque. Of course, the density range of slides is intended to reflect highlights and shadows as seen in the real world. Slides are the most difficult of all to scan because of their density range.

Negatives have much less density range. I can scan both color or black-and-white negatives in an Epson flatbed without running out of range, though for really thick black and white negatives I have had to use a wide-gamut color space to avoid clipping the extremes. A digital camera should have no trouble capturing the density range of negatives.

Sensors are pretty linear--moreso than film--but one assumes it would be fairly easy to correct the scanning bias using a general tone curve adjustment. The sensors in digital cameras are certainly as good as the tri-linear sensors on scanners.

Rick "thinking we might all have to construct such tools when the existing generation of scanners pass away" Denney

Ben Syverson
3-Jan-2012, 12:24
I know the comment about iPhones was facetious, but it may be more right than you know... Within a couple years, your phone will do fantastic macro at 12-15 MP. More than enough for stitching.

From my end, it's much easier to write iPhone software that takes photos and tiles them immediately, as opposed to separate camera control software and desktop software to collate and stitch.

I still think it's a mistake to make the first version motorized. You undoubtedly WILL have stability, vibration and focus problems.

rdenney
3-Jan-2012, 12:27
Anyone could contract a system like I've described above, and retail it for under $250. If no one does it in the next year or two, Justin and I will do it...

You provide what you have described for $250, and I'll be first in line. I don't mind moving the camera manually if there are reasonable detents. I can stitch in Photoshop.

Now, as to whether it would be better than an Epson, a more refined question is what level of camera and lens would be required to beat an Epson?

Here's a modest setup: I have an old 90mm SP Tamron with its 1:1 converter that is not bad. Let's say it resolves 60 lines/mm. At 1:1, that would put 60 lines/mm on the sensor. That would take the 5D's sensor right about to its Nyquist limit with 125 sensor sites/mm. I'm quite sure that lens is good enough to do that, even though it's old and relatively cheap. But 125 sp/mm is 1600 spi.

Nope, not good enough. The Epson can do maybe 2000ish. Maybe the camera would do a better job with tonal range and so on. Maybe not. But the resolution has to be high enough to take advantage of the negative.

So, we'd need a lens that is real good at, say, 2:1, to get 3200 spi. Or a camera with a higher sensor density, but not so high that it outstrips an affordable lens. 3200 would match a high-end flatbed if it could deliver it accurately with affordable equipment.

How about a proof of concept? Can someone with a prospective camera and lens mount the camera such that a proper photo can be made of a piece of a negative with all the moving and stitchin? A 1:1 macro lens would be the baseline, with the next step up being a very good enlarger lens (possibly even reversed) on a bellows, perhaps. Let's see what real resolution we can get at different price points of lenses and cameras. If it takes $4000 worth of camera and lens to beat a $1000 Epson, interest wanes. If I get time this weekend, I'll compare my "modest" setup with a negative that I have already scanned on the Epson.

Rick "thinking the concept needs proof before building a machine" Denney

tlitody
3-Jan-2012, 12:52
Now let me see. We like to use film because of all the detail we can capture. And we like the grain in film. So now we are going to photograph the film with a digital camera which we don't like because it doesn't have grain and will not capture the grain in the film anyway, and then use the digital file to output digitally.
I have a really great idea. Why don't we just use a digital camera and cut out the film altogether. And there's this little thing called HDR which means we can get way more dynamic range than film anyway. And since we would be stitching the images of the film we can use the digital camera on a pano tripod head and stitch the images to get more resolution than even film could produce if we use the right lens and get it right.
I bet no one thought of that before. Hang on a minute, I think some people have been doing that for quite a few years. I guess we got left behind somewhere.

Or is it just the aesthetic of a scanned neg which is what we really like? You know that sort of soft look that uncorrected lenses give.

Ben Syverson
3-Jan-2012, 13:06
Rick,

With a 5D Mark II and a 1:1 magnification, you're capturing 4000 raw SPI. Perhaps more pertinently, you're capturing somewhere in the neighborhood of 60-70 lp/mm with good contrast, depending on lens. So you're essentially matching the Coolscan 9000, which rates around 67 lp/mm.

BTW, 60 lp/mm requires 3048 SPI. I think your math is off somewhere. 1600 SPI is only enough to resolve 31.5 lp/mm, and that's in an ideal world.

Anyway, given that color negative film doesn't have much contrast past 70 lp/mm and few LF lenses can achieve that, I'd say it's enough. But if it isn't, or you shoot fine grained B&W, just wait a year or two and upgrade the camera to eke out more resolution. That's the good thing about a generic rail system. Not good enough? Just wait five minutes.

The Epson doesn't really compare. In no way does it deliver 2000 high contrast samples per inch. But even if it theoretically could, that only works out to 40 lp/mm. Scanning on my 4990, I maybe see 30-40 lp/mm, but they're very mushy low-contrast lines. After some very destructive sharpening, you can bring out that level of detail, but you'll bring out plenty of noise and sharpening artifacts as well.

Mike Anderson
3-Jan-2012, 13:09
...
Rick "thinking we might all have to construct such tools when the existing generation of scanners pass away" Denney

You're not alone:

http://theonlinephotographer.typepad.com/the_online_photographer/2011/12/burning-our-bridges.html

...Mike

Kirk Gittings
3-Jan-2012, 13:15
Tlitody, That is a good question going forward from here possibly without film but doesn't address the issue looking back at (in my case) 40 years of 4x5-only a fraction of which has been scanned.

Personally I think this is a idea with some potential usefullness as scanners fall by the technological wayside.

Peter De Smidt
3-Jan-2012, 13:18
I still think it's a mistake to make the first version motorized. You undoubtedly WILL have stability, vibration and focus problems.

I don't plan on making my first version motorized. For me, a setup that allows a decent and fast "scan" for a digital version of a contact sheet would be useful. It would also allow experimentation into hi-res options.


Rick "thinking the concept needs proof before building a machine" Denney

Absolutely right.

I can test a Nikkor 55mm macro, a Nikkor 105mm AF micro, and an 80mm Rodenstock Rodagon F4 on a bellows. My tests will be limited by my ability to focus my camera at those magnifications. I've found that extreme magnifications are very difficult to focus.

Jim Michael
3-Jan-2012, 13:27
Is the test target that was used for the scanner comparisons available?

Peter De Smidt
3-Jan-2012, 13:38
I have a really great idea. Why don't we just use a digital camera and cut out the film altogether. And there's this little thing called HDR which means we can get way more dynamic range than film anyway. And since we would be stitching the images of the film we can use the digital camera on a pano tripod head and stitch the images to get more resolution than even film could produce if we use the right lens and get it right.


I do regularly use a digital camera, including HDR and multirow stitching. What I've found is that the stitching works really well for a distant subject, but if you're fairly close, and using a wide field of view, the distortion gets really nasty, and since the stretching is so extreme, the end resolution gets really low on the edges.

I recently photographed a church interior, both with digital and with my 4x5. With my setup, the 4x5 has much more detail than my digital setup. Sure, there are much better digital systems available, but I can't afford them.

sully75
3-Jan-2012, 13:45
[QUOTE=tlitody;826564]Now let me see. We like to use film because of all the detail we can capture. And we like the grain in film. So now we are going to photograph the film with a digital camera which we don't like because it doesn't have grain and will not capture the grain in the film anyway, and then use the digital file to output digitally./QUOTE]

This thread already had a buzzkill.

sethlatimer
3-Jan-2012, 13:52
The macro copy stand would need to beat the epson by a fair bit, say 3 times better or 7200 dpi.
That is better than 50 megapix per sqaure inch.
So you need more than four but not as many as 9 non overlapping images in the inch to better that resolution. What is the most affordable solution to get a shot of a third of an inch square area?

Ben Syverson
3-Jan-2012, 13:57
tlitody's comments are a good but ultimately unfounded challenge to this idea. A good camera and 1:1 lens can reproduce grain and achieve a very faithful digital representation of the negative. It doesn't look "digital."

If one-shot digital photography is good enough for your needs, then by all means, do it. However, as numerous tests have shown, you can capture much more detail and latitude in one shot with film.

If you're shooting static interiors, digital HDR is a viable alternative. If you need to shoot people or landscape, you're dealing with a moving subject, and digital HDR becomes more problematic. Negative film gives you a one-shot "HDR" image, which really just means your highlight information isn't thrown away.

And of course, with LF you have movements, unique lenses and the luxury of a large ground glass which makes for faster easier focusing.

So all of the above gives an outline of the reasons to shoot and scan film.

tlitody
3-Jan-2012, 14:02
Tlitody, That is a good question going forward from here possibly without film but doesn't address the issue looking back at (in my case) 40 years of 4x5-only a fraction of which has been scanned.

Personally I think this is a idea with some potential usefullness as scanners fall by the technological wayside.

OK rent or buy a better scanning back for your 4x5 camera. Mount your best 4x5 enlarging lens on your 4x5 and photograph you negs from a very bright light table in one go. I think the graphics repro industry have been doing this for some while. Art work is digitised in this way. I just think that we are trying to reinvent the wheel here.

or

ICG still make drum scanners last time I looked. So they are not cheap. So instead of investing your $1,000,000 dollars in R&D, why not just pool enough to buy a highend drum scanner, employ someone to do scanning and offer the service country wide and those that contributed to the scanner get cheap scans done. Problem solved.

Ben Syverson
3-Jan-2012, 14:02
The macro copy stand would need to beat the epson by a fair bit, say 3 times better or 7200 dpi.
That is better than 50 megapix per sqaure inch.
So you need more than four but not as many as 9 non overlapping images in the inch to better that resolution. What is the most affordable solution to get a shot of a third of an inch square area?
50 MP per square inch is in the overkill category, unless you're talking about fine grained B&W shot on Leica or Zeiss.

With that said, most of the superzooms / high-end compacts have relatively small sensors with great macro lenses. Find one that can go to 1:1 or beyond, and you could probably produce a gigapixel image of 6x6. Whether or not it's rational is another question...

The nice thing about APS-C or FF DSLRs is that they cover a lot of area with fantastic quality, so you can cover the negative with fewer exposures.

tlitody
3-Jan-2012, 14:05
but if you're fairly close, and using a wide field of view, the distortion gets really nasty, and since the stretching is so extreme, the end resolution gets really low on the edges.


get a Leica M9 with a Zeiss zm 2/35. There is no distortion.

Frank Petronio
3-Jan-2012, 14:08
OK rent or buy a better scanning back for your 4x5 camera. Mount your best 4x5 enlarging lens on your 4x5 and photograph you negs from a very bright light table in one go. I think the graphics repro industry have been doing this for some while. Art work is digitised in this way. I just think that we are trying to reinvent the wheel here.

or

ICG still make drum scanners last time I looked. So they are not cheap. So instead of investing your $1,000,000 dollars in R&D, why not just pool enough to buy a highend drum scanner, employ someone to do scanning and offer the service country wide and those that contributed to the scanner get cheap scans done. Problem solved.

The point is that that there are trillions of film images still out there, I have a billion or so on my office bookshelves. I don't think any scanners are being made beyond Epsons and a few inexpensive slide scanners, so they are only a product cycle or two from being phased out. In a few years, spare parts my dry up. So the idea is how might we scan all this existing film, not to mention any new materials, 10-20 years down the road?

But yes, if I want a quality drum scan tomorrow then I'd just send out to Lenny or some other good provider. But what happens when Lenny gets old and his machines aren't repairable any more?

A digital photography rig as is being discussed makes a lot of sense because it is relatively "future-proof" and can be upgraded indefinitely.

Ben Syverson
3-Jan-2012, 14:18
get a Leica M9 with a Zeiss zm 2/35. There is no distortion.
Overkill... all macro lenses are designed to have virtually 0 distortion at working distance.

sethlatimer
3-Jan-2012, 14:37
It sounds like the recommendation is to attach a lens on a macro bellows? to a live view DSLR? Is there a particular lens or type of lens that would work best? 50mm 2.8 enlarger lens?

On another topic, I have been working a bit with the imagemagick and panotools software. These are open source tools that enable batch processing and scripting of files.
I know there is some software that can control the camera but I am not that familiar with that. The camera control software needs to be linked to the moving table(prolly arduino). That linking is beyond my ability.

I could definitely work on the stiching and exposure balancing with the imagemagick and panotools, and maybe a little gui with other open source tools.
Seth

Peter De Smidt
3-Jan-2012, 14:45
...OK rent or buy a better scanning back for your 4x5 camera. ...

....ICG still make drum scanners last time I looked....

...So instead of investing your $1,000,000 dollars in R&D….

...get a Leica M9 with a Zeiss zm 2/35. There is no distortion....

If I had $1,000,000, I'd buy a premier or two. Scratch that. I'd just have Lenny scan my negatives. But I don't. Nor do I have the change for any of the other suggestions. You seemed to have missed the "inexpensive" criteria that all of us are including.

You don't like the idea. That's fine. There's no reason that you have to take part.

Peter De Smidt
3-Jan-2012, 14:49
It sounds like the recommendation is to attach a lens on a macro bellows? to a live view DSLR? Is there a particular lens or type of lens that would work best? 50mm 2.8 enlarger lens?


That would allow very high resolution, but we don't know yet if it's overkill. It does add difficulty in sourcing and cost, mainly for the bellows.

Most Macro lenses go to 1:1, and extension tubes can help them get much closer.




I could definitely work on the stiching and exposure balancing with the imagemagick and panotools, and maybe a little gui with other open source tools.
Seth

That's good to know.

Ben Syverson
3-Jan-2012, 14:55
FWIW, you may have trouble with extension tube stability, and I would be very hesitant to push a macro lens beyond its engineered range. You may get okay performance, but a 1:1 macro lens has been laid out by the optical engineer for lowest distortion and optimal performance at exactly 1:1.

At the very least, testing is warranted. But honestly, I don't see the rationale for pushing past 1:1. With pixel densities as they are, you're getting 4000 samples / inch or better.

Kirk Gittings
3-Jan-2012, 14:58
To me the issue would be the software because present day stitching software IME does not do well with detail poor areas like skies. If you broke down a 4x5 into say 16 tiles and the top half were of an overcast sky you would have a difficult task of alignment at hand.

Peter De Smidt
3-Jan-2012, 15:05
Kirk's worry is, imo, the most serious one facing the idea.

I'll try taking some test shots tonight.

Ben Syverson
3-Jan-2012, 15:55
Kirk has a point, but a stitcher with specific knowledge of the application will do better. For example, dust would be a natural way to connect images.

Brian C. Miller
3-Jan-2012, 15:58
Vibration: Sequence is photo, move then stop, photo, etc. There is no movement vibration, and, hey, stop shaking that table!

Stitching: This isn't random. Stitching software has a problem because it has to join up a number of unknown sequences, while these sequences are known. X across, Y down, all known. This makes things much easier.

Why not a scanning back: the cheapest back from Betterlight is $6,500 for an 18Mp back. Canon Rebel T3 is $500 (12.2Mp), Sony Alpha DSLR-SLT-A33 is $480 (14.2Mp), Nikon D3000 is $500 (10.2Mp). All have macro lenses available, and there's extension tubes for the Canon and Nikon cameras.

Why not dump all the film gear and go digital: Uh, hello, anybody read Tim Parkin's comparison of 8x10 to MF digital? 80Mp is not enough for me. Blow it.

Here's the problems of using a digital camera to scan the negatives: Aliasing and the Bayer filter. It would takes something like Hasselblad's new "200Mp" camera with the moving sensor to overcome this. No consumer stepper motor is going to be moving one sensor pixel in position. Look at the comparison (link (http://www.landscapegb.com/2011/12/big-camera-comparison/)) and tell me it's kicking ass even on the Mamiya 7 on that resolution chart. It is going into distortion before it hits the 6 mark.

rdenney
3-Jan-2012, 16:12
Rick,

With a 5D Mark II

Ben, no offense, but this tiny factual mistake set off a series of observations that would not be needed with a few careful tests. My camera is a 5D, not a 5DII. The sensel size is 8 microns. That's 125 sensels/mm, which can record half that at best (60 lines/mm), and that's only in the horizontal and vertical directions. That's my 1600. Yes, it might be less, but it certainly will not be more. The point was that a 13 megapixel camera isn't going to out-resolve an Epson at 1:1, no matter how sharp the pixels are or are not, or how sharp the lens is. It will take much higher magnification, which will require a different lens arrangement and more challenging mechanics.

A sensor with twice the density won't help, if the lens can't keep up with it.

Gotta travel now, but I have the camera, the lens, the negative, and the light table. Maybe modest, yes, but that's what I have and I can't afford to replace it. This weekend, hopefully, I'll put a data point out there.

Rick "gotta run" Denney

Nathan Potter
3-Jan-2012, 16:40
Rick, at 60 lp/mm the contrast of the Epson is piss poor. See somewhere here a plot I made of the Epson 750 of resolution vs contrast. IIRC at about 60 lp/mm (I think I went up that high) the contrast was below even 5%. The digital equivalent would be much much higher especially with premium optics. Additionally one can play with the illumination source and use one highly collimated to further enhance contrast in the copy, a feature not available for the Epson. I found it! Rick per below. Actually I see the contrast is about 0 at 60 lp/mm, so pretty much useless while the DSLR is vastly superior (not counting field of view of course).

Nate Potter, Austin TX.

http://farm7.staticflickr.com/6171/6154033421_f5c72e4169_z.jpg (http://www.flickr.com/photos/argiolus/6154033421/)
EPSONcont-web-1 (http://www.flickr.com/photos/argiolus/6154033421/) by hypolimnas (http://www.flickr.com/people/argiolus/), on Flickr

Peter De Smidt
3-Jan-2012, 18:15
Here's a dynamic range test. I hope that's the right term. Please remember that I was a liberal arts major. I photographed a Stouffer 31 step step wedge on a Portatrace light box with a d200 and a 105mm Nikkor AF-D Micro lens. F-stop was f5. This is only meant to test how my equipment will see density with one exposure. The raw file was processed in Lightroom. All of the density/contrast settings were zero. I added what the eyedropper in Photoshop told me the LAB L channel values were and also visual density numbers from my x-rite densitometer.

http://i955.photobucket.com/albums/ae37/peterdesmidt/Stouffer_One_Exposure.jpg

Next is 3 exposures, 1 stop apart. I used Photomatix's tone compressor to tone map the HDR file. Obviously, there's a lot of room for adjustment, and you can always take more exposures.

http://i955.photobucket.com/albums/ae37/peterdesmidt/ToneCompressor.jpg

el french
3-Jan-2012, 19:09
My studio macro setup is designed to do focus stacked multirow spherical panoramas by rotating around the entrance pupil of the lens. http://img.photobucket.com/albums/v649/etfrench/Macro%20Pano%20Head/AutoFS_02s.jpg It can also do orthographic stitches by moving the subject.

Here's a 500 MP image created with it: http://photosynth.net/view.aspx?cid=39c28546-4202-4821-8b2d-4567cfceae6a Note: Zooming all of the way in is equivalent to pixel peeping at 200%.

While this setup could probably do an 8x10 scan, it would be better to use a vertical setup. Vibration is the biggest problem to overcome. Mirror slap and shutter vibrations are the hardest to deal with. I'm using a 2 second delay to handle the mirror slap and short duration flash to handle the shutter vibrations. Currently only Canon has electronic first curtain shutter. Another option would be to use an external leaf shutter.

Off the shelf linear slides like Stackshot, http://www.cognisys-inc.com/stackshot/stackshot.php, can easily do 1 micron steps. Building a dedicated XY stage with salvaged parts from a scanner should be easy enough to do. I use an Arduino to do the focus steps and it wouldn't be very hard to update the program to control an XY stage with micron resolution. Getting it to move 1 micron without stiction (stick/slip) will be the hard part.

Optically, you'll need a lens that is sharp corner to corner, no distortion, no CA, etc. There are not many inexpensive lens that will qualify. I have a JML 21mm f/3.5 lens that would work very well, but I don't know if any are available new.

Peter De Smidt
3-Jan-2012, 19:28
Fun! I had planned on using the t-channel aluminum extrusions for the support structure.

Peter De Smidt
3-Jan-2012, 20:56
This test is very imperfect. But for what it's worth, here is a scene photographed with a Toyo AX, 120SA, Acros, divided Pyrocat MC and scanned with a Cezanne.

http://i955.photobucket.com/albums/ae37/peterdesmidt/Church_BW.jpg

The scanned file could produce a 30x40" print at 300 dpi. (Actually the final shot is cropped in a bit, and so it'd be more like 26" x 40")

Using a light box, pb-4 bellows, d200, and 80mm Rodagon, here's a shot of the bible:

http://i955.photobucket.com/albums/ae37/peterdesmidt/Rodagon_Bible_fullframe.jpg

This would produce a picture 8.5x11" at 300 dpi. On the negative the bible is about 5/8ths of an inch long.

Next is a crop of the adjusted Cezanne scan:

http://i955.photobucket.com/albums/ae37/peterdesmidt/Cezanne_Bible_whole.jpg

Peter De Smidt
3-Jan-2012, 20:57
Here is a close-up of the hodge-podge scanner "scan":

http://i955.photobucket.com/albums/ae37/peterdesmidt/Rodagon_Bible_Crop.jpg

And here is a crop of the Cezanne scan:

http://i955.photobucket.com/albums/ae37/peterdesmidt/Cezanne_Bible_detail.jpg

Besides making correct comparisons, which I doubt that I did, the biggest challenge was focusing.

argos33
3-Jan-2012, 21:24
This is something I too have thought about, and may give it a try at some point. After seeing all the amazing macro work done with microscope objectives I would consider using some of those for the project. There are supposedly pretty good ones that can be bought new for $60-$150, but I do not have experience using them.

Consider Nikola Rahmé's work for instance:

http://www.flickr.com/photos/eurythyrea/sets/


It seems to me that you could make this setup more rigid, and make a simple but sturdy stepper-motor controlled stage with a light source underneath it:

http://www.flickr.com/photos/eurythyrea/6618862065/in/photostream

The biggest problem I would foresee with using these microscope objectives would be the incredibly narrow depth of field. I wonder if focus stacking software (like this guy and other macro shooters use) would be confused by the 2D surface or not?

As others pointed out, I would be worried about the stitching software being able to handle it. However that is one area of software development that seems to be progressing quite fast, and I'm guessing it will only get better and better.

Evan

argos33
3-Jan-2012, 21:31
Wow Peter! That is surprising. Perhaps there is some merit in this idea after all - thanks for the quick comparisons.

Mike Anderson
3-Jan-2012, 21:46
Here is a close-up of the hodge-podge scanner "scan":

....
And here is a crop of the Cezanne scan:

....
Besides making correct comparisons, which I doubt that I did, the biggest challenge was focusing.

That's very promising. The sample from that hodge-podge setup you put together is close to the Cezanne's. A Nikon d5100 ($700, 16meg) in that setup might best the Cezanne.

Imagine what this coming year's 24meg+ mid level cameras would do.

...Mike

el french
3-Jan-2012, 22:51
Using a light box, pb-4 bellows, d200, and 80mm Rodagon, here's a shot of the bible:


What was the aperture and bellows extension?


That's very promising. The sample from that hodge-podge setup you put together is close to the Cezanne's. A Nikon d5100 ($700, 16meg) in that setup might best the Cezanne.

Imagine what this coming year's 24meg+ mid level cameras would do.

...Mike

Adding more pixels may not make any difference at all if it's diffraction limited. Simple solution is to get a faster lens :)




This is something I too have thought about, and may give it a try at some point. After seeing all the amazing macro work done with microscope objectives I would consider using some of those for the project. There are supposedly pretty good ones that can be bought new for $60-$150, but I do not have experience using them.


Generally, regular lens work better than microscope objectives up to 5X. At 10X the microscope objectives are better. A Nikon BD Plan 10x microscope objective's DOF is about 12 microns. Film is about 18 microns thick. A 10X 'scan' of an 8X10 using a 4/3rds sensor would need 360 odd frames without focus stacking and at least 720 frames with focus stacking. An APS-C sensor would only be about 10% more efficient.

Frank Petronio
4-Jan-2012, 06:02
Ctein is talking about this on http://theonlinephotographer.typepad.com/the_online_photographer/2012/01/scan-film-with-camera-1.html but at the rate he plods along it will be Summer before he makes an exposure.

Freaking bearded men are slow as molasses. He should get a pipe and get even more relaxed.

Peter De Smidt
4-Jan-2012, 07:14
Front of the camera to the back of the lens was about 180mm. The lens is an 80mm Rodagon with a fixed aperture. It doesn't say what it is, but my guess is the lens was designed for slide duping at 1:1, with an aperture of F4, but I could easily be wrong. Just in case Bob reads this, it's serial number is 10451006.

Regarding microscope objectives: first off, that's some really neat photography. Second, I'm leaning to el french's view that it might be a little overkill, but if someone has one of these, by all means give it a go. It could be just the ticket for scanning 35mm Tech Pan. I did learn that some of Canon's cameras have a special silent shutter mode that really cuts down on shutter induced vibration. Naturally, I'm committed to the Nikon system. I shot at 1/4 second (main exposure) with an electronic cable release and mirror lock up. As a result, the results could've easily been affected by shutter vibration, especially since I was shooting with a macro bellows on top of a manfrotto 410 head, which was attached to a super clamp hanging on the center post of my tele Studex.

In my tests, I expect that imprecise focusing (and vibration) limited the obtainable quality, and the necessity to re-sample the image limited the usefulness of the comparison shots.

My conclusions, nonetheless, are that the quality was close enough to a very good scanner that the idea still has merit. In addition, alignment was easier than I thought. I simply lowered the lens to light table surface and used the 410's control to make the front of the lens parallel to the light table, and this was just by eye-balling it.

Ben Syverson
4-Jan-2012, 09:49
That's 125 sensels/mm, which can record half that at best (60 lines/mm), and that's only in the horizontal and vertical directions. That's my 1600.

Rick, your math is still going awry somewhere. Nyquist tells us that 60 lp/mm requires at the very least 3048 samples per inch (60 * 2 * 25.4).

The original 5D has 121.3 x 121.3 pixels per mm, which as you're suggesting works out to a theoretical max of 60 lp/mm. However, the OLPF, Bayer array and general fudge factor haven't been factored in. Per DPReview, the final number is around 42 lp/mm.

An Epson may be able to capture 42 lp/mm, but it would be buried under such a blurry haze that you'd need to sharpen it to oblivion to find that detail. The Canon gives you that information in a crisp, high-contrast, low-noise image.

I should note that there is an expensive lens (MP-E 65mm) for the Canon system that allows for 5X magnification, so with the 5D, you should be able to capture your negatives at up to 210 lp/mm.

Peter De Smidt
4-Jan-2012, 12:13
The MP-E 65mm is a nice looking lens, and it's more compact in the field than a bellows unit, but other than that, what would be the advantage of using it over an enlarging lens on a bellows?

Am I right to assume that at 1:1, there's no advantage to reversing a lens?

rdenney
4-Jan-2012, 12:25
Am I right to assume that at 1:1, there's no advantage to reversing a lens?

Sounds right. But reversing an enlarger lens optimized for 1:5 would work at 5:1, which might be necessary. But I'll give a try this weekend with a cheap lens that works at 1:1, and compare the results to my scanners. Then I'll know.

Rick "needing a bit of empirical insight with what I can afford" Denney

Peter De Smidt
4-Jan-2012, 13:03
It looks like the fixed aperture on my Rodagon is F8.5, and so a 50mm 2.8 enlarging lens should do better, especially reversed. Unfortunately, I don't have one any more.

For sensor illumination, the effective f-stop is more important than the set f-stop, but does a smaller effective f-stop also reduce lens aberrations, as stopping the lens down would do?

This isn't directly helpful, but it's fun: http://www.youtube.com/watch?v=XeXjpZpaZns

The following might be more useful: http://thegreatgeekery.blogspot.com/2011/04/auto-macro-stacker.html

See also: http://www.ebay.com/itm/4-Z-AXIS-SLIDE-CNC-ROUTER-PLASMA-linear-actuator-/130625119045?pt=LH_DefaultDomain_0&hash=item1e69dd2345

http://www.ebay.com/itm/X-Y-Z-Axis-Lab-Project-Table-Fixture-W-Animatics-SmartMotors-CNC-Mill-/110781462379?pt=LH_DefaultDomain_0&hash=item19cb16f76b

Mike Anderson
4-Jan-2012, 14:33
Ctein is talking about this on http://theonlinephotographer.typepad.com/the_online_photographer/2012/01/scan-film-with-camera-1.html but at the rate he plods along it will be Summer before he makes an exposure.

Freaking bearded men are slow as molasses. He should get a pipe and get even more relaxed.

There are some good tips in the comments to that article. It looks like many people are working in this direction.

...Mike

Ben Syverson
4-Jan-2012, 15:28
The MP-E 65mm is a nice looking lens, and it's more compact in the field than a bellows unit, but other than that, what would be the advantage of using it over an enlarging lens on a bellows?
I would imagine the advantages are rigidity, better alignment of sensor plane and focal plane, and better optical performance... Enlarging lenses are not required or expected to perform as well as macro lenses at 1:1.

But there are some really great enlarging lenses, and I do think you could have very good results with a bellows unit on a DSLR. It just adds a level of complexity which is in my view unnecessary. Good 1:1 lenses are cheap.

John Whitley
4-Jan-2012, 15:33
So instead of investing your $1,000,000 dollars in R&D

This is straw-man nonsense. The whole point of this thread is that the vast majority of the R&D has already been done, and is continuing to be done, by the major digital camera and sensor manufacturers. Getting the details sorted is likewise also in progress by high-end professionals. To wit, see the article The Future of Scanning (http://theagnosticprint.org/future-of-scanning/) by Willam Blackwell at The Agnostic Print. A quote below to whet appetites; see the article for the gory details:



I’ve recently built such a system for the University of Vermont Slide Library. Where MIT spent many “cutting edge” years hand-scanning every slide in their archive, we hope to digitize our 150,000 slides in under three months (pre metadata inclusion) with only $3400 in equipment. That is all made possible by the full frame pro dSLR.


Acknowledged, this is "merely" for 35mm capture. Nevertheless, that's all that's needed to prove the principle. The rest is a fairly boring matter of engineering. Depending on how good the copy camera is, the need for a scanning-stitching station may also go away (see the article for more on that as well).

tlitody
4-Jan-2012, 15:47
This is straw-man nonsense. The whole point of this thread is that the vast majority of the R&D has already been done, and is continuing to be done, by the major digital camera and sensor manufacturers. Getting the details sorted is likewise also in progress by high-end professionals. To wit, see the article The Future of Scanning (http://theagnosticprint.org/future-of-scanning/) by Willam Blackwell at The Agnostic Print. A quote below to whet appetites; see the article for the gory details:


See post #2 of this thread and then post #1 of this thread for the source and direct your complaints there.

Peter De Smidt
4-Jan-2012, 16:03
<snip>Good 1:1 lenses are cheap.

Ben, which ones do you have in mind? I agree that avoiding a bellows is a good idea, if the requisite magnificantion/quality can be gotten by other means. I have a 55mm Nikkor 2.8 P Auto. It'll go to 1:1 on it's own, but it is optimized (according to online discussions) for 1:10. I'll still give it a try.

Ben Syverson
4-Jan-2012, 20:10
Ben, which ones do you have in mind? I agree that avoiding a bellows is a good idea, if the requisite magnificantion/quality can be gotten by other means. I have a 55mm Nikkor 2.8 P Auto. It'll go to 1:1 on it's own, but it is optimized (according to online discussions) for 1:10. I'll still give it a try.
I've been using the Sigma 50/2.8, and it's fantastic at 1:1 and ƒ/8... I bet the 55 Nikkor is similarly great—I wouldn't be surprised if the bit about it being optimized for 1:10 is just an internet yarn. Extreme corner performance is not really an issue, because you'll be overlapping frames anyway. You're throwing away all of your corners.

All of these macro designs are able to be much better corrected than general purpose lenses, because no one expects them to be faster than ƒ/2.8. The lens designer's ƒ/1.4 handcuffs have been removed. (It's strange that in an age with zillion megapixel cameras, great ISO 3200, and highly advanced lens design software, lenses keep getting faster rather than better. I would love to have a Canon 35mm ƒ/4 with an asphere or two. It would be the smallest and lightest SLR lens, but far and away the sharpest. But I digress.)

Peter De Smidt
4-Jan-2012, 21:45
Hi Ben,

That sounds reasonable. Since one of the major reasons for this project is affordability, looking at extremely expensive (and rare) lenses doesn't make much sense.

I've been reading about microscope eyepieces. See, for example:http://coinimaging.com/Lens_tests.html He ends up mostly using micro Nikkors for his own photography. In addition, according to http://www.microscopyu.com/articles/optics/objectivespecs.html, "Microscope objectives are usually designed to be used with a specific group of oculars and/or tube lenses strategically placed to assist in the removal of residual optical errors." While some of the newer ones might be fully corrected, there's no guarantee that they'll be better than the standard macro lenses in our application. (If someone has some of these, by all means test them.)

I'm going to test my 55mm Nikkor, which appear to be common and available for $100-150, and I have a reverse adapter on the way to see if that helps. (They're about $3 shipped on Ebay.) I'll also test a componen-S 50mm, but I'll have to make a custom mount.

Peter De Smidt
5-Jan-2012, 15:23
I've had some correspondence with an expert on macro lenses. He built just the type of system we're suggesting for a client. Although he can't share details, both he and the client were very happy with the results. He said that he's not a big fan of microscope objectives "...as their projected image is rather small (18-20mm), need long tubes (160-210mm) or additional hard to find tube lenses (for infinity corrected objectives) and the biggest problem is their very shallows DOF, so each ever so slight non-planarity of your film kills the result."

I've been thinking about how we should attach the camera to the support structure. If we use a macro unit with bellows that allows camera positioning, that's fairly easy, but if we want to avoid bellows use, we'll need a way to accurately position the camera.

My idea is to use a cheap arca compatible clamp and nodal slide, along with an inexpensive camera plate. The clamp ($30 or more) would get bolted to the front of the support structure in a vertical direction.

http://i955.photobucket.com/albums/ae37/peterdesmidt/smallclamp.jpg
[Pictures used with permission]

This would allow the Nodal slide, such as http://www.ebay.com/itm/SUNWAYFOTO-DMP-100-100mm-Rail-Nodal-Slide-w-Arca-Compatible-Clamp-Sunway-NEW-/360398724446?pt=LH_DefaultDomain_0&hash=item53e9701d5e, to move up and down in the clamp.

http://i955.photobucket.com/albums/ae37/peterdesmidt/shortslide.jpg

We could adjust the clamp around the bolt axis, perhaps using set screws along the long edge of the clamp, one near the top and one near the bottom, to allow fine adjustments. To adjust the camera tilting front and back, I plan on having bolds in threaded holes at the bottom of each of the legs of the support structure.

The issue left would how to very careful adjust the position of the camera plate in the clamp. This would be used to move the camera closer or farther from the film. There are some very expensive lead screw focusing rails, such as: http://reallyrightstuff.com/ProductDesc.aspx?code=B150-B&type=0&eq=&desc=B150-B%3a-Macro-focusing-rail&key=it, that could replace the slide and clamp, but they're pretty expensive. My idea is to use a threaded rod (bolt, knob, ...) in a block bellow the clamp. Turning the rod will move it up and down. If it is situated so that mates against the the front of the nodal slide and it clears the clamp, it could be used to adjust the position of the slide in the clamp quite precisely.

Does that sound reasonable? Suggestions?

Jim Michael
5-Jan-2012, 16:04
I have a device that allows x/y positioning of a camera mounted on a tripod. Looks like it could be used to move a stage around in the x and y axis. Hand operated though. Is there an end goal of an automated solution? Seems like you'd want autofocus working for that, so a macro lens (and tubes if more mag needed) would be the most likely lens setup.

Peter De Smidt
5-Jan-2012, 16:12
Hi Jim. Yes, there's an end goal of automation, but I don't think we'll need auto focus. Instead well keep the negative very flat, using a clam shell holder, and move the holder on a very flat surface, i.e. a sheet of thick glass.

Hi Mike, that's exactly what I had in mind.

Mike Anderson
5-Jan-2012, 16:33
...
Hi Mike, that's exactly what I had in mind.

I realized after I posted that you'd pretty much covered what I said so I deleted my post.

Anyway, here's a cheaper focusing rail (http://www.adorama.com/MCFRS1.html), but it's got the unnecessary extra direction of movement. And here's a real cheap one (http://www.ebay.com/itm/200597578062?ssPageName=STRK:MEWAX:IT&_trksid=p3984.m1423.l2649#ht_3676wt_1097).

...Mike

Jim Michael
5-Jan-2012, 17:49
Hey Mike. Great minds think alike, the device I was describing looks like that first link.

I'm going to build something but I think I'm going to stick with the plate moving over predefined positions. My gut tells me that autofocus will be needed on an automated system. You'll want the sweet spot aperture on the lens, so maybe 5.6, and seems pretty difficult to totally eliminate film curvature, plus general inaccuracy in the system. But I'm no engineer and hope someone comes up with a cool automated solution.

Ben Syverson
5-Jan-2012, 20:01
ƒ/5.6 is often the sharpest aperture on a standard 50-60mm macro, but there are two compelling reasons to choose ƒ/8 instead: increased DOF and reduced vignette.

By ƒ/11 you really begin to see the destructive effects of diffraction.

Autofocus is probably a bad idea in an automated set up. If the macro lens can't immediately lock the focus, it may rack back and forth over a gigantic range from 1:1 to infinity. At the end, it may refuse to take the photo, leaving you with a gap in your tiling sequence.

Nathan Potter
5-Jan-2012, 20:07
As John Whitley suggests, this is really only an engineering and construction job using as many OEM parts as possible. The choice of lens depends only on the image quality needed (resolution) that is compatible with the sensor resolution. B&W only is intriguing because monochromatic illumination would be an option using extreme resolution lenses that would considerably exceed the resolving power of standard macro color optics. Something like the Ultra Micro Nikkor comes to mind with a FOV of around a cm. or so and capable of maybe 800lp/mm. at 450 nm. In such an instance it would have to be used at real magnification so that the image detail is compatible with the sensor pixel pitch. That would far exceed any Aztek machine except for Dmax limitations inherent in the sensor. Of course magnification means a smaller Field Of View for fixed size sensors so increased number of steps.

Nate Potter, Austin TX.

Peter De Smidt
5-Jan-2012, 20:28
Nate, I've read about those ultra-micro-Nikkors, but I have a question: If we use a monochromatic light source, won't we lose input from a large number of the imaging chip's sensors?

Another question: what depth of field will we need assuming we keep the film as flat as in a good glass sandwich carrier?

el french
5-Jan-2012, 23:06
Nate, I've read about those ultra-micro-Nikkors, but I have a question: If we use a monochromatic light source, won't we lose input from a large number of the imaging chip's sensors?

Another question: what depth of field will we need assuming we keep the film as flat as in a good glass sandwich carrier?

17 microns minimum.

Jim Michael
6-Jan-2012, 06:54
Autofocus is probably a bad idea in an automated set up. If the macro lens can't immediately lock the focus, it may rack back and forth over a gigantic range from 1:1 to infinity. At the end, it may refuse to take the photo, leaving you with a gap in your tiling sequence.

Yes this is a risk with AF. I've seen the same behavior with my Gigapan unit and have to monitor it during exposures. The Gigapan allows you to pause and back up to the previous frame so you can correct the problem. Even with a non-AF setup I think you'd want a pause & backup option, or else a means to reset position to a particular x-y coordinate in your grid to reshoot a matrix element.

Ben Syverson
6-Jan-2012, 09:58
Unfortunately that means you wind up with an interface as convoluted as the Gigapan's...

Nathan Potter
6-Jan-2012, 10:25
Nate, I've read about those ultra-micro-Nikkors, but I have a question: If we use a monochromatic light source, won't we lose input from a large number of the imaging chip's sensors?

Another question: what depth of field will we need assuming we keep the film as flat as in a good glass sandwich carrier?

They would be used in real magnification mode such that the resolution at the object (the film) matches the poorer resolution at the sensor pixel pitch - or there abouts, and yes you only have use of those pixels that detect the illumination source wavelengths.

Mentioned above is 17 um DOF, but that depends on what version of that series we are talking about. Dunno, I don't have my old spec sheet handy but 17 um for any of them seems way too large if one were to spec 800 lp/mm. based on a single image capture.

For that kind of resolution the focus has to be dead on, maybe within a micron or so. But the field of view is so small due to the magnification factor that the file size would be enormous in order to cover an appreciable image size.

The main reason I mention this is to emphasize that there are already a range of imaging lenses available that will approach and exceed the resolution capability of the equivalent drum scanner. Noise floor and Dmax is another issue.

BTW use of such high resolution lenses implies a tight control of other factors such as ambient temperature, vibration and focus maintenance during stepping. A high quality consumer unit is largely free from these tight controls so can be put together at much lower cost - maybe as low as $ 3K.

Nate Potter, Austin TX.

Peter De Smidt
6-Jan-2012, 12:50
Here's another neat video:http://www.youtube.com/watch?v=zyT7l-CZego

In thinking about this on the midday dog walk, it occurred to me that we might be focusing a little too much on ultra-high resolution. While that's certainly a nice goal, a much simpler system might be more practical for most people.

For instance, I don't know about you, but I usually printed 35mm film full-frame on an 8x10 sheet of paper with at least a 1/2" border. Even a d200 can capture enough resolution for that with one shot with a normal macro lens, and it would be extremely fast to operate.

If more resolution is needed, you could always turn the film so that the long end of the sensor is parallel to the short end of the film. Take two shots, sliding the film the required amount. It would be easy to make a holder that allowed very precise two shot negative movement. I bet you wouldn't even need to stitch. With a lowly D200, that's give a file 3872x5832 pixels, which would allow a 10.75" x 16.2" print at 360 dpi., or a 12.9" x 19.4" print at 300 dpi. That's big enough for me, but if it isn't for you, you could always increase the magnification and add more samples. Above about 4 samples, though, a manual slide type system might become a bit tedious. (Four would be easy. Cut a square out of flat material, such as an acrylic sheet, such that the negative holder will be in the right place when you slide it to each of the guides corners. Attach guide to top of glass. ...)

At these lower magnifications, other things might become easier as well. First, as mentioned before, no bellows would be needed, as a standard 50-60mm macro lens would work well. Second, we might get away without using a top cover glass. For example, I used to enlarge medium format film with a black metal mask on the bottom of the carrier and an anti Newton sheet of glass on top. Film flatness was good enough for me at the enlargement sizes I made, and there was less dust. In this case, the AN would be on the bottom, the film base would be up against it, and something like a mask from and enlarger could be placed on top, which would hold the negative down. This should do a good job with all but the most squirrelly film.

We also don't need the latest dslr. Something like a D90 (or Canon's equivalent) would work well, and the cost might be such that we can dedicate it to this use, with the concomitant savings in setup time.

This doesn't mean that we shouldn't pursue higher resolution versions, but the type of system I'm talking about might be more useful for more people, and it would provide a base for hi-res experiments.

Does that sound reasonable?

jon.oman
6-Jan-2012, 12:53
With regard to flatness of the negative/positive, would it be possible (or practical) to use a vacuum on the movable stage?

Peter De Smidt
6-Jan-2012, 13:09
With regard to flatness of the negative/positive, would it be possible (or practical) to use a vacuum on the movable stage?

I don't think so, since we have to be able to shine light very evenly through whatever is holding the negative up.

tlitody
6-Jan-2012, 13:42
So the idea is how might we scan all this existing film, not to mention any new materials, 10-20 years down the road?


If you are really that worried you would get all your negs drum scanned now by one of the cheap asian outfits before their prices go up. If you ain't really worried and are willing to wait 10-20 years, the resolution on a dslr will be so high by then that you won't have any problem photographing your negs from a light table.
So what exactly is the problem?

Peter De Smidt
6-Jan-2012, 13:53
I suggest that we ignore posts from people that seem strangely worried about what we do with our time. If we argue with them we're giving them what they want.

tlitody
6-Jan-2012, 14:15
I suggest that we ignore posts from people that seem strangely worried about what we do with our time. If we argue with them we're giving them what they want.

You start a clique if you want to. I'm just stating the obvious which seems to have escaped you.

sully75
6-Jan-2012, 16:18
I suggest that we ignore posts from people that seem strangely worried about what we do with our time. If we argue with them we're giving them what they want.

I'm always surprised to see an old fashioned internet troll, they are a rare breed these days. Don't miss them much.

rdenney
6-Jan-2012, 16:24
I suggest that we ignore posts from people that seem strangely worried about what we do with our time. If we argue with them we're giving them what they want.

:)

Rick "wondering if that x-y macro platform Mike linked would be precise enough to use for camera movement instead of mere focusing" Denney

Brian C. Miller
6-Jan-2012, 16:24
In regards to autofocus, the only places that focus needs to happen is at the edges. Since the edges are well-defined, then the focus can be locked or calibrated between one edge and the next. The user interface will remain simple.

Frank Petronio
6-Jan-2012, 19:09
If you are really that worried you would get all your negs drum scanned now by one of the cheap asian outfits before their prices go up. If you ain't really worried and are willing to wait 10-20 years, the resolution on a dslr will be so high by then that you won't have any problem photographing your negs from a light table.
So what exactly is the problem?

That you're a tool.

Ben Syverson
6-Jan-2012, 19:25
Tlitlitlody: I think you're right. All I need to do is ship all my negatives to China, or wait 20 years. Both are appealing to me.

el french
6-Jan-2012, 20:12
In regards to autofocus, the only places that focus needs to happen is at the edges. Since the edges are well-defined, then the focus can be locked or calibrated between one edge and the next. The user interface will remain simple.

I think autofocus should only be needed for the first image, if at all. Assuming the film is held flat, there should be no reason to refocus for the next frame. You will need to define tolerances, probably within 5 microns if shooting at 10X and much larger if shooting at 1X.

Moving reliably from one frame to the next can be controlled very precisely using closed loop systems. Inexpensive encoders are available with 14bit resolution. If you need to return to a previous position within 1 pixel, then you will need a system with no backlash.

My macro setup is mounted on an 80 pound block of granite (It's actually an 18" x 24" surface plate with Sorbothane feet.) It is still not rigid enough to eliminate all vibrations. The subjects I'm shooting aren't as stable as a sheet of film should be, so you may be able to get by with less vibration dampening.

The stitching software I use is Microsoft ICE, http://research.microsoft.com/en-us/um/redmond/groups/ivm/ice/. It is very tolerant of alignment mismatches and also has an orthographic stitching mode. Autopano Pro, PTGui, and PTAssembler could also be used. I'm not a big fan of Photoshop for stitching as it doesn't handle very well the types of images I make.

Peter De Smidt
6-Jan-2012, 20:57
With medium and large format film, I'm not sure of what would be gained scanning over 1x magnification, and 2x should be plenty for even the most fine-grained/high detail 35mm frame. The prints would need to be gi-normus. But the nice thing about this setup is with a different lens, and more time, someone could go as crazy as they wanted.

It would be nice, although not a requirement, if we could position the negative precisely enough not to require overlapping frames. We'll have to see.

Regarding vibration, it is an important point. I plan on testing out some flashes to see if they are consistent enough. In addition, we could use mirror lock up, rear curtain sync (or Canon's silent shutter mode), and, say, a 1 second exposure. In that case the shutter would fire, the vibration would go away, and the flash would go at the end of a second. That's a technique that Charles Krebs uses for some of his macros, and it works.

Another option would be to put a shutter bellow the lens. It could be mounted to it's own frame, and so it's vibration would not effect the system. That would be a good use of Polaroid Copal shutters that don't have apertures.

Peter De Smidt
7-Jan-2012, 08:55
The macro folks say that these stages work very well for macro photography, including straight down: http://www.velmex.com/manual_cross_sections.asp?series=2500 They're supposed to be very rugged, and they're a bit cheaper than RRS. With the preceding link, I was thinking of precise camera movement.

Look here for xy negative stage movement: http://www.velmex.com/xslide/motor_xslide.html

A 12" one axis motorized slide is $567 new. A 6" if $438.

Mike Anderson
7-Jan-2012, 09:47
The macro folks say that these stages work very well for macro photography, including straight down: http://www.velmex.com/manual_cross_sections.asp?series=2500 They're supposed to be very rugged, and they're a bit cheaper than RRS. With the preceding link, I was thinking of precise camera movement.

A cheaper option would be to be to have gross imprecise positioning at the camera and use the feet adjustments for precise adjustment.

...Mike

Kirk Gittings
7-Jan-2012, 12:23
There is a similar discussion going on over at Luminous Landscape.
http://www.luminous-landscape.com/forum/index.php?topic=60643.0

Mike Anderson
7-Jan-2012, 13:09
There is a similar discussion going on over at Luminous Landscape.
http://www.luminous-landscape.com/forum/index.php?topic=60643.0

You posted a link to this thread over there?!? Those guys will come over here and steal all our secrets!

Our thread was started first!

Seriously, it's interesting how all this is emerging at roughly the same time.

...Mike

Kirk Gittings
7-Jan-2012, 13:13
Me too. A sea change is underway which I think will lead to some real solutions.

sully75
7-Jan-2012, 13:31
Just a thought re: the stitching issue...what if you zoomed out for whatever camera you were using and got a full copy of the whole negative. Then you could bring that and your closeups into photoshop and tile the closeups over the full shot.

Peter De Smidt
7-Jan-2012, 17:23
A cheaper option would be to be to have gross imprecise positioning at the camera and use the feet adjustments for precise adjustment.

...Mike

I agree, but if you change magnifications a lot, one of these could be useful.

jb7
7-Jan-2012, 17:45
The macro folks say that these stages work very well for macro photography, including straight down: http://www.velmex.com/manual_cross_sections.asp?series=2500 They're supposed to be very rugged, and they're a bit cheaper than RRS. With the preceding link, I was thinking of precise camera movement.

Look here for xy negative stage movement: http://www.velmex.com/xslide/motor_xslide.html

A 12" one axis motorized slide is $567 new. A 6" if $438.

I think that that level of expense for half a positioning stage is wholly inappropriate. Totally over-specified for the load too; how much does a sheet of film weigh? I doubt if it would be more accurate than software, for accurate positioning down to pixel level, though I'm not entirely sure how that might work- particularly when there isn't any definable detail to stitch together, such as in continuous tones, like areas of sky, for example.

Grain, at a microscopic level, will surely have different shapes, at the edge of a frame, depending on lens projection and lighting? So maybe standard stitching software might also break down? Perhaps the order of the exposures could be incorporated into the stitching software, to provide a guide?

Does the gigapan software take account of the physical orientation of the camera, as well as the picture elements?

Peter De Smidt
7-Jan-2012, 18:14
Hi Joseph, I agree for the most part, although a sandwich glass carrier for 8x10 might weigh a fair bit. It's basically a worst case price-wise scenario of having to buy something completed. I expect that we can come up with something similar for less, although perhaps we could save with a group buy. Even if we spent $1200 on two of these, we should still be able to bring the total cost to $1500-$1800, minus the dslr. That's still a lot less than most 8x10 capable film scanners.

jb7
7-Jan-2012, 18:22
I posted a link, earlier in this thread, for a software controlled, threaded rod, stepper motor driven stage, which would cost less than $100...

Peter De Smidt
7-Jan-2012, 18:30
Sorry Joseph, my memory is questionable. I'll look for the link.

Basically, what I've been doing is investigating options. I'm not advocating any one system. The idea is to come up with some possibilities, hoping that they might either turn out to be viable, or provide insight into what might be viable. We have a lot of experimenting and investigating of what to do, and I expect that people will come up with and build various designs. For example, I expect that what would work well with 4x5 negatives might not be the same hardware that works well with ULF negatives.

Kirk Gittings
7-Jan-2012, 18:44
This surprises me:
New Plustek Scanner (http://www.geardiary.com/2012/01/04/plustek-to-unveil-new-line-of-scanners-at-ces/)

jb7
7-Jan-2012, 18:49
I'm looking into doing ULF, but I doubt if I'd want to scan that, that would be for reproduction at same size. I think 8x10 would be the largest size anyone would seriously want to enlarge, although the exception will prove the rule...

I think two sizes, up to 4x5, and up to 8x10, would be all that would be necessary- if someone wanted to go bigger, then it could be scaled up-

el french
7-Jan-2012, 20:07
I posted a link, earlier in this thread, for a software controlled, threaded rod, stepper motor driven stage, which would cost less than $100...

Could you repost? It seems to have gone missing.

I just recieved a linear slide from Igus, http://www.igus.com/wpck/default.aspx?pagename=filmtechnology. It's 1 meter long and price was about $125. I think it would work quite well for this application.

jb7
8-Jan-2012, 05:51
As has been mentioned, a high degree of accuracy is not needed, the stitching software can take care of that. Perhaps the RepRap 3d printer platform (or something similar) could be modified to provide the moving baseplate? It's a very cheap system, proven technology, both hardware and software, and not difficult to imagine how it might be adapted to accept a light box and camera mount- plus, it's all open source-


http://reprap.org/wiki/RepRap

This was the post I made, in relation to the positioning stage, which could be adapted from the 3d printer linked to above-

Thinking about it further, and taking account of the depth of field requirements as stated earlier, I'm not sure that any of these approaches are on the right track. There's got to be a better way...

That's a good looking rail...

Peter De Smidt
8-Jan-2012, 18:47
I built a support structure today, and hopefully I'll be able to take some tests with my 55 micro tonight, if I can get everything aligned. That's non-trivial. I also bought an inexpensive 4x microscope objective, one that the macro folks have had good results with, but I'll have to wait until an RMS adapter arrives to give it a try.

argos33
8-Jan-2012, 20:31
Peter, if you have a chance post some pictures of the setup. I would love to see what you come up with. I am working on finishing some other projects right now, but plan on making my own prototype as well.

Since I work in a metal/machine shop I think I can make a nice sturdy prototype without too much trouble. I am also learning about programming micro controllers like the Arduino, and eventually want to make an automated stage for the film. It seems the programming for the stepper motors isn't too complicated, however I am still learning and it may take awhile before I have something working. I'll keep everyone posted of my progress once I get going on it.

Evan

Peter De Smidt
8-Jan-2012, 20:44
Hi Evan,

That sounds terrific! Please keep us updated.

Here a full-frame picture from the 55 nikkor at 1:1:

http://i955.photobucket.com/albums/ae37/peterdesmidt/55_Micro_Whole.jpg

Below is a crop to the bible:

http://i955.photobucket.com/albums/ae37/peterdesmidt/55_Micro_bible.jpg

And finally an up-close crop of the text:

http://i955.photobucket.com/albums/ae37/peterdesmidt/55_Micro_close.jpg

rdenney
9-Jan-2012, 00:56
I've conducted my own proof of concept, and I think there is real potential.

I tried it with my Tamron SP 90mm macro, which is outstandingly sharp...in the middle of the frame. Or perhaps the field is not quite flat. In any case, it broke down in the corners. So, I switched to an EL-Nikkor 105/5.6. I have it mounted on a Pentacon Six body cap, which is mounted in a Pentacon Six bellows, and then mounted on my Canon 5D MkI using an adapter. I mounted the camera on a tripod adjusted to lean over my light table, and then used a telephoto lens stabilizer cradle from the front of the bellows to rest solidly on the light table. Focusing was...difficult. The camera needs f/5.6 for focus confirmation, and at near 1:1, the light getting to the camera was more like the equivalent of f/11. I used an eyepiece magnifier, and I have a Maxwell screen in my Canon which helps with focusing manually. I made exposures at f/5.6, f/8, and f/11, and f/11 was the best of the lot, once I figured out how to stabilize the camera.

My light table is a cheapie, and would not at all be evenly illuminated enough for this application. But it worked okay for the test. I just laid the negative on the light table.

The first image is a scan of the whole image made in the Epson V750 at 2400 spi. I adjusted the image according to my usual workflow. I made the photograph some years ago of a portion of the famous front door of the Espada Mission in San Antonio. Lens was a Schneider Super Angulon 121mm f/8, at f/32 with a 1-second exposure on FP4, developed in HC110 Dilution B to n+1. At f/32, there is a touch of visible diffraction, but that's what I needed for depth of field--that stone wall is a foot thick. I've made an enlarger print of this negative at 16x20, and it is extremely sharp at that print size.

The upper half of the second image is a full 24x36mm photograph of a piece of that image made using the 5D with the EL-Nikkor enlarging lens, as described above. Actual magnification was not quite 1:1--more like 1:1.15--which was as racked out as that bellows will go. I need a short extension tube (which I could not put my hands on at the moment) to get it all the way to 1:1. The effective resolution of the scan is 2670 spi--just a little more than the Epson. For comparison of general tonality, I've included the same piece of the image snipped from the Epson scan in the lower half. It's a bit smaller because of the difference in resolution. Yes, the upper image looks smoother and clearer to me, too.

I inverted the image and adjusted curves. I then applied a small amount of sharpening (0.8 radius at 100) to try to match the effect with the Epson. There is no sharpening at the downsampled resolution. Out of the Canon 5D, the histogram only covered the middle third of the tonal range output by the camera. I use RAW and expanded the range somewhat during the raw conversion, and then more in Photoshop.

The third image is the scan using the Canon 5D/EL-Nikkor shown in actual pixels. The fourth image is the scan using the Epson shown in actual pixels. There is no doubt that the Epson's butt is getting kicked here. Yes, the scan using the DSLR is a little bit lighter, which is, I think, a hot spot on my light table. But there is no doubt that the DSLR scan is seeing the tonality of the negative much better, and also retaining information in the thinner bits. And there is no doubt that grain is visible in the DSLR scan and not in the Epson scan.

What have we learned? 1.) This will work, even with modest cameras and affordable lenses. Enlarging lenses are good enough, at least at 1:1 or so using a camera with an 8-micron sensel. 2.) Rigidly mounting the camera is critical. My tripod alone was not sufficient--the camera required two points of support to be really stable. I used the self-timer to let the camera settle down before making the exposure, but it wasn't enough without two points of support, at least with the camera cantilevered over the light table. There is no question that moving the film will be the preferred approach rather than moving the camera. 3.) Forget autofocus. At f/11 (indicated) and 1:1, the camera is only seeing the equivalent of f/22, and the autofocus will not work. 4.) a lens with a flat field is critical. 5.) Dust is no less challenging with this approach than with any other. I spotted the Epson scan, but I didn't bother with the 5D scan and you can tell. 6.) I'm convinced that it is possible to significantly improve on an Epson V750 using this approach--my results seem to me a floor with lots of room to do better with the right apparatus.

Rick "less interested in what's possible with the state of the art and more interested in what's possible with what I already own" Denney

jb7
9-Jan-2012, 06:39
Good work, looks very promising- if I had a Micro Nikkor I'd be testing myself. I suppose the next stage will be to try a rudimentary stitch...

Peter De Smidt
9-Jan-2012, 07:44
Rick,

Thank you for doing you tests!


<snip> Focusing was...difficult. <snip>

Absolutely! I'm hoping that live view helps with that. Additionally, If one changes magnification regularly, some type of lead screw positioning system would be very helpful.

I agree with your conclusions.

Here's a very quickly built test bed:
http://i955.photobucket.com/albums/ae37/peterdesmidt/Front_Stand.jpg

http://i955.photobucket.com/albums/ae37/peterdesmidt/Stand_2.jpg

http://i955.photobucket.com/albums/ae37/peterdesmidt/Stand_wCamera.jpg

I used an F100 as a step in, as I needed my d200 to take the pictures. When I took the "scans", I aligned everything with a Versalab Parallel, although this is a worst case scenario for the laser alignment tool, since the distance from the front of the lens to the Parallel was so short. I also completely covered the light box except for the area being photographed.

rdenney
9-Jan-2012, 08:46
Good work, looks very promising- if I had a Micro Nikkor I'd be testing myself. I suppose the next stage will be to try a rudimentary stitch...

For me, not until I can devise a more even light source. Uneven illumination will be disastrous when stitching. It's almost tempting to build this over the base of something like a Bowens Illumitran. One advantage to using a DSLR is that the color temperature is easy to correct even when scanning color film. But it has to be even.

Live view might help with focusing, but I'm hoping the apparatus will be precise enough not to need refocusing after the initial focus setup.

For me, the next step is to wait for someone to construct a machine. While it needs precision for focusing and orthogonal alignment, I don't think it needs automation for the tiling process. I don't think it would be any more difficult to move the rails manually and stitch the results manually using something like Panavue or PS than it would to maintain a drum scanning capability. And I'm thinking 4x5 needs to be the target. The Epson is much more satisfying for 8x10, but 4x5 is in a gap between available film scanners and enough real estate not to need more than Epson resolution.

Rick "lacking the resources to construct the film moving stage and camera support" Denney

Mike Anderson
9-Jan-2012, 09:32
Or crazy idea #17:

In the comments to Ctein's article (http://theonlinephotographer.typepad.com/the_online_photographer/2012/01/scan-film-with-camera-1.html) someone mentioned using a computer display as the light source. Not sure if it was really successful.

But if it is feasible it could help with alignment. Instead of displaying it's whitest plain field, the monitor could display a sparse scattering of alignment marks which would show through film and make stitching bulletproof. So 2 captures, 2 sets of tiles would be made, one with the alignment marks and one without. The stitching software would have no problem stitching the patterned tiles, even if film image is a solid gray. Having stitched together an image containing alignment marks, the software would use data from the unmarked tiles patch in the over the alignment marks.

Of course the control software is more complicated now having to coordinate the transport, the camera and the monitor-light-table. And custom stitching software would be needed.

I know it sounds crazy I'm just throwing it out there.

Peter De Smidt
9-Jan-2012, 09:55
While it needs precision for focusing and orthogonal alignment, I don't think it needs automation for the tiling process.

I agree. As long as there's a good guide to slide the negative holder against, many people should be happy with the quality available with manual positioning, especially with 35mm and medium format.

My next steps are to built a thick glass stage to slide the carrier on. The top of my light box is not all that flat. Once that's done, I'll build a guide system to allow simple stitches. After that, I'll experiment with light sources, starting with a light mixing box from an old enlarger that I don't use anymore and a flash.

jb7
9-Jan-2012, 09:58
That test bed looks like it's on the right track- I suppose 3 adjustable feet on the film stage might allow for another planar adjustment.

Rick, while automating the positioning stage might be unnecessary, it might also make the job of scanning into much less of a chore, depending on how many scans you have to make. Scanning fatigue might be a real problem, particularly if it's compounded with stitching fatigue- which may turn out to be a bigger problem than focusing-

sully75
9-Jan-2012, 10:19
Peter in checking out your system, I wonder if flare might be an issue? Perhaps a mask over the negative would be necessary for better results? Just a thought.

It would be nice to include 5x7 in all of this, which is on the bleeding edge of adequate with Epson scanners but would be well served by this method.

Peter De Smidt
9-Jan-2012, 10:27
Peter in checking out your system, I wonder if flare might be an issue? Perhaps a mask over the negative would be necessary for better results? Just a thought.

Absolutely. For the tests, the entire light box is masked off over the negative, such that only a space a tiny bit bigger than then imaged area is not blocked.

The system should be scalable to whatever size is needed. The glass table would need to be at least twice as big as the negative.

rdenney
9-Jan-2012, 11:10
Rick, while automating the positioning stage might be unnecessary, it might also make the job of scanning into much less of a chore, depending on how many scans you have to make.

Given my generally low productivity, and the adequacy of the Epson for many routine scans, and the negligible size of my pot of disposable cash these days, I'll take a cheap but labor-intensive apparatus over an automated but expensive product any day.

But let's prove the viability of the whole process first, and start putting it into practice so we can really understand our needs and requirements, and then the entrepreneurs out there will know what to build so that it will be the right product for the cheapest cost.

Rick "remembering a slab of granite and a Corian table top in the shed that might prove useful, seeing Peter's prototype" Denney

rdenney
9-Jan-2012, 11:20
Or crazy idea #17:

Computer displays have a pronounced pattern that would have to be profoundly out of focus, negating any value for producing alignment marks. Those can be made on the glass stage if they prove useful in any case. And then you would have problems with the timing of the raster scan on computer monitors.

An LED bank with the right sheet of diffusion glass might be better. Or, that same diffusion glass over the face of a Vivitar 285 with the manual power control.

Rick "noting that not much illumination is actually required" Denney

Jim Michael
9-Jan-2012, 14:14
As was previously suggested it might pay to use a collimated light source. I think the 2 lenses from my D2 might be overkill though.

rdenney
9-Jan-2012, 16:49
Is there a reason this is in the Lounge? Seems to me this is a legitimate Digital Hardware topic.

Rick "thinking it bears on a couple of questions asked just today by people who would not have seen it in a search" Denney

Mike Anderson
9-Jan-2012, 17:17
Is there a reason this is in the Lounge? Seems to me this is a legitimate Digital Hardware topic.

Rick "thinking it bears on a couple of questions asked just today by people who would not have seen it in a search" Denney

So the guys at Luminous Landscape can't come over here and steal our ideas. Although we could probably release crazy idea #17 to the public without it compromising the operation.

Peter De Smidt
9-Jan-2012, 18:23
For those with a Nikon, this software looks interesting: http://www.controlmynikon.com/purchase.html

Has anyone used it?
Peter

Nathan Potter
9-Jan-2012, 19:31
I know nothing about stitching images but I find it hard to believe that density variations from frame to frame can't easily be compensated by existing software. Also alignment in areas of poor detail can be done by pin pricking the film in those areas randomly. If you don't like damaging your film use a separate clear sheet with pin pricks or tiny opaque spots that can be cloned out in PS later.

Seems clear that DSLR copy technique is highly feasible and likely considerably superior to existing Epson 750 scanner and maybe even a Nikon 5000 ED.

It boils down to mostly a mechanical engineering job depending on how sophisticated one wants to get. Possibly a nice custom business for someone younger than me.

Nate Potter, Austin TX.

Ben Syverson
9-Jan-2012, 19:38
There is absolutely no need to damage the film in order to aid alignment... People have trouble stitching digital frames without much contrast (like big areas of sky) because digital files have extremely low noise. Film has grain and dust, both of which can serve as anchors. Dust in particular will be a great aid in stitching.

Jim Michael
9-Jan-2012, 20:21
Wait till Frank sees the bill from the engineering department.

el french
9-Jan-2012, 21:59
There's no reason the camera has to be above film. Just put it where the paper usually sits on your enlarger :) Now you should have a nice flat film holder and even lighting.

Corran
10-Jan-2012, 00:53
Genius! I'll have to rig some kind of camera holder to try that. I guess with b&w film you could change the contrast with filters too.

Frank Petronio
10-Jan-2012, 02:43
Is there a reason this is in the Lounge? Seems to me this is a legitimate Digital Hardware topic.

Rick "thinking it bears on a couple of questions asked just today by people who would not have seen it in a search" Denney

Cause I started the thread thinking you Einsteins would laugh at the idea. Of course building a cheap spinning drum scanner was a silly lounge-worthy diversion but now I think you are getting onto something pretty cool and semi-practical!

While I have zero ~good~ engineering input to offer, please keep running with it and hopefully keep it in the shared, open-source, home-brewed skunkworks vein!

Might be good to have two teams working concurrently and get a little competitive action going... this is the Mac team, Luminous can be Lisa.

rdenney
10-Jan-2012, 08:00
Cause I started the thread thinking you Einsteins would laugh at the idea.

A few decades ago, I was known for being persistent on engineering topics (NO! Say it ain't so!). Around that time, the whole chaos theory fad emerged, and some geekezine published pseudocode to draw a Mandelbrot set. Somebody up the line where I worked wanted to program it up but didn't know how.

One day soon after, I found that magazine sitting on my desk, opened to that article, with a sticky note, "Betcha can't get this to work." You know how this story ends. The running joke was that they would walk by my office, throw it in, shut and lock the door, and simply wait.

I have a feeling there are many about whom that story would apply on this forum.

Is this worth doing? Now that we have a sense of the potential with some proof-of-concept testing, it's worth writing down, perhaps.

I think this line of thought got sparked by Ctein, not by LL or by us, though it's apparent that lots have tried it with smaller formats. And it's already been making the circles that using a slide duplicator and a DSLR is an adequate, and far, far more efficient way to scan a lot of, say, color slides. I've considered it myself.

Ctein has recommended that nobody bother with this, even while speculating on what it would take to do it right, but I think in his case he is not considering the value it has for large format, particularly 4x5. There are good and relatively inexpensive options for roll film and small format, and even 2x3 sheet film could be put in a Nikon film scanner. And, as Kirk has shown us, even now a new film scanner of that basic design is being announced.

For 8x10, and particularly for negatives, the Epson is really not a bad choice. The limitation is on density accuracy more than resolution for the size prints most of us are willing to make. People using larger negatives than 8x10 generally contact print, but they can also scan those large negatives in pieces using an Epson and stitch them.

But 4x5 falls into a hole. I get about the same number of pixels (and each pixel is more accurate) scanning 6x7 in my Nikon than I get scanning 4x5 in my Epson.

Using my 13-MP 24x36 camera at 1:1, I would be able to scan 4x5 at 180 megapixels, with pixel accuracy already shown (with cobbled-together apparatus) to be better than an Epson. I think it will support prints up to 30x40 with good performance, at least with negatives.

It would take about 20 images from my 5D, at 1:1, to cover 4x5, with sufficient overlap for stitching. I think I could make those images in about five minutes, with a good mechanical apparatus for moving the film stage. Then, pop the memory card into the computer, batch-convert from RAW with all images using exactly the same settings, and drop them into the stitcher. That would take maybe another 10 minutes. And then push the stitch button to process the stitch and go for coffee, or put the memory card back into the camera and start the next 4x5.

In level of difficulty, this falls between a single-shot flatbed/film scanner and a drum scanner, not to mention making prints under an enlarger. This is definitely in the range of reasonable, for those negatives needing something better than what the Epson provides. And for those who cannot afford sending it to someone like Lenny, or who want to execute the entire process themselves just for their own satisfaction, this process is entirely reasonable. I do not see a need for specialized software or an automated apparatus, which seems to me to just add cost.

It may take a bit of persistence to work it out, but we've covered that ground already.

I think it will take more effort to make it work well for contrasty transparencies, and I'm not sure a real PMT-based drum scanner is at risk just yet. But there is room for improvement without having to attain that lofty goal.

Rick "thinking Frank threw it through the door, shut the door, and is waiting" Denney

rdenney
10-Jan-2012, 08:21
There's no reason the camera has to be above film. Just put it where the paper usually sits on your enlarger :) Now you should have a nice flat film holder and even lighting.

Ctein already mentioned that a color head was the ideal light source--far more even than a light table and devoid of the texture that the diffusion surface of a light table often adds.

But the more I think about it, the more I'm convinced we need to keep the film vertical. That way, we can avoid having to use glass to flatten the film, and I really think we should keep any surface other than the film well outside the focus plane. And I think we need to keep glass out from between the film and the lens, for the same reason that we don't like photographing through windows. Many of us don't much like using filters, and those are optical glass that is made to be optically flat.

If we keep the film vertical, we will be mounting the camera horizontally. One aspect of the difficulty of focusing the camera is looking through it. DSLRs do not have removable prisms and other aids to right-angle viewing. Even using Live View would require viewing the LCD from the rear of the camera with most cameras. For me, that would mean putting the apparatus on the floor so I can get my eyes to where they need to be to focus.

So how about this for an apparatus: Use a color head for the light source. Given the size of the camera, a color head for medium format would work fine. Probably, any head would work, not just a color head. (Ben--evenness between exposures can be corrected in software, perhaps. But lack of evenness within the frame, as is clearly visible in my test, would not be easy to correct at all.)

Build a film stage that holds the film vertically. A glassless enlarger carrier might work well for this. I have one for my D3 that should serve the purpose. In fact, I might repurpose the whole top end of my D3, though it is designed for vertical mounting and the loose bits would have to be locked down for horizontal mounting. Mount the thing to a stable bench. The thick cutting board that Peter used looks interesting. Devising the movement mechanism for the film stage is the biggest challenge.

Put mount the camera on the other end of the bench.

I'd be willing to bet that this could be done with the parts of a gear-driven view camera, but perhaps not as cheaply.

Man, I need a weekend to devote to this.

Rick "who has some spare Graphic focus tracks that might prove useful" Denney

Frank Petronio
10-Jan-2012, 08:24
Well I always thought the yahoos on places like RFF who "scanned" their film with their DSLRs were hacks. They still are. But the ideas here have real potential.

I figure it will become useful when the last of the "better than Epson" scanners become unserviceable. I had a thread a few months prior to this one asking what, if any, scanner might still be usable in 20-30 years? (i.e. Well into my seasoned citizen status.) And the consensus was "none".

Peter De Smidt
10-Jan-2012, 08:38
This has been talked about for years, but there wasn't much interest due to the availability of good 35mm and medium format film scanners for a moderate amount. LF folks tended to be more traditional and a little later to the scanning party.

I'm interested in Rick's horizontal idea, but I'm guessing that alignment will be more difficult. If you don't want any glass or acrylic above the negative holding it down, you can always use something like a glassless negative mask for an enlarger on top of the negative. It would only hold down the negatives from the edges. Since the negative would be completely supported underneath, and since any heat from the light source would make the negative try to bow down in the middle, the film should stay quite flat, especially thick LF film.

That said, I've done my tests with a screen holder. It has a very fine anti-Newton texture on both interior negative holding surfaces, and yet a Cezanne is capable of about 6000 spi. I tested wet-mounting to optical glass and compared it to scans with the Screen holder on my Cezanne, and I couldn't tell a difference. (BW film.) But this is a different system, and it would be worth checking out.

Steven Scanner
10-Jan-2012, 08:54
This is an interesting thread. I've read page 1 to 4 and the first thing that comes to mind is: "Overhead projector projecting an image on a regular flatbed scanner that scans part of the image, than moves down one position and scans again. Repeat this untill the whole image is scanned. Stitch with software."
A 4x5 image projected can become about 8 to 10 times larger. If you can't make the hardware more acurate, enlarge the scale.

rdenney
10-Jan-2012, 08:58
I'm interested in Rick's horizontal idea, but I'm guessing that alignment will be more difficult.

I don't really see how. It's easier in many ways to measure plumb than level, but both are fairly easy. I can see that an advantage to a horizontal film stage is that the movement system might not have to be locked down because gravity would not be trying to move it. But I sure would like to be able to look through the camera without laying on the floor or standing on a ladder, and I would like to put this on a counter and not on the floor.

My plan is to lock down the bellows using a hard mount, and then hang the camera on the back end of the bellows. The bellows I'm using were designed for medium format using a camera much heavier than the Canon, and it is a beefy arrangement with tripod mounts on both standards. But it only has one extension rod, so I'd like to be able to lock down both ends of the bellows onto one of those long Arca-style plates and put the whole thing on a screw-type Arca-style clamp, which would allow fine focus.

But I never had any problem putting my 4x5 negatives into my enlarger using a glassless holder. The Omega D holder seemed to do the job, and I checked focus at the center and corners under the enlarger using a Peak grain focuser, so I know the film was flat enough to resolve grain.

Rick "thinking through the mechanics while waiting for the time to try something" Denney

Peter De Smidt
10-Jan-2012, 09:13
I certainly agree about the support structure liking horizontal better than vertical. My PB-4 bellows is strong enough to hold everything vertically, but it makes fine adjustments of the track more difficult. I'm going to make a cheap lead screw mechanism for camera position.

What I like about the vertical method is the ease of keeping things aligned by sliding the negative stage on a heavy, smooth plate, as well as the simplicity of using bolts in a triangular pattern for adjustment, which are held in place by the weight of the structure. The whole contraption isn't very tall, and with the unit sitting a normal table, it's no problem for me to look down into the finder. If you want to use 1 shot to get the whole negative, then this might be more of an issue, since you'll need great distance between the film and the camera. In any case, I expect that fine focusing is best done through tethered observation of the image on a computer screen, especially with live view. With some autofocus lenses, you can even adjust focus right from the computer in very small steps.

But there appear to be significant pluses and minuses to both orientations, and they both should be investigated.

rdenney
10-Jan-2012, 10:20
In any case, I expect that fine focusing is best done through tethered observation of the image on a computer screen, especially with live view. With some autofocus lenses, you can even adjust focus right from the computer in very small steps.

That does not fulfill the stated requirements, which was using whatever inexpensive DSLR might be available, preferably without any dependence on software which might well go stale.

By the way, it's easy to build things at right angles. Just go to Home Depot and buy cheap carpenter's squares and use those as angle brackets between platforms.

Rick "noting that prototypes often have to be discarded to find a different design paradigm" Denney

Peter De Smidt
10-Jan-2012, 11:55
Diversity in approach is a good thing. I hope we can come up with various designs, with no doubt some being better suited for individual applications, with varying film size, resolution, available space, equipment, construction abilities...all taken into account.

Using live view doesn't seem any more against the spirit of the endeavor than using stitching software. Neither would have to be used, but they might make the project easier or give better results than manual methods. Most Dslrs for the last couple of years allow live view. It's used regularly in professional work, especially table top work. I doubt that it'll go away any time soon. In any case, the builder can include what they like.

My off-the-top-of-my-head issue to be resolved with a horizontal version would be how to keep the negative holder flat between shifts of position in an easy way. That doesn't mean that there isn't an easy way, and it's certainly not my being defensive about my first prototype, which Rick seems to have implied, but I haven't though of it.

My horizontal 8x10 enlarger used a glass sandwich carrier that was held in place by a springs that pulled the light stage up tight against it. It was made removable by a cam on the end of a lever that move the light source back a bit. Something like that could certainly be done here, or one could use magnets, friction, xy lead screws ....

Changing topics, those interested in positional automation might want to check out:http://www.photomacrography.net/forum/viewtopic.php?p=99687#99687

Ben Syverson
10-Jan-2012, 12:30
That does not fulfill the stated requirements, which was using whatever inexpensive DSLR might be available, preferably without any dependence on software which might well go stale.
The great thing about keeping it agnostic is that you can use absolutely any camera capable of focusing at the required distance. That's what makes it future-proof. Live View will certainly aid in focusing, but obviously it's not required.

With that said, Live View has been standard on entry level DSLRs for over three years. And this project is just as well suited to large sensor mirrorless cameras like the NEX series.

After using Live View for macro-level focusing, I would not want to go back to manual focus. It's about ten times slower to take a photo, check the focus, retry, etc.

Jim Michael
10-Jan-2012, 12:33
Peter, perhaps tape the neg to the holder at the top, then use a hole & peg arrangement for positioning. I like the horizontal idea for eliminating the sag from a glassless carrier, which decreases dust. All that retouching after the fact is no less time consuming.

Peter De Smidt
10-Jan-2012, 15:35
One issue, whatever way you point the camera, is how to arrange for camera movement.

In my first prototype, there's some room to get the camera closer to the subject but not much farther away, such as you might want if you want to take a whole negative with one digital camera frame, a type of contact sheet scan, as it were.

You could simply have a slot and camera screw, but then every time you move the camera, you'd have to adjust for alignment. You could make a wooden carriage with a clamp, such that the carriage has a straight edge parallel to the lens axis. This way, you could use a flat strip parallel to the lens axis to hold the carriage up against when you slide it forward and back. This is what I use in my DIY 3d panorama head, and it works well in that use. (A quick search showed that I'm missing one of the components. I'll hunt it up and take some pictures soon.)

Another way so to attach a longer rail to the camera. For example, here's my DIY nodal slide:

http://i955.photobucket.com/albums/ae37/peterdesmidt/NodalSlideTop.jpg

http://i955.photobucket.com/albums/ae37/peterdesmidt/NodalSlideBottom.jpg

http://i955.photobucket.com/albums/ae37/peterdesmidt/NodalSlideonstand.jpg

Doing it this way, though, does lead to more vibration in the support when the camera section is not directly over the clamp section. Probably a better way would be to have the Arca style slide bolted directly the structure, and then use a female to female Arca clamp fixture. Something like what is seen at: http://www.ebay.com/itm/Set-Arca-Swiss-Clamps-Kirk-Markins-Wimberley-Benro-Foba-Acratech-/330644278615?pt=LH_DefaultDomain_0&hash=item4cfbef4d57

rdenney
10-Jan-2012, 16:09
The great thing about keeping it agnostic is that you can use absolutely any camera capable of focusing at the required distance. That's what makes it future-proof. Live View will certainly aid in focusing, but obviously it's not required.

With that said, Live View has been standard on entry level DSLRs for over three years.

I wasn't thinking of Live View, but the use of a tethered computer.

Rick "whose camera doesn't have LV, but who wishes it did" Denney

Peter De Smidt
10-Jan-2012, 16:45
Tethered live view is by far the best way to get precise focus in a studio environment with a stationary subject. Not only do you have a much bigger screen than on the camera, you can make adjustments without touching the camera. We've used this for years at the big studios where I work from time-to-time. It's not a fad.

Next best is to shoot tethered, because of the hugely bigger screen of a computer, and there's no need to shuttle a memory card back and forth.

If you look into what the top macro people are doing, when they're in a studio environment, they're overwhelmingly shooting tethered, live view or no.

But as has been said many times, if someone doesn't want to work that way, then they don't have to.

rdenney
10-Jan-2012, 19:47
Tethered live view is by far the best way to get precise focus in a studio environment with a stationary subject. Not only do you have a much bigger screen than on the camera, you can make adjustments without touching the camera. We've used this for years at the big studios where I work from time-to-time. It's not a fad.

Peter, I posted a bit in haste, and certainly my two posts above were in haste--trying to keep up with the conversation on my iPhone while playing the airport game today. My thumbs are fat and I'm always irritable when trying to type anything on an iPhone. I should have just waited. Now that I'm in my hotel (I've been in Virginia, Baltimore, and Nashville today), let me try again. If I offended you, please forgive me--that is never my intent.

Firstly, you've built a prototype stand, which is more than anyone else has yet done, and that deserves commendation. Of course, being first has its disadvantages, too, by providing a nice target for jerks like me to opine about. But that was not my intent in any case, I was just trying to carry it back to what I could do with what I already own, which is one of my stated goals in my concept of use. I'm a systems engineer, and it is built into my engineering process to start with what the user will do with the system, and with the user's constraints. That's why I described my issues with what I ran into during my proof of concept. I wasn't just making sure there was potential to improve on the Epson, I was also getting some experience with what it would take to operate such an apparatus. One of the things that made focusing difficult for me was having to stand on tiptoes when I did it.

From the outset, I have stated that I applied the constraint of what I already owned in terms of expensive stuff. I do not have the money to replace my 5D with something newer, and I probably won't have the money to do that for some time. My days as a consultant are over; now I get paid by a source that freezes my (reduced) salary merely to score political points. So, I'm constrained by features.

Live View is available on my wife's D300 (she's Nikon and I'm Canon, which promotes marital bliss). But it requires delving into the menu system to get to it, or at least it did when I tried it. I think it's implemented more easily on Canons newer than mine--my 5D doesn't have it at all--and perhaps more easily on newer Nikons. You would know that better than I would. Displaying Live View on a computer monitor? I didn't even know such features were available. I reacted earlier based on the assumption that one would have to use Capture One or something like that--and obviously that assumption was wrong. I do not use tethered computers even in the "studio". But whether or not, "fad" was a rotten choice of word.

My comment that sometimes we have to be willing to start over during prototyping was not intended as a knock on your design at all--I aimed it at me as much as anybody. It is intended to mean that at all times, our prototypes are evaluated against the requirements that emerge from how we will use the system. My requirements may differ from yours. You have solved the problem of the camera being too high, but using a solution that is beyond my budget, and, I suspect, beyond the stated desire to make use of cheap and maybe even older DSLRs. You also stated that alignment would be more difficult, and I did acknowledge at least some aspects of that. Maybe when I try my own hand at a prototype, I'll acknowledge all that reasoning and have to start over myself. I may even come to that conclusion just in trying to think it through. I will not take offense at all if you say I told you so.:)

There is certainly nothing wrong with competing ideas, as long as we keep how we will use the system, and the requirements that emerge from that concept of use, in the front of our minds. In my real work I have to remind myself of that about ever five minutes.

Rick "who has thrown a lot of prototypes in the trash, many of which earned the laughter they received" Denney

Jay DeFehr
10-Jan-2012, 19:47
No offense intended, but I think you're all looking in the wrong direction, namely, backwards when you should be looking forward. Innovation is rarely achieved by looking backwards. The idea that the best solution is a better version of a past solution is a classic trap. The best solution is probably simpler and cheaper than any past solution, and relies on technology that didn't exist when the past solution was considered a good one. Did anyone see my post (http://www.largeformatphotography.info/forum/showthread.php?t=85515) today about the superlenses based on metamaterials? You might want to check out the link.

rdenney
10-Jan-2012, 19:55
No offense intended, but I think you're all looking in the wrong direction, namely, backwards when you should be looking forward. Innovation is rarely achieved by looking backwards. The idea that the best solution is a better version of a past solution is a classic trap. The best solution is probably simpler and cheaper than any past solution, and relies on technology that didn't exist when the past solution was considered a good one. Did anyone see my post (http://www.largeformatphotography.info/forum/showthread.php?t=85515) today about the superlenses based on metamaterials? You might want to check out the link.

Jay, this is a big discussion in engineering circles. I'm a systems engineer, and thus my built-in process is to first document how the user will use a system without regard to any technology, and then to extract requirements from that concept of use that also do not assume or impose technology. Once the requirements are documented, then technologies can be evaluated. A good engineering solution is the cheapest alternative that fulfills all the requirements.

It is quite tempting to always look for an advanced technology and try to build a solution around it, but many engineering projects have failed utterly taking that approach. The potential for success may be higher, but the risk is also greater. I deal with projects that have failed or that have been delivered late or well over budget all the time, and nearly always it traces back to a solution looking for a problem. Sometimes, a few pieces of plywood screwed together, a cheap enlarger head, a macro rail, and some Arca-style plates and clamps can fulfill the requirements for cheap enough to do a lot of good work while we wait for technology to present new solutions as implementable products.

I read your post on superlenses, and maybe that will create a new technology that will change everything. But they are talking about something about which they are still doing basic research, while we are building prototypes and standing our existing cameras over negatives and doing tests right now. Given the approach we are taking, there will be no real loss if a year or three from now something comes along that changes everything.

Rick "cliche of the moment: the great is often the enemy of the good" Denney

Peter De Smidt
10-Jan-2012, 20:09
Hi Rick,

Thank you for your thoughtful post. I don't have live view at home either, but my forays into macro photography have made me wish very much that I did.

I agree that a horizontal approach has a lot of advantages, especially if one wants to be able to deal with 5x7 or 8x10 in one shot. With my 55 mm, I need about 110mm from the subject plane to the front of my lens at 1:1, but to get a whole 4x5" frame in one shot, I need about 13". Obviously, longer lenses will need even more extension, which could be a problem for a vertical approach if one needs to focus through the camera finder.

Another advantage of a horizontal approach is that there would probably be more space for a light source.

SURF
10-Jan-2012, 20:09
Someday I will try that too. :) I often think about it. It will be a very funny DIY project. The goal is to make something better than Scitex using DSLR if that possible. One thought: what light source will be the best? Galogen, proof light fluorescents or special grade fluorescents (like in Scitex scanners)? Ideas?

Peter De Smidt
10-Jan-2012, 20:18
Someday I will try that too. :) I often think about it. It will be a very funny DIY project. The goal is to make something better than Scitex using DSLR if that possible. One thought: what light source will be the best? Galogen, proof light fluorescents or special grade fluorescents (like in Scitex scanners)? Ideas?

For color, the highest CRI source at about 5-5500K would be ideal. Right now, that would be something like a Solux bulb or a flash. The downsides of a solux type bulb is the heat, and the downside of a flash might be consistency. There's a lot of room for experimentation here, including diffuse versus collimated light sources...

rdenney
10-Jan-2012, 20:29
Someday I will try that too. :) I often think about it. It will be a very funny DIY project. The goal is to make something better than Scitex using DSLR if that possible. One thought: what light source will be the best? Galogen, proof light fluorescents or special grade fluorescents (like in Scitex scanners)? Ideas?

Better than a Scitex is quite a demanding goal. I would be quite satisfied with better than an Epson, at least to start out with. But I think it would be possible to improve on a Scitex with the right components. But it solves a different problem. The point of the high-end flatbeds is to provide a film-scanner quality on a very large platen, while the point of this effort is to provide film-scanner quality for large-format film in a way that doesn't depend on proprietary software and hardware solutions. Different goals.

As to light sources, Ctein suggested using a color head from an enlarger. That would work quite well, I think, and those are based on halogen bulbs that have a very high persistence--high enough to smooth over the 60Hz power pulse. I worry about the pulsing of fluorescent tubes, though. I had that issue when experimenting with my light table--my shutter speed had to be longer than 1/30 to avoid an interaction between the 60Hz pulsing of the tube and the travel of the focal-plane shutter. This is not an issue for the scanning approach used by a Scitex scanner, I suspect.

I'm quite tempted to consider an LED bank under a diffusion panel. I might have to make my own--when powered by DC, LEDs don't pulse at all, but they are often pulsed to reduce power consumption and duty cycle.

Color is not so much the issue--color corrections are easy to make in software.

Rick "thinking the pulsing is what caused the hot spot in my test" Denney

polyglot
10-Jan-2012, 20:34
Hi guys, just found this thread and confess I only read a few pages in the middle ;)

However, this is, as you seem to have discovered, totally feasible. There are plenty of kits for stepper-driven X-Y stages available on the internet that run on open-source software and hardware, constructed using cheap chinese acme-screws and sliders. They're designed for creating CNC milling machines and the like, so they're quite rugged enough to position a camera.

I don't know why you're all focused (haha) on putting a macro lens on a DSLR though, when an enlarger is designed to project a grain-sharp image vertically and you all already own one. What I would suggest (for the simplest option) is mounting a DSLR with no lens face-up on an X-Y stage under an enlarger. We know the enlarger lenses are capable of producing grain-sharp images. Focusing is a little interesting (you probably want a live-view camera or a 90-degree prism on the viewfinder) but basically you treat the sensor as the paper and move the camera around to capture the whole image. You might (or might not) get issues with non-parallel light (Bayer masks don't like it) in the image corners but that would be improved by using a longer enlarging lens.

Consider a cheap 3-year-old APS DSLR of about 12MP, that's about 6 micron pixel-pitch, say 10 micron real resolution after you do the Bayer interpolation. If you set your enlarger to print to an 8x10" surface coincident with the DSLR's sensor, you get 20k*25k resolution (500MP), which is more than enough to completely capture any 4x5" film. Bigger enlargement = more resolution, and vice-versa.

If you want to capture arbitrary-sized media in stupid-high resolution, then by all means you can move the lens around with the DSLR. And you'll get higher quality from it too, but it's mechanically more-annoying. In that case, I would suggest that the best option is an enlarging lens (a classic 6-element 50/2.8 gives you 2x the coverage you need if using a 135-format DSLR) used in its proper orientation, i.e. facing away from the film or "reversed" with respect to the DSLR. Set it to about 2x magnification and off you go with 100lp/mm captures.

If you're automating this stuff using a microcontroller (I would suggest that this is a really good idea; it would take only a few hours to write the code and you can use really cheap off-the-shelf driver electronics), you can easily include a flash lightsource with the control. That eradicates any stability worries (particularly from shutter-shake) you might have had and means you can basically eliminate settling-time from the motion planning. I would be horrified if you couldn't capture frames at 1Hz with an automated system (hope your flash's power supply is well-cooled!), which with a bit of overlap means a completed capture in under 1 minute.

rdenney
10-Jan-2012, 20:37
Hi Rick,

Thank you for your thoughtful post. I don't have live view at home either, but my forays into macro photography have made me wish very much that I did.

I agree that a horizontal approach has a lot of advantages, especially if one wants to be able to deal with 5x7 or 8x10 in one shot. With my 55 mm, I need about 110mm from the subject plane to the front of my lens at 1:1, but to get a whole 4x5" frame in one shot, I need about 13". Obviously, longer lenses will need even more extension, which could be a problem for a vertical approach if one needs to focus through the camera finder.

Another advantage of a horizontal approach is that there would probably be more space for a light source.

The light source is one thing I'd thought about, too. But I don't need to do it in one shot--that would reduce the quality to well below what is possible with the Epson, which doesn't fulfill objectives, it seems to me. I think that was Ctein's complaint with the whole concept. Our tiling approach is what makes it appealing to me.

I've thought through the camera mount, and I think I could do mine either horizontally or vertically. But the film stage is still an issue for me, though it may be that the solution to the light source will help. A properly mounted and aligned D2 head, even with its regular light bulb, might provide a nice, flat table on which to construct a the stage. I need to dig for the head of my D2 and start trying something. But it will probably be a while--I spent most of this weekend doing a plumbing task--plumbing in my temperature control valve so that I can start processing film again--and only a little on playing with this concept. But it will probably be February before I have another weekend.

Rick "who can get a lot done if it can be done in a hotel room" Denney

Jay DeFehr
10-Jan-2012, 20:52
Jay, this is a big discussion in engineering circles. I'm a systems engineer, and thus my built-in process is to first document how the user will use a system without regard to any technology, and then to extract requirements from that concept of use that also do not assume or impose technology. Once the requirements are documented, then technologies can be evaluated. A good engineering solution is the cheapest alternative that fulfills all the requirements.

It is quite tempting to always look for an advanced technology and try to build a solution around it, but many engineering projects have failed utterly taking that approach. The potential for success may be higher, but the risk is also greater. I deal with projects that have failed or that have been delivered late or well over budget all the time, and nearly always it traces back to a solution looking for a problem. Sometimes, a few pieces of plywood screwed together, a cheap enlarger head, a macro rail, and some Arca-style plates and clamps can fulfill the requirements for cheap enough to do a lot of good work while we wait for technology to present new solutions as implementable products.

I read your post on superlenses, and maybe that will create a new technology that will change everything. But they are talking about something about which they are still doing basic research, while we are building prototypes and standing our existing cameras over negatives and doing tests right now. Given the approach we are taking, there will be no real loss if a year or three from now something comes along that changes everything.

Rick "cliche of the moment: the great is often the enemy of the good" Denney

Rick,

Your approach is very practical and, well....professional, but I was responding to suggestions of raising $1M for R&D of a DIY drum scanner to last decades. DSLR workarounds are perfectly legitimate, as far as they go, but beg the question; why not use digital capture in the first place and avoid the intermediary? If you've been following the development of metamaterials and various display technologies, you might have seen several contenders for a kind of contact scanner, but again, why not use the digital media for capture? I think this is what I'm getting at; if we follow any technology that will deliver very high resolution scans of film negatives, it might be a good candidate for replacing film as a capture medium, in one form or another. In other words, there's more than a few reasons no one is investing in the development of new dedicated film scanning technology, and the best we're likely to do is to co-opt technology developed for other uses, but rationally speaking, developing film technologies is.....not.

That being said, the decentralization of manufacturing/ 3D printing/ open source movement/ DIY laboratories, etc., could converge to make any project like this one, or even making a small run film coating machine, feasible. There's a whole lot of genetic engineering going on in kitchens and makeshift labs right now, and more every day, and some of them are producing real science, even now. This might turn out to be a two-edged sword, killing off the big mfg, but enabling the DIY'er, and possibly for the first time, it might be easier to find someone to tell you how to make something than to offer to make it for you/ sell it to you.

Peter De Smidt
10-Jan-2012, 21:16
I like the idea of being able to do a full negative in one or maybe two frames. That would be so fast that I could have it done in the time it would take for me to set up the trays to do a contact sheet. I don't have an Epson. I do have a huge old pro scanner with it's own g4 mac. It takes awhile to get going, and it takes a long time to scan. Moreover parts are extremely expensive, and so I only want to use it for my own max quality scans. Thus, for me the best use of a Petronio Scanner would be speed. If it also helps some people who'd like higher quality than what they are getting, well, that's terrific. Plus, I've always wanted to build a robot, and so I'm going to work on a robotic x-y positioning stage. In other words, I reserve the right to work for my own objectives.

Regarding Jay, he's trolling. Don't spend time looking to the past, says the guy on a website dominated by large format film photography. And he's trolling by reacting to a post made by another person who was trolling...$1mil for research and development. What a joke. Entering ignore Jay mode.

Regarding Polyglot's suggestions, sure I could work on my enlarger, but I'd rather not, for no other reason than my enlarger is in the basement, and my light room is on the second floor. In addition, a number of people here might no longer have a photo enlarger. For those who want to do this, there's no doubt that it could be made to work. Built a prototype for holding your camera, investigate techniques for focusing on the sensor....take some pictures and report back.

rdenney
10-Jan-2012, 21:24
DSLR workarounds are perfectly legitimate, as far as they go, but beg the question; why not use digital capture in the first place and avoid the intermediary?

There is no one-shot technology for digital capture with cameras of even 6x7, let alone 4x5 and larger. Gigapan isn't a one-shot solution, and requires software manipulation, some of which is not easy at all, to achieve any of the effects that are easy with a view camera and a three-dollar sheet of film. Scanning backs are not a one-shot solution, and require a static subject or at least impose a limited application with moving subjects. And they aren't as big as 4x5 in any case. And they are expensive. And they are not sustainable given the frequent abandonment of the required software platforms and the apparent difficulty in keeping software up to date with current platforms.

If someone made a 4x5 digital back for a price I could afford (which would be under $2000 these days), I'd buy it and forget this project, though I'd still worry about sustainability. But if I want to use a large-format camera with all its features, including my choice of how time is rendered (based on how long the shutter is open), then none of the current digital capture solutions fulfill the resulting requirements. Film does, and at least black and white film is quite sustainable into a pretty decent future.

That day may come when 4x5 (at least) sensors are good and cheap, but my sense is that it will take a breakthrough technology to make a large, affordable sensor possible. And then someone will have to productize it--that is not exactly kitchen-table stuff. With current technology, I reference the price of the IQ180 ($50K).

A scanner has a light source, a lens, and a sensor. With the design we are attempting, the light source can be just about anything, the lens is something really good but not outrageously expensive, and the sensor is a CCD or a CMOS already packaged in a handy product. I don't see a difference from any other scanner, except that we are talking not about scanning but rather digitizing in tiles.

The tiling is what makes the concept interesting. 20 tiles of a 4x5 negative will yield 12000 x 15000 decent-quality pixels in the resulting file. What single-shot digital capture system could do that?

Rick "wishing there was an affordable digital capture solution" Denney

Mike Anderson
10-Jan-2012, 21:26
Rick,

Your approach is very practical and, well....professional, but I was responding to suggestions of raising $1M for R&D of a DIY drum scanner to last decades. DSLR workarounds are perfectly legitimate, as far as they go, but beg the question; why not use digital capture in the first place and avoid the intermediary?...

The system they're working on here in a sense multiplies the capability of one's digital camera - by a factor of 10 at least, I think, assuming 4x5 film. And it will continue to do so as one's digital camera's capability increases - it scales with the (digital) camera.

The other issue is that many here have preexisting film they want digitized.

Frank Petronio
10-Jan-2012, 21:40
http://www.ebay.com/itm/Leitz-Aristophot-Macro-Photography-stand-Linhof-4x5-back-/200698900513?pt=LH_DefaultDomain_0&hash=item2eba963021#ht_11180wt_1348

Here's a nice hunk of hardware, a Leitz Aristophot Macro Photography stand with Linhof 4x5 back. Tinker with that.

Heck through a 45 scan back up there and do like Paul Graham's book "Films.": http://www.photoeye.com/bookstore/citation.cfm?Catalog=ze536

Peter De Smidt
10-Jan-2012, 22:26
Unfortunately, the nicest thing in the picture is the Luminar, which isn't included.

Mike Anderson
10-Jan-2012, 22:43
How will the falloff problem be dealt with? Even with an APS-C sensor and a small aperture there will be some falloff on each tile. There will need to be an extra processing stage in which the tiles are run through a Photoshop script to undo the falloff, (until there's application specific software for this system). A falloff negating (Photoshop) mask will be needed specific to the camera/lens/subject-distance used.

polyglot
10-Jan-2012, 23:56
How will the falloff problem be dealt with? Even with an APS-C sensor and a small aperture there will be some falloff on each tile. There will need to be an extra processing stage in which the tiles are run through a Photoshop script to undo the falloff, (until there's application specific software for this system). A falloff negating (Photoshop) mask will be needed specific to the camera/lens/subject-distance used.

All of the good stitching tools, including (free) PanoTools, deal with this transparently. They also have solvers that will undo any lens distortions as well, which is obviously important for the (cheap) wideangles commonly used for landscape stitching.

peter ramm
11-Jan-2012, 09:58
Wow, 23 pages. I can't read it all but perhaps someone (Frank?) wants to summarize what is a very interesting topic.

We used to make camera-based scanners that created step and stitch images at any resolution you wish. These were mounted on microscopes and did a nice job, but we never claimed to have the DR of PMT-based scanners. Never happen except with hi-bit cryogenic cameras viewing self-luminous targets. Works OK (only OK for complex reasons) at about 4D of dynamic range under those conditions. No resemblance to a film scanner, though.

The core problem (not the only problem) is flare. As soon as you are dealing with a detector array and an illuminating beam, you have flare. That flare limits you to much less than 4D of DR, even at low precisions.

In contrast, a scanning PMT limits flare because the beam is tightly constrained (especially if it is a confocal scanner). That's why drums are better than any array-based scanners (including linear arrays), and why camera-based scanners cannot match the performance of scanning PMTs. You can be fairly clever in limiting the flare component in a linear array scanner, but that only goes so far.

Making a crappy scanning PMT device is easy. Making one with the tolerances to deliver both spatial and bit-depth precision is very demanding (= expensive). So, how many of us LFers would shell out for a new product? Aztec would know. The numbers are far too small.

Will Frostmill
11-Jan-2012, 11:02
Live View is available on my wife's D300 (she's Nikon and I'm Canon, which promotes marital bliss). But it requires delving into the menu system to get to it, or at least it did when I tried it. I think it's implemented more easily on Canons newer than mine--my 5D doesn't have it at all--and perhaps more easily on newer Nikons. You would know that better than I would. Displaying Live View on a computer monitor? I didn't even know such features were available.

Rick,
As much as I'd like to sell you my Olympus E-520* (with live view! can mount Nikon lenses!), you have the tethering software you need on the disk that came with the 5D. Here's what google told me:
http://www.ehow.com/how_7330275_can-canon-5d-tethered-computer_.html So, it looks like you can park your laptop next to the setup and check focus instantly.

Will, who has no nifty tagline, or ready cash, Frostmill

*seriously, I'm going to put it up on the classifieds any day now.

Ron Marshall
11-Jan-2012, 11:23
Digital camera mated to a microscope. Multiple exposures and stitching.

The one in the link has a built in digicam, but at that price point probably not acceptable image quality.

http://www.microscope-depot.com/seriesU.asp

But if someone already has a decent DSLR then a microscope with a camera mount, stitching and perhaps HDR should be doable for not too much.

Peter De Smidt
11-Jan-2012, 11:29
Wow, 23 pages. I can't read it all but perhaps someone (Frank?) wants to summarize what is a very interesting topic.


To be frank, I'm not Frank, but I'll give it a go anyway.

My goals are to make a scanner such that:
1. It's quality is significantly better than an Epson flatbed.
2. It's easy to use.
3. It's fairly easy to build from non-esoteric or extremely expensive parts.
4. It can be used in a manual mode.
5. It can be automated, at least to an extent.
6. It can be used quickly for proofing, or more slowly for higher quality.
7. Finally, a "how-to" manual will be made which would help interested parties build their own scanner.

Those are my goals, and other people's will probably differ, which is great. I'm a little tired, though, of people telling me what my goals should be.

The idea is to use a copy stand, whether vertical or horizontal, style system with a dslr and a macro lens. The negative will be move on a flat surface so that small areas of the negative with be "scanned", perhaps with multiple exposures for extended dynamic range through HDR, and the tiles will be stitched together with something like panotools.

Some tests have been done and at least one prototype has been constructed. Both HDR and samples at 1:1 have given positive results. In addition various technical experts have been called upon, and the result is that this is a very viable project. Thus, I won't waste any more time arguing about feasibility.

The time has come to try things out and see what works. I hope that others beside Rick and I will also build something. If someone isn't interested in actively take part in testing, construction, or offering positive suggestions, I politely ask that they spend their time posting in other threads.

Kirk Gittings
11-Jan-2012, 12:10
Peter, I appreciate your efforts on this. I think you are heading in exactly the right direction.

Ron Marshall
11-Jan-2012, 12:28
Peter, I appreciate your efforts on this. I think you are heading in exactly the right direction.

+1 Thanks!

peter ramm
11-Jan-2012, 12:29
To be frank, I'm not Frank, but I'll give it a go anyway.

My goals are to make a scanner such that:
1. It's quality is significantly better than an Epson flatbed.
2. It's easy to use.
3. It's fairly easy to build from non-esoteric or extremely expensive parts.
4. It can be used in a manual mode.
5. It can be automated, at least to an extent.
6. It can be used quickly for proofing, or more slowly for higher quality.
7. Finally, a "how-to" manual will be made which would help interested parties build their own scanner.

The time has come to try things out and see what works. I hope that others beside Rick and I will also build something. If someone isn't interested in actively take part in testing, construction, or offering positive suggestions, I politely ask that they spend their time posting in other threads.

Peter, this has been done, is being done, and will be done better in the future - perhaps by you. I think if you search on medical image stitching, for example, you will see lots of stuff. Then there are the consumer products like Panavue. People have tried doing films with those and there are citations in the literature discussing the issues.

The technical aspects of seamless image montaging are formidable. Trying some commercial software will demonstrate that. You should be able to get a demo from someone who wants to sell you a montaging system. What about the GigaMacro thingie cited earlier? An hour with an existing system will give you a feel for what the issues are.

However, the stitching bit that the montaging system deals with is quite independent of your first point - whether or not the rectangular area detector in a camera can deliver quality better than the linear array in an Epson. Based on past experience, my prediction is no - but I would love to be wrong.

Peter De Smidt
11-Jan-2012, 12:40
Hi Peter,

Thank you for your input. Part of the fact that much of this has been done before is a good reason for optimism. We are at a good place technologically at the moment, with digital cameras, diy robotics, hdr, and stitching technology making things practical that weren't just a short time ago.

Maybe I'm wrong, but I hope that the samples posted earlier in the thread speak to the achievability of goal #1.. See: http://www.largeformatphotography.info/forum/showthread.php?t=84769&page=11 Posts #105 and #108.

See also Rick's test at: http://www.largeformatphotography.info/forum/showthread.php?t=84769&page=17 post #170.

Kirk Gittings
11-Jan-2012, 13:06
Peter, look at this exchange for Auto Pano Pro (http://www.kolor.com/forum/p60679-2010-03-23-11-52-58). I think there may be some help here for the software issue via "templates". I used to own this software but never tried this. It seems to allow you to tell the software how you shot the tiles which allows it to arrange the tiles appropriately regardless of detail content.

Peter De Smidt
11-Jan-2012, 13:22
Thanks Kirk. A brief skim showed that the template idea might be very helpful. I'll look into it.

rdenney
11-Jan-2012, 19:38
Rick,
As much as I'd like to sell you my Olympus E-520* (with live view! can mount Nikon lenses!), you have the tethering software you need on the disk that came with the 5D. Here's what google told me:
http://www.ehow.com/how_7330275_can-canon-5d-tethered-computer_.html So, it looks like you can park your laptop next to the setup and check focus instantly.

You can't check focus instantly. You can make a preview exposure and see that quickly. Live View, of course, gives you a real-time moving display. I actually find I can check focus pretty well just zooming in on a test exposure using the LCD on the back of the camera.

But I can also see through the viewfinder well enough to focus, as long as I can get my eyeball to the viewfinder window without having to perch precariously on one toe.

Live View on a camera with an LCD that can be twisted around to point forward is an excellent solution, but I don't have one of those.

Rick "whose skunkworks is not well funded" Denney

rdenney
11-Jan-2012, 19:48
Thanks Kirk. A brief skim showed that the template idea might be very helpful. I'll look into it.

We've been operating under the assumption that two images can't be stitched easily if they cover flat, featureless subject matter. Some have argued that grain and dust will be enough to allow that, but I think I'll see if I can get Photoshop to stitch two images across a featureless stitch boundary to see if it's possible. Of course, I might not know if it fails, which seems telling in its own right.

Rick "thought booked for the next few weekends and limited to talk for a while" Denney

Dave Jeffery
11-Jan-2012, 20:13
Wow, 23 pages. I can't read it all but perhaps someone (Frank?) wants to summarize what is a very interesting topic.

We used to make camera-based scanners that created step and stitch images at any resolution you wish. These were mounted on microscopes and did a nice job, but we never claimed to have the DR of PMT-based scanners. Never happen except with hi-bit cryogenic cameras viewing self-luminous targets. Works OK (only OK for complex reasons) at about 4D of dynamic range under those conditions. No resemblance to a film scanner, though.

The core problem (not the only problem) is flare. As soon as you are dealing with a detector array and an illuminating beam, you have flare. That flare limits you to much less than 4D of DR, even at low precisions.

In contrast, a scanning PMT limits flare because the beam is tightly constrained (especially if it is a confocal scanner). That's why drums are better than any array-based scanners (including linear arrays), and why camera-based scanners cannot match the performance of scanning PMTs. You can be fairly clever in limiting the flare component in a linear array scanner, but that only goes so far.

Making a crappy scanning PMT device is easy. Making one with the tolerances to deliver both spatial and bit-depth precision is very demanding (= expensive). So, how many of us LFers would shell out for a new product? Aztec would know. The numbers are far too small.

If we were wanting to build something anyway just to try and do better than an Epson V700 / 750 what would you recommend for a set up using a DSLR?

I was raised in Grimsby, Ontario so can you throw an old fellow Canuck a bone? :)

argos33
11-Jan-2012, 21:02
Rick,
I agree the problem of the whole unit becoming too high is an issue, however I am going to stick with designing a vertical setup for now. I think for most normal use you will be using the DSLR close enough to the film to stitch that it is low and it shouldn't be an issue. For those wanting to do a quick 1 or 2 frame capture it might start to get high, but perhaps some of the aforementioned solutions would be acceptable (tethered, live view, etc). As a side note I remember seeing a neat video online about tethering a cheap portable DVD player to your DSLR. If the camera didn't have live view you could at least see the photo on a bigger screen to asses focus and exposure. The technology is getting better and better too - it's just a matter of time before you can plug your DSLR into a cheap tablet I'm guessing. Another thought might be to make the whole column pivot where it attaches to the base, so that you could use it in either orientation depending on the height, etc. Seems to me that with some slight modifications (resting points behind the column, reinforced lockable joint etc) it wouldn't actually be that much more work.

As far as the light source, I too was thinking LEDs with diffusion - as mentioned before the light output on a regulated DC power supply should be very consistent, and you could always add a gel filter (between two diffusion layers for example) to adjust the color if it is way off. The rest could be dealt with post-capture. I am thinking LED light banks similar to this one except maybe smaller depending on format:

http://www.amazon.com/LED-Wholesalers-2501WH-White-Square/dp/B001NDU9XG

I also don't see why you couldn't make such a light source dimmable so you could adjust for the density of the negative. This style of light source would also save a lot of room which would help to keep the whole thing from getting to big/high up.

Peter, as a friend of mine likes to quote: "Don't let the bastards grind you down!" This is a great idea and we all appreciate you sharing your photos and hands-on experience thus far. I think it is great that we can talk about an idea like this and develop it together in an open-source manner. Nice job on the prototype - as Rick pointed out it doesn't have to be pretty - just able to test the viability of the design, you can always improve the materials and look of it. I like the idea of using a track for general positioning like yours and on the Aristophot (thanks Frank), and then having a much more precise platform for actually moving the camera up and down to focus. That would help a lot for folks using process, enlarging, and microscope lenses.

I am in a similar boat as Rick in that I am pretty busy right now, but very interested in making my own prototype and have already started planning and setting some materials aside in the shop.

Evan

Kirk Gittings
11-Jan-2012, 21:39
We've been operating under the assumption that two images can't be stitched easily if they cover flat, featureless subject matter. Some have argued that grain and dust will be enough to allow that, but I think I'll see if I can get Photoshop to stitch two images across a featureless stitch boundary to see if it's possible. Of course, I might not know if it fails, which seems telling in its own right.

Rick "thought booked for the next few weekends and limited to talk for a while" Denney

I make a large part of living stitching images in PS for clients needs. The only problem arises when there is little detail in the overlap areas such as a featureless wall in the overlap area-then I have to go to a more sophisticated program. I know this would be a nightmare in many of my 4x5 negatives with allot of sky. I don't even need to test it on film as I basically test it every week in other ways. I don't think grain will matter any more that texture in stucco does. I think the "template" feature in Auto Pano Pro might overcome this issue because you can "tell" the software where the tiles go.

Nathan Potter
11-Jan-2012, 21:49
http://www.ebay.com/itm/Leitz-Aristophot-Macro-Photography-stand-Linhof-4x5-back-/200698900513?pt=LH_DefaultDomain_0&hash=item2eba963021#ht_11180wt_1348

Here's a nice hunk of hardware, a Leitz Aristophot Macro Photography stand with Linhof 4x5 back. Tinker with that.

Heck through a 45 scan back up there and do like Paul Graham's book "Films.": http://www.photoeye.com/bookstore/citation.cfm?Catalog=ze536

I use one of these Leitz Aristophot units in my darkroom for precision copying. It is essentially a vertical optical bench which is highly configurable. I'd consider it for a scanning DSLR copy setup with X/Y stage at the base and bottom illumination. Well built from steel parts, carefully machined and virtually an antique. The original ring light with 12 tungsten bulbs and opal diffuser is a wonder of uniform illumination.

Nate Potter, Austin TX.

Nathan Potter
11-Jan-2012, 21:59
Wow, 23 pages. I can't read it all but perhaps someone (Frank?) wants to summarize what is a very interesting topic.

We used to make camera-based scanners that created step and stitch images at any resolution you wish. These were mounted on microscopes and did a nice job, but we never claimed to have the DR of PMT-based scanners. Never happen except with hi-bit cryogenic cameras viewing self-luminous targets. Works OK (only OK for complex reasons) at about 4D of dynamic range under those conditions. No resemblance to a film scanner, though.

The core problem (not the only problem) is flare. As soon as you are dealing with a detector array and an illuminating beam, you have flare. That flare limits you to much less than 4D of DR, even at low precisions.

In contrast, a scanning PMT limits flare because the beam is tightly constrained (especially if it is a confocal scanner). That's why drums are better than any array-based scanners (including linear arrays), and why camera-based scanners cannot match the performance of scanning PMTs. You can be fairly clever in limiting the flare component in a linear array scanner, but that only goes so far.

Making a crappy scanning PMT device is easy. Making one with the tolerances to deliver both spatial and bit-depth precision is very demanding (= expensive). So, how many of us LFers would shell out for a new product? Aztec would know. The numbers are far too small.

Peter, you are correct about flare really limiting the Dmax capability of sensor array capture. I think I had suggested earlier in this blossoming thread that using a highly collimated light source would help to greatly increase contrast. Actual tests may show this is highly desirable.

Nate Potter, Austin TX.

Peter De Smidt
11-Jan-2012, 22:09
I like the idea of LED lights. Bright, compact, low power and low heat (for their brightness). They can be dimmed, and there's no reason to use super high bin parts. (I'm also a salt water aquarist, and LEDs are all the rage in that hobby, but light output is critical for a reef tank. I expect that getting enough light won't be too hard for us. One 3w Cree LED would probably be sufficient.)

Another benefit over flash is that you wouldn't need a separate viewing/focusing light. My one worry with LEDs, especially for color, is the discontinuity of the spectrum. The only way to know if this is an actual, opposed to a purely theoretical, problem will be to try it.. I'll be testing flash and halogen lighting for a start, simply because I have everything on hand.

Brian C. Miller
11-Jan-2012, 23:30
The only problem arises when there is little detail in the overlap areas such as a featureless wall in the overlap area-then I have to go to a more sophisticated program.

This is not an issue! The software already knows the size and position and overlap of each segment. You don't have to worry about syncing up with grain, dust, whatever. That is only needed when the software is assembling a jigsaw puzzle made by a delusional epileptic monkey on acid. The whole thing is computer controlled, so each capture is a known segment on a grid.

This isn't the problem you're looking for. Move along.

-- Brian "No Jedi mind trick required" Miller

Kirk Gittings
11-Jan-2012, 23:35
Move along? What are you referring to? What software? The PS5 I was referring to can't even do it with a simple flat stitch of two images if there is not sufficient detail in the overlap areas. I deal with this on commercial jobs every week. This will be a issue. On more sophisticated software you have to manually do control points-a PITA. So what do you know that you haven't said?

Ben Syverson
11-Jan-2012, 23:57
Knowledge of where the shot is situated in the overall image is probably not enough to eliminate the need for the stitching software to correlate the images. Even if the X/Y stage is computer controlled, it won't be accurate to the pixel level.

However, dust will provide a natural target for correlation. Even very clean negatives have plenty of dust. I don't foresee any big problem with stitching. Some custom software may be required, but that's not the end of the world.

Kirk Gittings
11-Jan-2012, 23:59
Ben based on my experience stitching, I don't think dust will be enough-maybe if there is a ton of it and you jack up the contrast-but I doubt it.

rdenney
12-Jan-2012, 00:03
This is not an issue! The software already knows the size and position and overlap of each segment. You don't have to worry about syncing up with grain, dust, whatever. That is only needed when the software is assembling a jigsaw puzzle made by a delusional epileptic monkey on acid. The whole thing is computer controlled, so each capture is a known segment on a grid.

This isn't the problem you're looking for. Move along.

-- Brian "No Jedi mind trick required" Miller

Sheesh, Brian, that's easy to say but not so easy to live up to. What software are you talking about? Software that hasn't been written yet? I don't intend to write software for this activity. And you are assuming that the software knows, within 4 or 5 microns, the position of the film stage, or that the film stage can be moved with that degree of accuracy. Anything I build will not provide that, though it will keep the film in the correct plane. Just because it's easy to articulate the solution to a problem means the resources are at hand to achieve it. Do you know of a product that already does that?

Others are happy to construct a vast complication, and that's fine, but I'd just as soon build mine from existing bits, preferably those I already own. I'm short of time and money, and requirements for much of either will make this project as inaccessible as a Scitex.

I have an old copy of Panavue but not on my newer machine. It allowed precision alignment using manually set alignment points, either based on landmarks in the image or on marks of our own making. With a distortionless flat-field lens (which is not that hard to achieve for bellows lenses that tend to be symmetrical), only two markings are needed and they could be placed on the support glass and removed later in Photoshop.

The software Kirk has suggested deserves careful study, which I will do at some point. But I wonder if a template would still require precision lateral positioning of each tile. I was thinking that approximate positioning, based on markings and manual positioning, would be the much easier machine to build.

Rick "keeping it simple" Denney

rdenney
12-Jan-2012, 00:19
Peter, you are correct about flare really limiting the Dmax capability of sensor array capture. I think I had suggested earlier in this blossoming thread that using a highly collimated light source would help to greatly increase contrast. Actual tests may show this is highly desirable.

I'm hearing this loud and clear. The histogram for my test scan, which included quite a range of tones from deep shadows (base fog on the negative) to fairly dense highlights, only filled the center third-to-half or so of the full range recorded by the sensor. To restore contrast, I stretched the tone curve, but didn't see any hint of posterization either visually or by looking at the histogram.

I got better contrast at the pixel level than does the Epson, not that the Epson presents a difficult goal, but I can see how flare would blind the sensor to low-contrast fine detail. I made no attempt at any masking whatsover, and my light source was diffuse and much wider than the negative, let alone the field of view. So my test results above are perhaps as bad as it gets in this regard. Also, I converted from RAW to a high-gamut color space (ProPhoto), which in hindsight was a mistake--lots of empty gamut left at the extremes of the histogram.

I'm not that sure such detail survives conventional enlargement any better, but that would be sacrilegious to mention. Nor do I think any other scanner except for a laser/PMT drum scanner is likely to do that much better. I have a condenser head that I can probably find with some digging and use, if for no other reason than to provide a very even light source. It will be more collimated than a light box, but not nearly as much as a laser source. I suppose I'll find out how much it does or does not help.

Rick "who has already bettered the Epson with no special care" Denney

Leszek Vogt
12-Jan-2012, 02:46
It appears to me that good portion of this gismo is worked out. I sometimes use PS8 for stitching and it works v. well. I believe the ptgui has better stitching software. The idea of the camera pointing down w/macro lens make sense to me. The camera could be sliding on solid rails (similar as the interior crane operates).....and set to better than 0.0001" accuracy. Indeed, this can be motorized by servo motors and controlled via CMC type software....or similar.

While the camera moves over the determined grid, it fires frames allowing 10-20% space to overlap...or even more if sweet spot of the lens is used. However, the bigger the file the more problematic will be the processing of it. I had a problem with 514MP pano file out of 7 shots and my computer has near 4GB of memory....so this is kind of heads up.

Once the sliding rails are firmly attached and calibrated, the camera/lens should be the same distance over the surface of the 4x5 or larger negative. Which also means that there will be a need to use live-view focus (13X on my D700).....and once that's accomplished (tether or manually)...the camera can then copy the 4x5 negative/positive. Naturally the WB should be set to manual....and on some cameras Kelvin degrees can be adjusted in increments. The camera has a direct hook up to USB so there would be no need to juggle the memory card....as well as the photos can be reviewed on large H. Def monitor. Using wi-fi camera controls from a laptop would allow all of this to work remotely. Also, the lens could have gearing and also can be rotated via wi-fi....in order to obtain precise focus.

True, the expense of the D700 is up there, but the files would definiately look clean. The only unresolved part would be the automation of running the camera on rails over the negative and triggering it at appropriate spot in order to cover the overlapping areas. However, that could also be done manually if necessary....and naturally it would take more time.

If i didn't convolute the issue, hopefully this was helpful.

Les

Dave Jeffery
12-Jan-2012, 03:10
"a highly collimated light source would help to greatly increase contrast"

I know very little so please don't jump on me but might it be possible to cheaply set up a linear array of collimated light sources even if they were staggered a little? Fiber optics? Tiny tubes aligned that direct light? Even if the light source was not configured in a straight line a precision timed movement of the light source, film, or sensor in one direction should provide even illumination?

If only a small array of light can be set up, mow an area the size of the sensor with collimnated light and a small, highly accurate x-y controller, then move the film with a larger, less accurate x-y controller with some overlap, scan again with the highly accurate x-y controller and then stitch the smaller areas together? Could this possibly reduce the cost of trying to make a large, highly accurate x-y table?

If a collimated light source could be set up in a linear fashion the same length of the short axis of a sensor, the movement of the light, film, or sensor would only need to be linear and this should be much cheaper to build with precision.

So the idea is just make a small scanning area, highly accurate, and hopefully cost effective to build, and then stitch the small scanned areas together?

I'm talking out my butt and just throwing out ideas in hopes that better ideas might come of this so don't read too much into this.

http://www.robotshop.com/ for Arduino controllers etc. There is an online community that can help with programming the Arduinos but the x-y thing is supposed to be pretty common apparently.

This is a great project!

Peter De Smidt
12-Jan-2012, 08:13
I like the ideas! Good work!

Regarding stitching, as I've said before I agree with Kirk that there might be some issues here, but we really don't know whether there's a problem or not until we stitch some samples at the magnifications we're talking about. Let's not get too involved in fixing a problem that might not even be there.

Regarding light sources, I'm a microscope neophyte, but don't some of them have collimated light sources? This might be a good option for experimentation.

Regarding moving the camera instead of the negative, this would certainly be feasible, and it would especially be good for people with 8x10" and larger negatives, since if you move the negative, the glass support plate needs to be twice as big as the negative with a moving x-y negative stage. My guess is this would be a little harder to implement with common materials, but I'd love to be wrong, and it'd be nice if someone investigated this. I really wish that I had a closet full of 80/20 (http://www.8020.net/) components.

The best next steps appears to be:
1) Making an even light source. My first one is going to be diffusion.
2) Making a sliding negative stage. That way we can investigate the stitching issue.

I'm certainly up to doing that, although I'm going on location for a shoot this weekend, and so it'll probably be next week before this gets done.

Regarding the robotics, I'm sure I can do this too, but it's been a long time since the electro-magnetism semester of college physics. If someone with more electrical experience wanted to investigate positioners, stepper motors, stepper drivers (probably an Arduino), Arduino programming... That would be a big help. See, for instance: http://www.photomacrography.net/forum/viewtopic.php?t=15673

peter ramm
12-Jan-2012, 08:20
If we were wanting to build something anyway just to try and do better than an Epson V700 / 750 what would you recommend for a set up using a DSLR?

I was raised in Grimsby, Ontario so can you throw an old fellow Canuck a bone? :)

Dave, as a fellow canuck I am hinting that you can dig this pit as deep as you like and there ain't no bone in it. The problems are not obvious until you have some montaging experience and that is why I suggest trying one of the commercial scanning systems before investing much time. On the other hand, you can learn lots and have fun if that is the goal.

Note, I have no direct experience using DSLRs to scan film. Everything I did was with scientific cameras. Therefore, feel free to consider me as much a newbie as anyone. Perhaps there is a special sauce in the DSLR.

Funny, we made our mark using camera-based systems to replace scanning PMTs in demanding applications. Fuji Film's medical division used to sell one of our scanners in Japan as the "Beautiful Image System". And here I am being a wet blanket. Sorry.

Montage of about 25 tiles. She's kinda dirty and missing a leg but not bad for 150 years old.

Brian C. Miller
12-Jan-2012, 08:27
Move along? What are you referring to? What software? The PS5 I was referring to can't even do it with a simple flat stitch of two images if there is not sufficient detail in the overlap areas. I deal with this on commercial jobs every week. This will be a issue. On more sophisticated software you have to manually do control points-a PITA. So what do you know that you haven't said?


Sheesh, Brian, that's easy to say but not so easy to live up to. What software are you talking about? Software that hasn't been written yet? I don't intend to write software for this activity. And you are assuming that the software knows, within 4 or 5 microns, the position of the film stage, or that the film stage can be moved with that degree of accuracy.

Beings that I make my living from writing software, and I have experience in machine automation, I write from the authority of personal experience when I say, this isn't a problem.

The GIMP is driven by Script-Fu. The Script-Fu language is what is engaged when you click the buttons on the GUI. Here we have a situation where images, without edge landmarks, need to be tiled and overlapped together. This means that certain edges don't matter, so they can be ignored. Ignore the edges, and position according to the tile information. It's just that simple. There isn't that much adjustment needed to feather in 64 white or blue or whatever tiles.

The reason that the conventional tiling software has a problem is because it is written such that the edges must have landmarks. But here, in our situation, for this purpose, the camera's position is known. 2 follows 1, again and again, and with a reasonable degree of precision. Falling back to camera position is just that, falling back. If the primary stitch target doesn't exist, then camera position is all that matters. The tiles get feathered together in sequence.

This would be a minor modification to the existing stitching scripts.