How to make a ULF "Camera" out of Flatbed Scanner?

[mark_tucker2]

Has anyone ever contemplated making a "camera" out of a

flatbed Epson scanner? I envision a tripod mount glued to the

bottom of the scanner, and then some type of LF Ground Glass

laid down on the glass of the scanner, to "accept" and illuminate

and receive the image. Then some type of basic black box built

around the scanner, then some type of lens mounted to this

black box.

 

The advanced "location version" would also include a Honda

generator to power the scanner and the laptop.

 

Given the size of the 8.5x14" glass, and the 4800dpi of the Epson

2450, the created file could/would be MASSIVE. Something equal

to the equivalent of 8x10 or 11x14 film. It would, in effect, be a

scanning back, so landscapes would be tough, if you had a

scene with moving clouds, or traffic, or people. Shutter speed

would be about a couple of minutes at high-rez.

 

Would ground glass be the material to use to illuminate the

image? That, I see, is the hard part going in -- how to get the

projected image bright enough for the scanner's sensibilities to

pick it up.

 

Any thoughts, suggestions? Thank you in advance.

 

-Mark Tucker, http://marktucker.com/

[scott_fleming1]

I recommend that you marry it to the plungercam and mount it on a cinematic high rise dolly on the back of your new large format transportation vehicle. Can't go wrong. When can we order one?

[jay_de_fehr]

Hi Mark. An interesting idea, and a question for those with much greater understanding of scanning mechanics than I enjoy, but I think I see where you're going . It would seem that a scanner capable of full bed transparency scanning would be in order, with the projected image as the lightsource. Potential problems might include, as you mentioned, light intensity, as well as exposure time. How long does an appropriately high resolution scan take to complete, and how would an image with parts in motion capture? Unfortunately I am unable to provide you answers to these questions, which I'm sure have already occurred to you. Just voicing my support and encouragement. You might consider posting this question in a digital forum. If their level of expertise is anything approaching that of posters on this one, you should get some very intelligent answers very quickly. Good luck, I'll be following this thread. Thanks for the brain teaser.-jdf

[mark_tucker2]

The reason I posted it in this forum is the factor of getting an

extremely bright projected image. It seems like it would need an

f2.8 giant lens; something like one of those wacko lenses that

Dagor77 puts on ebay. Something to transmit a very high

amount of light. It seems like a LF question to me; how to get

that image to be super-bright. It almost needs to be "amplified"

in some way.

[._._z]

Take a look at

 

http://www.rit.edu/~andpph/text-demo-scanner-cam.html

[brian_c._miller]

Suggestion: You need good coffee!

 

I roast my own in my kitchen, and I'll send you a pound of either Sumatra Iskandar triple-picked, or Kenya AA Kilimanjaro, or Ethiopian Sidamo, or Zimbabwe AAA Victoria Falls. If you want it ground for you, then it'll have to be sent overnight express, and you'll have to pay for that shipping. Otherwise, I'll mail it parcel post. Just send me a mail address.

 

After drinking good coffee, intead of that rot-gut brew which has been giving you hallucinations, you will realize that using a scanner to create a flatbed camera is utterly daft.

 

Idea analysis:

Scanners are designed to focus on a flat object. You would need to turn the glass plate into frosted glass, probably with an appropriate acid as an etchant. This will give you your image plane.

 

I think that you would either have to restrict your activity to very bright days, or else use enough hot lights on the subject to provide adequate light. And use subjects which can withstand being broiled by multiple 1Kw lights for several minutes. Flowers are probably out of the question.

 

Jack Daniels didn't pay you with product in lieu of cash, did they?

[roger krueger]

If you use a ground glass you're limited by the fineness of the grind on the glass--nowhere near the scanner resolution--and you also can never really get decent contrast or illumination. The scanner light source, at only an inch away, is really, really bright. You'd have to use immense movie lights on your scene to even get within a couple of stops. (And you might have to convince the scanner that having the bulb disconnected isn't a hardware error to be stopped for.)

 

You're better off focusing the image normally on the CCD. I'm not sure how the Epson 2450 is configured, but my ancient (5+ years old) Epson folds the lightpath twice with flat mirrors, then uses a regular lens to focus the image on a (roughly) 4-inch CCD. I know the cheapo modern scanners use a micro-lens on each pixel, but I'm not sure if that's the case for better scanners.

 

You can't just swap out the lens or change its focusing point--if the lens is part of the moving assembly, you only image a portion of your target equivalent to your scanner travel in the long dimension. In the short dimension, however, everything works normally. The result would be a massively dimensionally distorted image at all but one distance--with the lens it comes with, that one distance is in fact the distance to the scanner glass. Put another way, shooting a mountain off in the distance, you'd get an image that covered 14 inches of the subject in one direction and 50 miles in the other.

 

If you mount the lens to be stationery in the former position of the bed (pointing straight up) you than do in fact have a scanning camera that covers 4" x 14". (The image you get will be 8" x 14", but twice as tall as it should be.) Except (and this is where I gave up the project) your lens is optically too far away from the imager (unless it's quite a long lens) and you've still got two pieces of indifferent-quality mirror in the lightpath.

 

You could yank the mirrors and point the CCD straight up (with the folded lightpath it normally points forward, that is, towards the power cord), but this would have involved complete destruction of the original CCD mount, (and I needed to be able to return my unit to use as a regular scanner.) Also a problem is that my scanner imaged a white test strip located inside the case, just past the glass, on startup, and before each scan--a calibration thing. If it can't see something fairly similar to the strip it wants at that point, it just stops, and begins blinking the front panel lights, indicating a hardware error. I'm guessing you could put something white somewhere in the light path (and illuminate it!) to fool it, but I never got that far.

 

Even after all this you'd probably still not have enough light. I know some film scanners can use longer exposures, thus slowing down the scan, but I'm not sure if any flatbeds can do this. Seems like a really inexpensive way to increase your machines d-max, but I'm not a scanner designer--maybe there's a good reason not to include this feature. The person who might be able to give you an answer on this is Ed Hamrick, www.hamrick.com --the guy who writes VueScan--he probably knows more about scanner firmware than anyone on the planet.

 

Anyway, PLEASE post any results you get in this project--real scanning backs cost an immense amount of money, converting a flatbed is a VERY attractive option if you can make it work.

[clayton_tume]

Hi Mark

<p>

I tried this about 3 years ago but in a slightly different format, I have a 90 year old rotating Cirkut camera that uses 8 inch roll film and is essentially a scanning camera. The Cirkut exposes film through a narrow slit, the film is driven past the slit as the camera rotates, the movement of both is sychronised and the end result is a sharp image. Negative size is 8 x 60 inches or longer if long focal lengths are used.

<p>

So, this is the perfect format to experiment with. It required only mounting the flatbed sensor array in the exposure slot in the correct place. Now my memory is hazy on this one, I did get images off it but there were a number of software issues to deal with that prevented me from going further. Also the image quality wasn't that hot and was a long way off an 8 x 60 contact print.

<p>

I used a 300 ppi scanner and I remember it took forever to scan the scene, my camera wouldn't rotate slow enough to sync properly. Things have changed a lot since then and you can now buy rotating cameras with digital arrays but they're very slow and only equivalent to 35mm film......they've got a long way to go to catch up to LF.

<p>

Let us know if you have some success, it's fun experimenting!

<p>

Clayton

[nathan_congdon]

Well, the nattering nabobs of negativism should have a look at Bailey's post above! I have to admit I am impressed:

 

1. That this could be done at all

2. That anyone would go so far as to actually do it

3. That this amazing forum has once again succeeded in "closing the loop" on what I would have thought was an essentialy unanswerable question.

 

Coffee indeed! Mark needs to spend more time inhaling the concentrated aroma of the LF Forum!

 

All the Best,

Nathan

[jay_de_fehr]

MMMMM coffee.-jdf

[mark_tucker2]

Oh, My God. I just had a Kevin Bacon Moment.

 

Try to stay with me here, as I outline the Six Degrees of

Separation: My rep went to RIT. Therefore, they called her (us) to

speak there two years ago. I had a show there. I met one of the

professors there, who in turn, introduced me to this guy who

invented this scanner camera that Bailey linked. He's a card...

 

Then, even weirder, I was walking around the floor of PhotoPlus

at Javitts in December of last year, and I look over and see this

(wacko) professor with a photo studio set up, outside the RIT

booth. It was a Polaroid camera, something like a 180 or 195

that shot BW pack film, and he'd rigged up a very ugly motor

(think exposed wires and plywood) that VERY slowly pulled out

that white paper tab of the Polaroid. That same ugly motor was

synched up to a floor stand, that spun around in 360degrees.

 

He would approach people and give them a free portrait. They'd

stand on this spinning plywood thing, and hold perfectly still.

Their body would spin around in a complete circle, as the

Polaroid tab was being slowed pulled out of the camera, as the

picture was being exposed.

 

Go figure.

 

I always feel like I'm one of the weirdos at PhotoPlus. But after

meeting this guy, I left there feeling very

"Baptist/MissionaryPosition/Republican".

 

-MT

[brian_c._miller]

Wow, Nathan, I get to be a nabob? All right! My status in the world has risen remarkably! Merriam-Webster dictionary: "nabob: a person of great wealth or prominence."

 

I think that Mark wants to do this without too much fuss and bother. And I'm not sure that an Epson scanner would be the greatest parts source. What the prof did is more like a stunt than anything else. The scanner doesn't have the resolution of a real negative, it requires computer and power, and I bet that the more sophisticated the scanner, the more software interlocks it has.

 

Thus, I differentiate between something "being done" and something "being practical." And possibly destroying a $350 scanner isn't something I'm willing to recommend.

 

Mark:

That said, I'd recommend trying something first by simply covering the scanner glass with bleached onion-skin paper. The light from the camera lens will form an image on the paper. Set the scanner to use the light source in the hood. Scan. See what you get.

 

When I bought my 1934 Schneider 360mm lens, I tested out the coverage at the camera store by using a piece of large paper taped over the back of my Calumet C-1. The technique is good enough to have a decent image.

[mark_tucker2]

After reading these responses, and reading that RITstuff, for

some reason, the sentence appears in my mind: "You know,

some questions are worth pursuing. And others, maybe, have a

somewhat shorter lifespan".

 

This could easily fall into that category, where you spend

fourteen grand, and thirty-seven trips to Home Depot, and a year

and a half, and then you open the paper one day and read that

some Taiwanese company is introducing the very same product

that'll retail for $99.95, with a $50 rebate.

[george_jiri_loun]

Mark, you got the point. Those producing scaners are also able to see all its possibilites...

[john grunke]

Yeah.... Next!

[jim_galli4]

I'll take the coffee if Mark doesn't want it. I'm having enough trouble making a traditional image. I think the coffee would help.

[brian_c._miller]

Another thought that I thunk in my head is that with my 1650 it is cognizant of whether I am using the 35mm strip light in the lid or whether I am using the auxilary 4x5 light. When I change lights, I have to unplug the scanner (no off switch), swap the lights, and then plug it back in. Then the scanner fiddles with its initialization, and is happy.

 

If I swap lights without unplugging the scanner, it thinks that the previous light source is still attached. Both light sources have a spot on them where the scanner calibrates itself.

 

Thus, the Epson 2450 may be conned into thinking all is well, but the driver software will automatically mask out the preset portion of the light source. Thus, 4"x9" is all you'd get.

 

I think the best route would be to build two backs: one for the ground glass associated with the scanner (focusing), and another for the scanner itself.

 

What you need is a scanner which has a transparency adapter for 8x10, then fool the scanner into thinking the adapter is mounted, and provide a substitute white light calibration.

 

Actually, the scanner should automatically focus on the image. The film holders for my 1650 place the film appx. 1mm above the glass. Therefore, the scanner has a leeway of at least 1mm from the plane of the glass to the film surface.

 

Here I am, Too Much Coffee Man!! :-)

 

OK, so two backs (ground glass and scanner adapter), and the image area is 4-x-something, and its tethered to a computer, and it has less resolution than film, and you'll either have to photograph stuff which won't melt or else have expensive high-frequency flourescent lights or be limited to daylight, and an image will take a few minutes.

 

But it'll be a scanning back! Look, ma, no film!

 

Hey, try this out: put a slide into your enlarger, and the scanner underneath it. Focus for the top of the glass. Can the scanner get an image? If yes, you're on your way to an inexpensive digital camera back!

[doug_dolde]

I use my Epson 2450 frequently for "photographing" 3D objects. I leave the cover off and drape the grey side of my dark cloth over the object. It will hold focus for a couple inches. I get really good images this way.

[struan_gray]

My dickering about with my Epson 1200 says that for daylit

scenes the need for bright lenses is exagerrated. At f5.6 with a

piece of translucent drafting film as a diffuser I had no problems

getting a bright enough image using the reflective copy or

transmissive scan modes.

 

The real problem is getting the illumination even over the surface

of the scanner bed You have to elminate ground glass hot spots

without introducing artefacts like the lines you see in camera

screens with a Fresnel. I can think of ways of doing this using

diffractive optics pressed from sheet perspex, but for a one off it's

going to make a real scanning back look cheap.

 

So the only 'good' way for homebrewers would be to dismantle a

scanner and use something like Davidhazy's technique. For

really good results, you're going to have to write your own driver.

 

That said, if you can live with hotspots, or with the residuals from

removing Fresnel rulings from your images, and if you're not

looking for ultimate resolution, Canon make some nice thin

scanners for around $100 which are small and light enough that

with a few modifications they can be slotted into a camera like a

normal back. They run off USB bus power, so portable use is

even conceivable.

[mark_sampson]

It should be noted that there are already scanning backs for large-format cameras. I've tried the Betterlight in a workshop at RIT a few years back- it's very impressive. So someone has invented this wheel already.

[brian_c._miller]

The cheapest Betterlight is $6,500. Compared to $150-$350 for a tabletop scanner, $100 worth of hardware and $??? of sweat, blood, and tears, the potential is captivating for those who want a scanning back.

[tim_atherton2]

"I use my Epson 2450 frequently for "photographing" 3D objects. I leave the

cover off and drape the grey side of my dark cloth over the object. It will

hold focus for a couple inches. I get really good images this way."

 

I've been doing this on a larger flatbed to scan/photograph small museum artefacts - bone needles, arrow heads, tools etc - works great.

 

I also saw some stunning flower scans made this way. which I mean to try sometime.

[pete_andrews]

I was going to make a panoramic scanning camera out of an old colour hand-scanner once (remember those things, and their snaky, shaky scans?). The project never got further than the drawing board.<p>Adapting a flatbed shouldn't be too difficult, though. All flatbed scanners already contain a lens, so there's no need for a groundglass and an additional lens. A simple refocussing of the existing lens is all that's required, with maybe a large front-silvered mirror to redirect the view of the scanner.<p>Remove the glass platen of the flatbed, disable the built in fluorescent tube light, readjust the lens focus (fiddly), or stick a negative diopter of about -2 in front of the lens, add one 12v battery and a laptop computer to make the system portable, and away you go.<p>Hopefully the software of the scanner will allow the gain of the CCD to be adjusted sufficiently to adapt to varying levels of daylight.<br>(Most scanner software preadjusts the gain or exposure time by 'looking' at a white reference strip through the glass platen, just under the plastic cover where the CCD and mirror assembly is parked. I suggest that this opaque white strip is removed, and a slot cut in the scanner cover, so that a piece of translucent diffusing material can act as the white-point reference.)<p>Some scanners (eg HP 5100c), will only operate correctly if the scanner head detects a black patch in the lid at the 'parking' position. These scanners won't work without the top case cover in position. Most scanners probably won't track correctly either, unless they're positioned horizontally.<p>Good luck!

[roger krueger]

Pete: Nope, won't work very well to just refocus the built-in lens. Your subject area parallel to the direction of travel is limited to the amount of travel, your subject area perpendicular to the direction of travel is normal, hence massive dimensional distortion--see my previous post in this thread.

 

Even if this weren't the case, you've still got a lens set up for 1:2.25--it's not going perform as well at infinity, and it could easily not even have enough coverage. Don't count on having enough room in the focus mount to move the lens back this far either.

 

On my Epson, it's also an immense struggle to refocus the lens at all--you loosen a locking screw and push the lens fore or aft, then retighten. Then reassemble the scanner, and make a test scan (an inclined ruler works best). Evaluate the test scan, tear it all apart again, move the lens (how much?? hard to say, there are no markings, and you're dealing in fractions of a millimeter) and try again. Repeat over and over. Took me two hours. On the one hand, my scanner is now focused better than it was at the factory. On the other hand, wild horses couldn't make me loosen that d*** little screw again.

[pete_andrews]

That was a severe lapse of thinking on my part Roger. Yes, you'd have to translate the linear movement of the lens and CCD carriage to an angular displacement. Which is what I planned to do with my redundant hand scanner originally.<p>OK. You could uncouple the drive cord or rubber belt from the scanning stepper-motor, and gear it to rotate a front silvered mirror - that should work. Maybe you could use one of the mirrors already in the scanner. The mirrors are only clipped in place in many flatbeds.<p><i>"On my Epson, it's also an immense struggle to refocus the lens at all -"</i> Yes I know. Been there, done that! That's why I thought a negative diopter might be easier.<p>Even though the built-in lens isn't of amazing quality, it would still give far better results than trying to scan an image projected on a sheet of ground glass. That ideas a definite non-starter.