Build Thread: Tulane CIPLC (now with a TON of photos!)

Built a scanner? Started to build a scanner? Record your progress here. Doesn't need to be a whole scanner - triggers and other parts are fine. Commercial scanners are fine too.

Moderator: peterZ

Post Reply
benjamin
Posts: 58
Joined: 04 Mar 2014, 00:53

Build Thread: Tulane CIPLC (now with a TON of photos!)

Post by benjamin »

Well, this first post is kind of just a teaser, because while I've taken photos of every step so far, I don't have them with me. That said, the "beta version" is complete, and I just wanted to share- will add to this thread as more photos are available, and as the scanner develops. The goals for this project were a little different than those of folks building personal scanners- we have to scan and OCR several dozen linear feet of government publications; which have the benefit of coming in only two horizontal dimensions (and only two fonts). To accomplish this, our scanner needs to be sufficiently foolproof such that it can be operated by research assistants with minimal training; capable of being transported by car to area libraries; and flexible enough that when the project is finished, it will retain utility for other interesting as-yet undefined applications.

So, here it is, well, the "rough draft," anyway:
Image

...as you can see, we're restricting camera movement to one plane (at least, for now), and using a two-part sliding cradle which sits on a drawer-slide base. We've cut a lot of weight by reducing the base footprint and using a sliding center section rather than MDF. The center column uses components from a sliding door; and the platen is counterweighted on pulleys running cord through the column. The platen and coroplast are easily removed by hand for transport. The USB cables from the cameras themselves will run to a computer, where I'm cobbling together a capture, storage, and processing script that's triggered by a footswitch (described in my last hardware thread).

99% of the non-camera materials for the project came from a) local dumpsters or trash piles; b) architectural salvage stores (Habitat ReStore & The Green Project); and c) scraps left around the house. We did pay retail for the nonslip mat, thumb screws, and drawer slides (though it turns out we could have gotten better slides for 50 cents/each at Habitat).

There's definitely a lot more work to do, but I'm looking forward to feedback from the community to help the fine-tuning process as we move into the next phase!

Whole ton of photos/documentation t/k...
Last edited by Anonymous on 27 Jan 2010, 02:59, edited 1 time in total.
User avatar
daniel_reetz
Posts: 2812
Joined: 03 Jun 2009, 13:56
E-book readers owned: Used to have a PRS-500
Number of books owned: 600
Country: United States
Contact:

Re: Build Thread: Tulane CIPLC

Post by daniel_reetz »

IT'S GORGEOUS. Are those camera arms from crutches??

From your description and the few things I've seen, this looks like a pretty thorough rethinking of the original design. I can't wait to see more but glad you decided to share today.
benjamin
Posts: 58
Joined: 04 Mar 2014, 00:53

Re: Build Thread: Tulane CIPLC

Post by benjamin »

Crutches indeed. Free from the Green Project. Felt a little guilty, pretty sure they were antique- varnished wood everything including arm pads & handles, and the rubber tips were wrapped with twine around the edges.
benjamin
Posts: 58
Joined: 04 Mar 2014, 00:53

Photos

Post by benjamin »

As promised, the real build info.

I started out by keeping an eye out for promising garbage while walking the dog. I quickly found this yard sign in a trash can. That supplied the coroplast and this cool wood which could be used for a sliding cradle mechanism.
Image

Shortly afterwards, I found a renovation dumpster full of mostly unused wood. Without writing a full essay on the hierarchy of dumpsters, renovation dumpsters typically contain a fair bit of "cut-off," leftover scrap wood from whatever framing/repair the contractors were working on. Exercise caution, though, because they'll also contain whatever nastiness was removed from the building. Note that having a helper to trim long pieces can be handy.
Image

Remember to check for and remove leftover nails and staples!
Image

I cut a section from the signpost to use for the cradle support.
Image

From the beginning, I wasn't crazy about the hinge design of the original Instructable, and the only drawer sliders I could find were the side-hung kind. I played around for a while using two hinges- one at the platen and one on the column, but wasn't happy with the result. Then at the Green Project (http://www.thegreenproject.org) I found this funny little object with four wheels and a bolt sticking out of it, and started to make a track for it.
Image

I had to cut pieces of wood to make rails for the slider. To finish it, I would have had to make a top for the rails with a gap behind for the bolt, but in initial tests the wheels started binding on the cap.
Image

Fortunately, at the Habitat ReStore (http://www.habitat-nola.org/restore/) I found more of these objects being sold with rails- turns out they're tracks for sliding doors. This one can take any 1/4" bolt and uses two of them (not shown), which adds some rigidity. Note that using washers will keep the bolt from chewing up the side of the platen over time.
Image

Then I cut the aluminum sliding door rail to fit the column. This size was pretty arbitrary, it just has to be high enough to turn the page underneath. I had a three foot piece of cutoff, so I cut the track to leave space for a) bolting the column to the base and b) cutting holes for the pulley above it. I didn't put a stop on the track, but should. This picture is totally fake, btw. Shot it, then started to cut, then gave up and tried the circular saw, which ripped through the aluminum like butter.
Image

I cut the coroplast to 11x15 using the leftover track as a straight edge. Utility/exacto knives are great for precision but can "wander" pretty easily in my experience; so I frequently opt instead for a sharp BFK.
Image

I made the base out of dumpstered 2x4's, which had been ripped to like 2x3. The sizes were a little arbitrary- 15x24". The weight of the 2x3's and the column at the back mean the scanner won't tip even though the cradle actually extends about 2 inches forward. Smaller base means less weight/more portability! I attached the sliding cradle support to two pieces of plywood and attached them to a pair of side-mounted drawer slides inside the base. I then mocked up a cradle support out of cardboard & scrap wood and taped some shims to it so I could start brainstorming camera arms.
Image

Then I cut the cradles. Here too the size of the triangle is arbitrary, I just cut 45 degree triangles out of the front of a very cheap drawer that had fallen apart, and used a piece of fence post for the bottom, cut to fit inside the sliding base. I used panel nails ($1/pound at the Green Project) and wood glue, and glued in two little shims for extra support. I drilled two 1/4" holes exactly halfway through the base- into one goes the bolt to attch the cradle, glued into the other is a dowel to keep the thing aligned straight. Here the dowel isn't yet glued/pushed down. If I had more time, I would have used a lot more dowels in this project- I love the seamless finishes you can get, they're less likely to loosen over time, and they're an excuse to use my vintage Stanley 59 jig.
Image

Here's the finished base. I added velcro to attach the coroplast to the cradle. This is helpful for two reasons: a) you can remove the pieces for transport; b) it's easier to adjust the spine width with them removed. In future versions I won't bother with the slide, just use the bolt + dowel spaced exactly an inch apart, and drill 1/4" holes every .5 inches or so across a flat piece of plywood inside the sliding base.
Image

Clamped everything back on and started experimenting with camera angles using a piece of 2x cut with a miter box. Our scanner will be used by law students who are both super-busy and not really imaging geeks, so I didn't want to use an articulating arm and introduce all that room for error. We know that the angle from the floor will always be 45 degrees to match the 45 degrees of the platen/cradle, and that as long as the book is centered on the cradle, that angle can always be achieved with books of varying sizes or spine sizes by adjusting up or down and zooming in/out (using only the optical zoom). This photo represents the highest the camera would ever need to go- capturing the entire cradle area. A later version may introduce a second sliding mechanism allowing the camera to go in/out, eliminating the need to zoom, but it's unclear if in order to be usable, the camera would have to "zoom" in to the point where it was interfering with the sliding platen, and potentially getting too close to the boundary point between the camera's regular and macro focal distances.
Image

Glued the nonslip mat to the coroplast using some automotive headliner spray adhesive that I had lying around. I used the leftover aluminum track to smooth out the bubbles, then glued it & let it dry under weight (I used 45lb fitness weights). Finished the edges with black duct tape. Then attached the velcro and clamped an angle into the exact center in order to doublecheck my arm height and angles.
Image

It took me a while to find a single-plane adjustable arm. Finally at the Green Project I saw a nice pair of wooden crutches. They should be cut such that at their lowest point the tip of the crutch is level with the edge of the cradle, and their highest point is just above that line drawn in the last mockup.
Image

I put a handle on the platen (which is just two picture frames with one side each removed, cut at a 45 degree angle on the miter box, then nailed to two more triangles made from that busted drawer. I was planning to paint everything using a can of leftover flat black paint ($2 at Big Lots); but it had gone rancid and the smell was horrible... so instead I used some leftover quick-dry stain. In an ideal world I'd still use stain, but a much darker one (and I'd use new wood, route the edges, etc etc etc...). I may still need to paint the crutches black if reflection is an issue. Held my breath and used a tiny bit of the black paint to color the edges of the platen triangles, entirely to cover up some sloppiness in the cuts.
Image

Crutches attached. Note that here they're still a little too long. Cameras are currently attached with corner braces and thumb screws, but that will change to a mount plate- again, to remove the possibility for user error. As it stands it's pretty easy to turn the camera until the bottom of the frame is aligned with the binding of the book, but again, this is something that can be calculated and fixed in advance. Remember that it's important to align the camera arms such that the lens is centered over the image, it's easy to accidentally center the arm instead, which will create a 1-1.5 inch offset! Easiest way to do this is using a string with a weight held at the center of the lens and lowered to the base. Also in this design we have to compensate for the extra ~1.5 inches that the cradle hangs over the front edge; it's easy to center the thing over the base and be off by another .75 inches.
Image

I drilled two holes in the column, one to support the two pulleys, and one for the line to pass through. Two pulleys are necessary to keep the cord from dragging. I used a double-pulley because I found it cheap, but it leaves the door open to make a lever or foot pedal to lift & raise the platen in the future. The cord is just tied to the slide.
Image

I found a cool iron cylindrical 4lb window sash weight, but it turns out it's a little too heavy to properly counterweight the platen. Until I get up the nerve to cut it down or find an alternative, I'm using a 1.25lb fitness weight.
Image

Right now for lighting I'm just using a gooseneck clip light. I already have two light fittings that can support higher wattage bulbs, but need to make an enclosure. Many of you seem to have your lights directed towards the pages, but my reading suggests this is a bad idea. Glare is light reflected back into the camera. A little bit of research on reducing flash glare suggests this is least likely to happen when the light source is 45 degrees from the camera; ie, perpendicular to the floor in this design. Of course, the light spread means that not all the angles are a perfect 45 degrees, but having the light directed down and shielded from the sides sure helps. Note also that lumens decrease exponentially the further you get from your subject; so the closer you can get the bulbs to the book, the better (allowing for reflection, etc). I toyed with the idea of mounting the lights to an attachment rising from the platen itself, but decided that the constant movement would reduce bulb life when using filament bulbs to the point where it's not worth it.
Image

Note that I didn't have to worry about a trigger button, because we'll be connecting the cameras directly to a computer. Instead, I built a foot switch that sends a signal to the computer via USB. This is possible because we bought Canon SX100 IS cameras, which have more options available in the Canon SDK & therefore in both commercial Windows multicamera remote programs like PSRemote & open source software like GPhoto2. I've documented the USB button here: http://www.diybookscanner.org/forum/vie ... ?f=1&t=252
User avatar
daniel_reetz
Posts: 2812
Joined: 03 Jun 2009, 13:56
E-book readers owned: Used to have a PRS-500
Number of books owned: 600
Country: United States
Contact:

Re: Build Thread: Tulane CIPLC (now with a TON of photos!)

Post by daniel_reetz »

Now THAT, my friend, is a build thread. Awesomeness level = max.

Easiest way to do this is using a string with a weight held at the center of the lens and lowered to the base.

Clever, clever. Wish I'd thought of a plumb-bob when I built the scanner for Aaron... his scanner is a little off. ;)
benjamin
Posts: 58
Joined: 04 Mar 2014, 00:53

Re: Build Thread: Tulane CIPLC (now with a TON of photos!)

Post by benjamin »

Plumb-bob. That's the word I couldn't remember. Also I've now replaced the counterweight with a cloth bag stuffed with assorted hardware bought for this project but unused; fine tuned so that the platen is biased slightly downward, which helps seat it a little bit.

On a personal note, this brought back an intense flashback to high school AP Physics, where on the exam we were asked "How would you calculate the weight of this object without placing it on a scale?" I was the only one in my class actually taking the AP exam, so we hadn't gotten to the finer points of all that weight/ mass/ accelleration/ gravity stuff. I'm pretty sure my whole answer was "Get a pulley and a bag of sand, achieve balance, place the sand on a scale." I drew a diagram and everything. Needless to say I did not enter college with a Physics credit, however this literal reading of questions did eventually help quite a bit in law school.
cratylus
Posts: 30
Joined: 04 Mar 2014, 00:52

Re: Build Thread: Tulane CIPLC (now with a TON of photos!)

Post by cratylus »

This is awesome! How did I miss it before? Anyway, great write up. Keep us posted on the foot pedal developments.
User avatar
daniel_reetz
Posts: 2812
Joined: 03 Jun 2009, 13:56
E-book readers owned: Used to have a PRS-500
Number of books owned: 600
Country: United States
Contact:

Re: Build Thread: Tulane CIPLC (now with a TON of photos!)

Post by daniel_reetz »

If anyone reads the forum, but not the news, AND is in the New Orleans area, you can see Ben's scanner at a totally cool art show.
Post Reply