Page 3 of 4

Re: Which one is better? Platen's angle 100 or 90 degree?

Posted: 15 Aug 2017, 12:38
by duerig
Ah. You did mean a divider between the two lights. I didn't understand that when I first read your post. This is very interesting. I might give this a try in the near future to see what happens. I happen to have a relay and plenty of lights and can rig up a test lighting module.

-Jonathon Duerig

Re: Which one is better? Platen's angle 100 or 90 degree?

Posted: 15 Aug 2017, 13:23
by dpc
How about synchronizing and sequencing the lights and cameras? Camera 1 fires with only the RH light on. Camera 2 fires with only the LH light on.
Sure you can do that. Downsides of this are that now your scanner requires two lights instead of one and you'll need a switching circuit or a moveable mechanical baffle that blocks the light temporarily. I suppose you could also create a scanner that has a single moveable light that would shift right or left between the two shots (or move the platen cradle left/right?).

I've modified the drawing and shaded the area with a yellow color where the light for camera 1 could be placed and not show up as direct or reflected light by camera 1...

Image

So where would you put the light within that yellow shaded area? Realize that by lowering the light it will also be moving it closer to the outside edge of the platen glass. This will result in the light hitting the pane at more of an angle, resulting in less light being directed back into the camera (image will appear darker).

Since the light is closer to one edge of the page the image will be brighter on the outside than farther away in the gutter (intensity = 1/d²). This problem also occurs when the light is located farther away from the platen in the conventional single-light design, however due to the increased distance between the light source to the outer edge and the light source to the gutter the effect isn't as pronounced (I can provide an example of this if it isn't clear).

-d

Re: Which one is better? Platen's angle 100 or 90 degree?

Posted: 15 Aug 2017, 15:31
by duerig
Wow. You two are geniuses. You have now put together all the pieces for a new solution to lighting.

@dpc is absolutely right that there are two issues. One is the direct reflection of the lighting. The other is the indirect reflection of the opposite side captured by the camera. Let's talk about how to bring together your ideas to solve both of these issues.

First, As @dpc notes, we simply put the lights in the yellow regions. This alone will solve the direct reflection of the lighting. It may mean that the lights are further away from the platen compared to the cameras. But they will still be much closer than they would otherwise need to be.

So now only one light is ever on at a time. And the lights are separated by a fair distance, probably 200mm or so based on my calculations. We add a baffle to each light individually so that each light only lights up its facing page. The baffle specifically prevents direct light from falling on the opposing page. We can use the same kind of calculation here for designing the specific placement of that baffle.

Now we still face the second problem to some extent because the closest 100mm or so of a page will potentially contain the reflected image of the other page (at 90 degrees). But the other page is now in darkness. So any reflection will be a reflection of only a very small amount of indirect light and shouldn't impact the quality of the image.

How does that sound? We end up looking similar to @jck57's original sketch except that the two lights are a bit higher up and further apart. And then each one has its own baffle to prevent its light from spreading to the other side.

This seems worth prototyping to me.

-D

Re: Which one is better? Platen's angle 100 or 90 degree?

Posted: 15 Aug 2017, 16:04
by dpc
You'll still need to toggle the lights on/off for each shot or come up with a covering baffle that totally eclipses the light source so that it won't show up in the reflected platen image. I'd be concerned about turning the lights on/off that much during a scan but maybe that's not an issue.

Instead of a baffle all you need is a piece of cardboard covered in black velvet and place that over the platen glass that isn't currently being photographed. Nothing will reflect off of it and affect the adjacent platen. You'd flip this back and forth between the platens depending on which side you're shooting.

What problem are you trying to solve by lowering the light position? You'll have the problem of uneven lighting to deal with and now two lights that you'll have to toggle on/off (or some sort of mech baffle to move around). Is doing this worth the trouble?

Re: Which one is better? Platen's angle 100 or 90 degree?

Posted: 15 Aug 2017, 16:17
by duerig
I'm not certain what jck57's goals are here.

I know that for myself, a lot of expense and time when building Archivist Quill kits goes into handling the 100 degree platen angle. We custom grind the glass to a nonstandard miter. Many of the aluminum beams are custom cut to 50 or 40 degree angles. We lasercut a lot of custom metal brackets because standard 90 degree brackets won't work. And the whole assembly is quite tall because of the distant lights. So if a system like this would allow a more compact system that requires less labor and expense to manufacture, then I could reduce the price of a kit and so make them available to more people.

-D

Re: Which one is better? Platen's angle 100 or 90 degree?

Posted: 16 Aug 2017, 09:38
by jck57
Yeah, I want the 90 degree cradle for the same reasons. Once you get your cameras set up for the lights, you could have the individual lights fire just as their corresponding cameras fire, almost like flashes.

Re: Which one is better? Platen's angle 100 or 90 degree?

Posted: 17 Aug 2017, 11:42
by dpc
If you're going to use a 90 deg platen you'll get adjacent platen reflections no matter how many lights you use.

The light will be reflected off of the platen in the camera's view onto the adjacent platen, which then will be visible in the reflected viewspace. The only way to eliminate this is to photograph one platen at a time (as you've proposed) and temporarily mask off the adjacent platen with some material that eats light. This doesn't require two lights, but (as we've discussed) by using two switchable lights you can position them closer to the platen than a single overhead light. I would shy away from this though. The lights would be at a lower angle and closer to the platen causing more dramatic uneven lighting across the page. I also would have some concern about turning those lights on and off so frequently (does this cause additional stress? do they take time to stabilize to reach the proper color temp?).

If you do choose to implement these things, you'll end up with a scanner that might be easier to manufacture but it will be more difficult to operate. I would think the goal should be the opposite. We (as scanner designers) should be attempting to reduce the per-page toil on the operator and make it as simple as possible to acquire high-quality photos of printed material. We do the hard work up front (laser-cut brackets, non-standard angled cuts, etc.) so that it makes life easier for the person who is going to spend tens if not hundreds of hours scanning books.

Anyway, that's just my take on it. It's good that we're having these brainstorming sessions and discussing some of the problems with advanced scanner design. Things have been kinda quiet around here over the last year.

Re: Which one is better? Platen's angle 100 or 90 degree?

Posted: 17 Aug 2017, 12:41
by duerig
@dpc, I want to understand you better. So let's simplify the situation somewhat. Suppose we have a V-shaped 90-degree platen. Neither side is masked by black cloth. Let's just worry about a single light and a single camera. For this example, I don't care about the distance of the light. So let us suppose that it is relatively far away and provides an even coverage of the page. The single light is carefully baffled so that the direct light it emits is only directed at one side of the V-shaped platen.

So this means that the one page the camera is pointing at is brightly lit. While every adjacent surface to that page is much more dimly lit because of the various reflections from the glass on that page. One of the adjacent surfaces is the opposite pane of glass.

I've done some numbers and with a 90 degree angle platen, and with a reasonable distance between the camera and platen, the 50mm of this page closest to the gutter will have a reflected view of the opposite page overlaid on it. But I propose that if the opposite page is only very dimly lit by indirect light, which it will be in the above scenario, that the quality of that 50mm section of the scan may be just as good. I will hopefully test this idea soon. Do you think that even if the opposite page is only dimly lit that there will be a significant impact on quality in this case?

At a higher level, this discussion is about changing what kinds of trade-offs we can make. In jck57's case, he is working on his automated scanner. So even if this only works when a black covering is deployed on the facing page, it won't make the operator's job any more difficult. That is just one more mechanism he adds to the machine. And so whether it is worth it will depend on whether the complexity of the new mechanism is a good trade for simplified 90 degree brackets and construction.

In my case, this is only useful if it can provide high quality scans without the need for a movable black cloth. If it turns out that even a dimly lit opposing page impacts the quality of the target page without a black covering, then I'll have to pass this idea by. On the other hand, simply switching lights on/off via a relay is not an insurmountable difficulty for my designs because I can just add that functionality to Pi Scan. Or figure out a way to use external camera flashes which are designed for these kind of momentary triggers.

I don't think either one of us wants to design a scanner which is harder to operate. If we can't figure out a way to make it as easy to operate as current designs, then maybe the trade-off won't be worth it. I will say that there are interesting cases where a black covering is very useful. One thing I like about jlev's impost scanner design is that the cradle angle is adjustable. On most books, you can open it wider and scan both pages at once. But on fragile books, you can make it very narrow and use the black cover to scan more slowly.

Another way to think about this is that the current price of one of my kits is above what many people can afford for a scanner. So I am always on the lookout for ways to either make my kits more cheaply and expand access to scanners that way, or for ways to provide a new model that I could sell alongside the current one for a cheaper price. This is one case in which trade-offs that make kits easier and cheaper to manufacture might be worth a trade-off that makes it a bit more cumbersome to scan. I'd love to be able to say 'Pay $600 for a high quality scanner that lets you scan at 800 pages/hour or $1100 for a high quality scanner that lets you scan at 1200 pages/hour'.

Re: Which one is better? Platen's angle 100 or 90 degree?

Posted: 19 Aug 2017, 16:30
by dpc
Do you think that even if the opposite page is only dimly lit that there will be a significant impact on quality in this case?
Difficult to say. There are a number of variables that contribute to the overall impact of reflection artifacts. It could be that it isn't noticeable for black text on white pages, but could show up more pronounced when scanning pages containing photographs.

I think you'll have a real problem containing the light to just one of the platen panes. I don't think you'll be able to get your baffle close enough to the platen to mask the light effectively without it being in the camera's view.

Re: Which one is better? Platen's angle 100 or 90 degree?

Posted: 18 Sep 2017, 12:45
by daniel_reetz
dpc wrote: 19 Aug 2017, 16:30I think you'll have a real problem containing the light to just one of the platen panes.
This is the problem, as I see it. But it's still worth trying. A simple way to test the idea (if you have the gear) is to set up a digital projector (like a pico projector) and display an all-white image. Then use the sharp edge of the image at the gutter of the page. That's as sharp as any mortal is ever going to make that edge.

One of the odd confounding factors here is that you need the page of interest to reflect light. That's how the camera sees it, after all - light reflects toward the camera. But most pages are effectively lambertian, which is to say that they diffuse light and sent it in all directions. So even if you illuminate only the one page perfectly, some light will still impinge on the opposite side of the book, and the glass covering it.

I'm pretty sure there was, at one time, a commercial scanner that did the alternating lighting thing. I know I tried it a long time ago. I remember that the results were interesting, but at the time I was interested in other things so I didn't go much further.