Using SORAA Snap System Beam Spreader - Uneven Lighting

[camwow13]

Yeah I'll probably break down and buy some power adapters and a Rasberry pi for slicker control. Can Pi Scan import Raw files from the cameras as well as JPG? I'm going to be shooting in Raw since the A6000's default color profiles leave a lot to be desired.

I actually have a Kobo Aura One. Only just got it too as I found a crazy deal for it at an overstock store that didn't know its value. I do run KOReader on it The forum decided to say I have a Kindle for some reason.

[adi7]

These images looks very good.There is some reflection from light on the image which is not edited, but anyway,the image looks very good.
Konos93,your platen is at 90 degrees? Because there is a reflection from opposite glass on the page near binding.

There is someone who verified if the reflection ,which comes from the position of light/camera, is present on brand name bookscanners (Qidenus,Atiz)?
because at this scanners the light seems to be much closer to the glass. Thanx.

[camwow13]

I bought some Lenticular lenses direct from Tenrec and did a cursory comparison here: https://imgur.com/a/lbu5AjO

As you can see the biggest difference comes from when I properly align my cameras. The lighting is still uneven but it's reaching a point of acceptability to me. I'm a perfectionist and I have to admit this worst case scenario example is looking very decent. My "normal" test pages from other photo books look excellent and text based scanning is downright fantastic.
My biggest problem as seen in the examples is actually the reflection near the spine of the opposing page glass. This is really only seen with dark pages and I can't figure out a way to get rid of it besides covering the opposing page with something dark when scanning dark pages. Not really ideal.

I finally got my second A6000 and will be completing final setup for a proper two camera setup. I also verified ARW raw files can be transferred from camera to storage by Pi Scan, albeit very very slowly.

So I'll keep getting things put together, calibrated, and adjusted. I'll post a proper comparison of lighting systems once I'm production ready for everyone to reference.

[adi7]

Your platen is 90 degree? The reflection near the spine is linked with 90 degree angle?
You use Soraa lights. You use also SoraA snaps?

[camwow13]

It's 100 degrees, I CNC'ed and laser cut the parts directly from the Archivist plans so it's almost exactly to the specifications laid out on diybookscanner.org/archivist although with lots of deviations in fasteners.

This entire post is about the snap system, haha. As seen in my above linked examples, the 60 degree snap spreader works almost as well as the 10 LPI Lenticular lenses from the Archivist kit. Even with the kit lenses I'm still struggling to get the lighting as even as possible and reflections under control.

I only just got my second camera so now I'm sitting down to fine tune things and make more detailed comparisons.

[adi7]

The reflection from the opposite platen near spline was characteristic of our first 90 degree scanner. We changed the angle to 100 degrees and is gone.I'm not very happy to see that can be present in certain conditions even at 100 degree.
Our big problem is with the reflection of light near edge of an A4 book ( the book edge reflection seems to be more problematic then at 90 degrees).
In your pictures it seems that you moved the camera back,or zoom less, right? because I do not use zoom, is an Nikkor 40mm/2.8 lens and we must move the camera .For us this is a tricky problem because of time spent with camera alignment of the camera.
Because of light reflexions problem ,I scan only small books.

Maybe there is some special glass which eliminate reflexions?
Or some special lens filter?

[duerig]

The secondary reflection from the other side is only completely absent at 180 degrees (flat platen). As you decrease the platen angle more and more, the secondary reflection will slowly spread from the gutter side and eventually if you had a 0 degree angle (completely parallel), you would see a complete hall-of-mirrors effect. Assuming there was a way for light to get inside in the first place.

Given that, moving from 90 degrees to 100 degrees greatly diminishes the size of the secondary reflections, but does not entirely eliminate it. In general, making the V-shape flatter is a trade-off. Reducing secondary reflections means that the cameras are closer to the overhead lights, so the lights must be moved further back to reduce primary reflections. And of course the flatter the platen, the harder it is on the book. Given all of this, 100 degrees is a good compromise between all of these constraints. At this angle, the secondary reflection is on the (usually white) gutter of the book where there is typically no text because of print margins.

You can fairly easily set up a system in CAD software that represents the overhead lights and cameras as points. Then you can use lines and constraints with lines as rays of light that bounce off of the platen (thinking of the platen as a mirror) and return to the camera. Simulating both primary and secondary reflections can give you a visual idea of how everything works. Well, it is easier and gives a better intuition about it than solving the trigonometry equations directly at any rate.

[dpc]

Something to try....

Put a piece of white paper over one side of the platen glass (place OVER the glass, not underneath). Put a black piece of paper (or better, a piece of black velvet fabric) over the other side of the platen glass. Photograph the white paper. Next, replace the black paper with another white piece of paper and take the same photograph of the other side. Now compare the two images. Do you still see the uneven lighting? If so, it isn't due to light coming from the adjacent page.

I would imagine that the uneven lighting you're seeing is due to the outside edge of the page being closer to the light source and not from a diffuse reflection of the adjacent page. Moving the light farther away can help with this due to the 1/d^2 lighting intensity falloff.

[camwow13]

Thought I'd give an update on where I'm at. Thanks everyone for the help and suggestions! Really appreciate it

The explanation about the spine reflections makes sense. It's a bummer I can't get rid of them completely, but fortunately they only show up in my worst case scenario tests. Thanks to my shoddy construction skills I managed to chip and scratch my glass so I'll probably have to replace it. This time I think I'll talk with some local glass shops rather than my hardware store. Some of them carry some reasonably priced anti-reflection glass from Tru-vue and I'll be able to actually miter the glass to 50 degrees this time. I'm aware you can get some odd results if the AR coating starts to rub off, so I might just go with regular glass again.

I'm using the stock 16-50mm lens on the Sony A6000 which has to be zoomed to 25-26mm every time I power them up. This fills the frame with the cradle wing as recommended in the calibration instructions. I'm still trying to get the calibration dialed in. Since I'm aiming to scan a ton of yearbooks, I want to keep manual de-skewing to a minimum. I don't think it's possible to get it 100% perfect, but fortunately Lightroom's transform tool works great and can be applied 'en-mass to all the appropriate images.

As for "regular" books I've definitely hit production level with those. Got a few scanned in this week while playing with Pi-Scan and have been getting excellent results in conjunction with ScanTailor Advanced. The BW processing, margins, and various auto corrections make most of my precision pretty pointless haha, but the results are great

One note on Pi-Scan though. It doesn't really play well with A6000's. When it works, it works fantastically, but the connection can be finicky. To connect I have to turn one camera on (no cameras detected), turn the second camera on (detects the first camera), then turn off the first camera (detects all cameras), then turn on the first camera (camera doesn't see connection), and finally fire a test shot (both cameras will sync up and fire). During shooting I've had one of the cameras (both have done it) drop out which causes Pi-Scan to hang after pressing the shutter. I've found that disconnecting and reconnecting the USB cable and then shooting another shot fixes this issue, though I have to make sure to shoot the pages again as some pages will be missing in the end otherwise. I've also fixed it by rebooting the Pi and disconnecting the cameras and removing their batteries, but I think the aforementioned method works consistently. Whatever this bug is, it seems to have defined parameters that I'm learning to work around. I'm unsure as to the root cause. Pi-Scan is using a slightly older version of gPhoto, but the Sony A6000's tethering isn't the best and the camera itself might be causing the problem.

In any case I'll spend the next week\weeks figuring things out and report back with what I've found and some examples of what I've got dialed in with the lighting.