'Pyramid'-shaped scanners (Tamtus, Atiger)

[bnz]

I think I don't understand your drawings entirely. You plan to photograph two pages with one camera while the are lying on the pyramid top? You will have keystoned/warped pages in the result that you have to take care of by software. I'm not sure to what degree you can do this without big loss. I think I recall there was a forum thread about such dewarpings. You also may have some problems with the DPI per page if you capture two pages with one camera I would assume.

I find this design very interesting though as well, especially these constructions with mirrors.

[Anonymous1]

This looks epic. I am definitely looking into making something like this, as a single camera is all I have.

[bnz]

This is the thread I was referring to with the dewarping tests at relatively steep angles.

http://diybookscanner.org/forum/viewtopic.php?f=1&t=636

Incredibly interesting stuff. It probably needs some more work on the software side if you want to use this in some kind of automated fashion.

[ahmad]

I think I don't understand your drawings entirely. You plan to photograph two pages with one camera while the are lying on the pyramid top? You will have keystoned/warped pages in the result that you have to take care of by software. I'm not sure to what degree you can do this without big loss. I think I recall there was a forum thread about such dewarpings. You also may have some problems with the DPI per page if you capture two pages with one camera I would assume.

I find this design very interesting though as well, especially these constructions with mirrors.

You understand perfectly. Like Anonymous, I only have one camera. But it's a good one - a Sigma SD10. I've photographed A1-size architectural drawings with it and had acceptable results. DPI, like you say, may still be a problem, we'll just have to wait and see. There was a thread on dpreview.com forums about document archival with Foveon sensor-based cameras, but I can't seem to find it right now. But anyway, this is the last place to have a Foveon vs Baeyer debate

This is the thread I was referring to with the dewarping tests at relatively steep angles.

http://diybookscanner.org/forum/viewtopic.php?f=1&t=636

Incredibly interesting stuff. It probably needs some more work on the software side if you want to use this in some kind of automated fashion.

I still haven't figured out the exact angle I'm going to use - but needless to say, that thread will play a big role in deciding.

I'm fairly confident about the software side of things. The first thing I photograph on the scanner will be a 5x5mm grid. I will take that image into Hugin and spend some time generating parameters for keystoning and lens distortion. These values will then form part of a script, calling Panotools (the Hugin backend) which will 'develop' images to make them ready for Scan Tailor.

The camera will be tethered to a PC using Firewire. As soon as an image appears on the CF card, the script will

• Develop a greyscale image from the RAW file using Monoveon (through WINE)
• Call Panotools to correct distortion
• Call ImageMagick to crop image to platen borders
• Output a greyscale PNG

A major drawback of my approach is that up to 50% of the sensor space is being wasted - both through keystoning (which also means that I lose resolution as the image approaches centre), and at the sides of the frame because of the aspect ratio of a partially-open book.

So I'm not claiming it's the 'best' way of doing things. Just the best way for me, right now.

[bnz]

This sounds like a really interesting approach. I'd be thrilled to hear some feedback and samples about it once you have it working. I'd probably have tried to do all the processing in Lightroom, but I think your approach will be able to present you the results more directly. Panotools might be a good idea to correct the distortion. I almost forgot about the existence of this toolset.

[daniel_reetz]

Wow, an SD10. I have always wanted a better Foveon camera, I have one of the old Polaroid X530s that has a mini-Foveon.

The one thing that I foresee being an issue is that you'll need the entire book surface to be in focus. That means having a DOF equal to the depth of the hypotenuse of your inverted platen. I don't think that's a serious problem, in and of itself -- it's definitely something you can calculate. But if, at that short distance, you need to use your lens at f/8 to get the required DOF, it might mean that lighting will need to be very bright to counter the reduced incoming light.

Very excited to see what you come up with. I think this would be an interesting thing for lots of SLR owners.