Fix Keystoning via hardware adjustments

[dpc]

@Eppi,

That's probably the best way to ensure that things remain aligned. I suppose you'll have to be concerned with the camera casting a shadow depending on where your lights are positioned, but other than that it should work OK.

My camera has an onscreen display where it will draw an 'X' across the camera's display so it's pretty easy to know if it's pointed at the center of the page.

[Eppi]

Thanks dpc. I knew there was something obvious, like shadows, that I was not considering

Oh yes, now that you mention it, I have seen X's or grids on camera displays before. Seems like a great feature for a book scanner to have.

Thanks again for the advice

[jay]

Hi all,
I'm new to the forum and have been following this thread with interest. Now I'm just a noob at this (first build still underway), so not as familiar with all the nuances & nuisances of book scanner design as you folks. But I thought I'd add my two cents worth anyway...

I'm following the "New Standard Scanner" build which means that my cameras will each be mounted on a fixed column. However, I am considering incorporating a minor tweak which I hope will help reduce keystoning. The idea is to mount the cameras to each column via a simple sliding (and lockable) mechanism. The mechanism will allow the camera position to be altered at 45º to horizontal, along a line which is parallel to the cradle (see blue line in diagram below ( I'm hoping that a diagram actually appears below, if not then I has mucked it up)).

Blue axis.jpg

So if the camera is positioned along this blue line so that it is directly over the center of the page, then keystoning should be minimised (along one axis anyway). Distance along the surface of the cradle will then be 1:1 with distance along the sliding camera mount.

This plan appears to be similar to the route that the.traveller is taking (lame pun not intended) just using a different plane of movement. Can anyone see any major problems with my plan? I can see that there will be a bit of extra work initially but if I mark some points along the slide which correspond to known points on the cradle surface then I'm hoping that setup time at the start of each book will be reasonable.

The only issue I found was that if the cradles angle decreases from 45 degrees to 0 degrees the height of the camera ( has to be raised to keep the same shot angles) can become pretty high up which would mean digital zooming on the camera.

[pfarber]

arm.gif (2.26 KiB) Viewed 8086 times

The green is something like 1/2 tube where you can get those locking collars for. 3 pivot points and you are done. The head of the arm where the camera goes has a flat reference piece (a few sq in) and you use a piece of squared off wood or metal to calibrate the angle of the head (which the camera is affixed to) and the surface of the platten.

00996298000.jpg (21.05 KiB) Viewed 8086 times

Replace the dial indicator with a camera and make it bigger.

Fixed height means camera is set and forget as well as the lights. You just slide the angled arm up and down in relation to the center of the book on the platten. The red dots are clamps that would allow the tube to be moved about. The brown line is the wood and is cut to your preferred height. You can have several calibrated wood sticks for varying heights for paperback, paper sized, or large format.

All you *NEED* is to keep the camera 90DEG to the paltten. All that math is nice, but nothing a calibrated piece of wood wouldn't fix

[the.traveller]

@ dpc: How did you get that x mark in your display? Is it standard in your camera display or is it something which you added into CHDK or SDK?

@Eppi: A very nice addition towards the explanation from jay.

That was the last part which I have not included in the Excel sheet yet, due to a short holiday. I will come to that in the next month probably.

Your addition is the correct conclusion of what Jay has been explaining about the change of positioning the center of the cameralens towards the book you take pictures from.

@pfarber: You are correct that a calibrated piece of wood will fix it. But for newbie's it is nice that math is there to prove what you are doing is correct. I am not a professional carpenter, I dare to say that I have 2 left hands. So if I can calculate everything and find that what I see from other members is actually feesable for me to do with the numbers I have calculated for my books then, yeahhh.

I hope you can tell us where you can get those locking collars? Your picture gave me a new idea.
The wooden stick is already on my list, because making these with aluminium and plexiglass was not in favour by me because of the potential damage to my lens. Or maybe when I make it shorter and just use a wooden stick placed in the corner and have it moving freely to reach the point where my cameralens is without touching it. I will be able to see that it is pointing to the center of the lens. Then zoom in to get the largest portion of the book in view and ready is my setup to start scanning.
All of this using Eppi's and your idea for sliding the camera into the correct cross section of the book.

@Eppi: The wooden sticks are mentioned in a thread which I can't find so quick to have a link to. In another thread of a member came a solution using plexiglass and a aluminium L shaped bar attached to it, to find the correct middle of the book and because of that the point where your cameralens should be.

[dpc]

The Olympus camera that I'm using at the moment has the 'x' and grid overlays as an option out of the box. I haven't used CHDK so I don't know if it offers the same feature.

BTW, in a V-shaped cradle, the center of the page will move during the course of a book scanning by the width of the spine (book thickness). You might consider picking a camera position that is halfway between the two extremes as a sort of a compromise.

[dpc]

Has there been a scanner build using a fixed cradle and allowed the platen to shift horizontally as well as move vertically?

If you did this and then attached the cameras to arms attached to the platen (instead of to the base), you would get the page in the same location within the captured image across the entire book scan. You'd minimize the amount of wasted pixels due to cropping and use more pixels on the actual page content which is always a plus.

[jck57]

Has there been a scanner build using a fixed cradle and allowed the platen to shift horizontally as well as move vertically?

If you did this and then attached the cameras to arms attached to the platen (instead of to the base), you would get the page in the same location within the captured image across the entire book scan. You'd minimize the amount of wasted pixels due to cropping and use more pixels on the actual page content which is always a plus.

But that's hows the standard scanner effectively works. The cameras and the platen are connected in the horizontal plane and the cradle moves laterally. The standard scanner design features a cradle that adjusts to different thicknesses and is also mounted on rollers. Because the book can move laterally, the platen will come to rest at nearly the same position both horizontally and vertically in relation to the cameras throughout the book scan. Since the spine is kept horizontal, spine thickness becomes irrelevant and camera reframing should not be necessary.