Scanner distortion

[YankeeBoy]

This may be old hat to most of you but it was an unpleasant surprise to me.

I was starting the build of an out of print Halinski F/A-18C and decided to scan it to be on the safe side when I (as usual) screw something up. I scanned the frame pieces. printed them, laminated them and then the fun began! Part of the frame is two vertical panels with a horizontal panel joining them. On the parts sheet the two vertical panels had their fore and aft dimension oriented up and down the page. The joining piece happened to have the same dimension oriented across the page. When I started to assemble the three pieces the joining piece was nearly 1/8 inch longer than the two side pieces. To make a long story short, my scanner, an fairly old Epson 1670 flatbed "stretched" the horizontal dimension of the scan while leaving the vertical dimension alone. I verified this by printing a 4 inch square, scanning it and printing the scanned image. It showed the same distortion.

I found lengthy explanations on the internet as to why this is a known characteristic of most single bulb flatbed scanners but it wasn't known to me.

Bottom line - if you scan for backup parts you might want to verify that your scanner is not introducing distortion in the scanned image.

If you download those "free" (scanned, copyrighted) models from pirate sites and the parts don't fit it is probably due to scanner distortion (and you got what you deserved.)

I'll have a test sheet with known dimension squares on it when I go shopping for a new scanner this week.

[Darwin]

And, to compound the problem, your printer can do the same thing. If you are really living right, the two might cancel each other out. With my luck, the will be additive.... A good reason for having some kind of ruled size references on the edges of the pages, even if it does take up some of the image area.

[YankeeBoy]

True, Darwin, true. Fortunately my printer is spot on (as repeatedly checked by Gremir Models' printer distortion check) so at least that hasn't been a problem.

[Gharbad]

Interesting, I hadn't thought much about that.

Oh well, more reason for me to stick with building my printed kits, if I ruin it I ruin it! I have enough models to fill my life time anyway.

[whulsey]

The first commercial printed kit I started ran into the same problem. It meant that sometimes I had to scan a page in segments so I could get the longer pieces running in the direction with no (or at least very little) distortion. Pain in the butt, but haven't figured out any other way to work around it.

[Leif Ohlsson]

Thanks for alerting me to this problem. I just recently lost my innocence as regards printers, but managed to solve that problem reasonably well. And now the scanner...

So I printed my printer test document, a rectangle 250 x 150 mm. After having changed printer driver that now prints OK. I then scanned the print in my old Epson Perfection 1260 (older than yours, YB?).

In Photoshop, the measures seemed to be alright. A printout showed no measureable difference for the long side, and ca 0.5 mm for the short side. This translates to -0.3 percent for the short side.

This is rather acceptable. But in the future I will draw two carefully measured lead pencil lines on each page of importance (i.e. parts to be cut out) I scan, 250 mm long side, 150 mm sort side. Any correction then can be made in the graphic programme used to save or read or print the scans.

Whulsey, wouldn't it be possible for you to correct scans in your graphic programme? I will probably have to make the short side 0.3 percent larger in the future, which is quite doable in Photoshop. Isn't there a similar option in Gimps, or what you use?

On another tack - doing this I discovered that exact printing also depends on the software you are using. On my Mac there is a very convenient application called Preview (or something like that; I have the Swedish version). It is very handy for viewing pdf, jpg, psd, tiff, and a score of other formats.

Today I discovered that you simply can't use it for printing. At least not parts for cutting out. My rectangle was off some 5 mm both ways. I think this is very disconcerting; that you can't trust your software.

So I will make it a rule to convert everything I want to print with exact measures, both scans and vector design parts, to pdf, and then print them from Adobe Acrobat. That application - so far - seems to print exactly right for me.

And scans will have to be adjusted immediately upon making them, before final saving,according to the calibration made today (+0.3 percent short distance). Then they can be used both directions for designing, printing, etc.

That's what I learned at school today...

Leif

[AlanG]

The distortions in scanners and printers can be even more complex. I have a fairly old Canon scanner (N676U), which I calibrated early on with a full-page squared grid. I found that the vertical scale varied substantially over the height of the page, though fortunately almost entirely over the top third. This comes out significantly stretched compared with the remainder, so I have to scan things twice, once upright and once upside down, and stitch the scans together. The result then needs to be reduced in height by 0.7% to square everything up. There remains a tiny shear distortion, and an even tinier keystone, but these are small enough to be swamped by errors in my cutting and fitting skills, so I ignore them.

To be fair to Canon, none of these distortions is big enough to be visible on ordinary document copies, and after all they don't sell this class of scanner as precision digitisers.

Alan

[legion]

Isn't this a mechanical 'fault'? If i am correct, most scanners/copiers travel along 2 rails/tubes that are suspended. Could it be that the scanning module causes the rails to sag, increasing the distance between scanning module and the object? Stiffer rails/tubes could solve this, perhaps.

[peter taft]

On the "Page Setup" check that you have "Keep Aspect Ratio" switched on ! What this means, is that if the scanned image is lets say a square 4" X 4", the output image may get stretched in one of the 2 dimensions - for arguments sake, lets say 4" X 5", this is corrected in the page setup with the "keep Aspect Ratio" - in other words, "AS IS" - What you scan is what you get. Hope this helps.

[YankeeBoy]

Legion - There is considerable discussion of the technical whys and wherefores of this problem with flatbed scanners many places online. Google "flatbed scanner distortion."

Peter - I'm sure that would be helpful if there was such a setting for my scanner. There is not.