This has been absolutely baffling me. I scanned an image both vertically and horizontally on the flatbed of my HP Color LaserJet Pro MFP M277dw using Windows Fax and Scan. I then used Gimp to rotate one of the images 90 degrees, to overlay it above the other image. But then, the image appears to have been stretched by the flatbed!

Not giving up, I tried on another scanner, the Epson Stylus C88+, and got the same results! Is there an inherent flaw with most flatbed scanners? Why does it seem to skew out the image when compared to its horizontal counterpart?

In the 3rd image below, if I lined up the "A", the "D" gets misaligned. The graphing paper also shows the effects of a horizontal scan.

Why does this happen? Are there any scanners that don't do this? Is it dependent on the angle that light is reaching the photo-sensors?

Finally, can this multi-angle scanner invention have anything at all to do with this effect?

First vertical scan:

Vertically scanned image

Second horizontal scan:

Horizontally scanned image

Merged both scans and flipped 90°:

Merged scanned image from vertical & horizontal images

  • 1
    \$\begingroup\$ How accurate are the scanners? If you scan the same picture twice in the same position do the pictures overlap perfectly? Scanners in all-in-one printers are rarely top-notch... And is it the scan head that moves over the paper or the paper that is moved? \$\endgroup\$
    – xenoid
    May 21, 2020 at 8:59

4 Answers 4


Why does this happen?

A flatbed scanner works with two different processes, on each dimension.

One with a rigid element, the CCD array, which has a fixed dimension.

But the other dimension is given by movable parts, so this movable mechanism is not that precise on your case. It is normally by a motor pulling a belt.

Is it dependent on the angle that light is reaching the photo-sensors?

No, it is more about the motor-belt mechanism.

Are there any scanners that don't do this?

They are not common anymore, the rotative laser scanners, where you mounted your target on a cylinder, and each rotation was better to control the advance of the point where the laser targeted.

But you needed to scan all pice at one pass because what was imprecise was the mounting of the target.

A photo scanner is not precise also because you can put your target photo at a slightly different angle.

The best option is to scan on one "pass", and not dividing your project into different passes.


You write "Using Windows Fax & Scan". Fine resolution Fax tends to have a resolution of 204×196 dpi. There is also "Superfine" at 408×391 dpi. While there are also symmetrical resolutions available in the respective standards, those slightly skewed aspect ratios may be more common and have a longer history.


You're using consumer grade devices and expecting laboratory grade results.

Consumer grade devices can be cheaply designed and manufactured precisely because "good enough" is an acceptable trade-off in exchange for the reduced cost. In your case, "good enough" seems to mean a certain amount of geometric distortion is acceptable.


Probably a lack of mechanical precision. I gave up on a $2k Polaroid medium format Sprintscan because it couldn't advance with acceptable accuracy/repeatability. Seems something cheap is even less likely to get it right.

My Screen 1045ai drum was perfect of course, but an extraordinary PITA in every other way.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.