On 2024-03-24 09:44, Daniel Bauer wrote:
Am 23.03.24 um 23:56 schrieb Carlos E. R.:
On 2024-03-23 14:44, Daniel Bauer wrote: ...
% > % How to stitch scanned papers? % > % I have a map that is larger than my scanners, so I took 3 partial scans. I % > [snip]
If it's only one image I'd simply do it with the Gimp, easy and probably faster than searching for a software, install, try...
Open one image in the gimp. Enlarge canvas to the full end size or larger. Load the other images as layers. Show only first layer and an adjoining one. Make the adjoining one semi-transparent and move it around until it fits. Make it nontrasparent again.
I have two problems here.
One, amazingly, the second scan seems to have a different "scale" or zoom that the first. This is imposible! But it happens. I superimpose horizontal lines, and one centimetre below it, other horizontal lines appear parallel, 1 mm apart.
I did the rotation as best as I could, setting the center at the end of the horizontal line mentioned above. And features anywhere else do not match.
It is amazing. There is distortion. Different distortion in the two scans of the same paper.
And then I try to undo transparency, but I am stuck. I'm probably not doing transparency right.
...
To do this, select the layer and then go to Layer > Transparency > Color to Alpha. ...
Ah, that's something else. Sorry, for me the basic Gimp functions are so very natural that I didn't explain any further...
I'm familiar with a limited portion of the tools in gimp, so new learning, no problem :-) Actually, that initial method makes aligning easier: I made transparent the colour of the paper, so that the map lines stood out. The problem is undoing that change later. Maybe there is an easy way to do that. I might, on the final result, do that transparency method before printing the map. You have seen the colour in the links I posted, it is the kind of chemical copies that were done in the past from tracing paper of a land plot map or house plan. ...
...
Now I have to rotate on the right end of the images, where I also drew a dot. I will use the first dot as centre of rotation.
[...]
Done rotating. See:
https://paste.opensuse.org/pastes/b09e2b4b1b3c
I am aligning on the tiny black dot at the end of the dash to the right of the "35". See how the "36" doesn't align?
Maybe this is a scanner defect in the movement of the head. I used 600 dpi.
Can be that the scanner isn't absolutely accurate, but before buying a new scanner ( :-P ) try:
leave the center of rotation in the middle of the layer, do not move that center to any other position. Then rotate.
Not sure if this helps, but could be.
The defect is only visible with a lot of zoom, the eye doesn't see it. So it is acceptable, but it is a surprise to me, I thought the scanner to be more accurate. Yesterday I thought it was the vertical axis, so it would be irregularities in the scanner motion; then I remembered that I had rotated the image 90 degrees, so it is actually the horizontal axis, and that would be defects in the sensor, differences in distance between the individual cells at the left and right ends of the sensor, which is strange.
(You'll have to start with a new image, not use the already rotated layers)
Now unzooming and undoing transparency... wow, the result is perfect to the naked eye, aside from a different colour (even though I did not change scanner settings in xsane).
The scanner probably tries to do a correction finding out which are the correct colors/brightness/contrast, so if the color mix on the different parts of the map is a bit different (e.g. more red lines), it will result in different correction.
Maybe you can turn off "automatic exposure/color correction", or save the correction from the first scan and apply it to the others, but I don't know.
The scanner doing a colour correction of its own between images is not a thing I thought it could do; I thought the colour correction was solely done by software, by xsane. And I don't have it on auto. I did read somewhere that scanners do a calibration run, for calibrating the lamp, when they power up. There is some kind of calibration matrix at the top of the glass beneath the plastic, so not seen from outside. At least that is what I read, I have not disassembled one to see.
You can try to adjust the colors in Gimp, but then I'm sure your next post will be in two days :-) There are so many different ways to do that... (Save an xcf with the moved/rotated layers and then work with a copy so you can always go back to the original, or duplicate the layers and work with the duplicates)
It also depends on how important is a perfect result. In most cases "good enough" is good enough...
Oh, absolutely, this is good enough. It just took me by surprise. -- Cheers / Saludos, Carlos E. R. (from 15.5 x86_64 at Telcontar)