BobHartung Loc: Bettendorf, IA

Let me lay out my problem:
I am scanning a large map in sections. The scanner is positioned in a void in a large sheet of plywood. The scanner is perfectly square to the plywood edges. I scan with at least 40-50% overlap of all images on both the side and bottom of the scanners field of view. The scan will require 6-7 rows of 6 scans in each row. The first two rows can be scanned, placed into a panorama (1 for each row) and then the complete rows can be merged into a panorama.

The problem begins with row 3 and worsens with each additional row. The merged rows (which by themselves look okay) all begin to tilt and this becomes progressively worse with each additional row.

LrC fails to merge into a Panorama. PS 2023 works for the first two rows on the "Auto" setting but then begins to fail as described.

i'm sure someone out there has tried to scan a large object and then recreate it as a multi row panorama.

And before you ask, yes I opened the images as "Layers in Photoshop" and I did select Edit->Auto Align Images before trying to merge.

Any helpful suggestions appreciated. If you just don't have a clue please do not respond. I'm looking for meaningful discussion not off the wall comments.

lamiaceae Loc: San Luis Obispo County, CA

How large of a Map is this? I have not tried "panorama-ing" more than two frames with my scanner. But I know what you mean about tilting and geometric distortions from the many stitched panoramas I have made with images from my camera and then stitched with Ps CS6. I think you are using Lr and Ps correctly. One thing should be obvious, you must move the map squarely left and right, and up and down as you scan each overlapping segment. But what you may not realize it that the optical field of your scanner is slightly curved like a WA camera lens. I have a high end Epson $1,300 V850 PRO Scanner and the resultant images are not 100$ square, they have a bit of barrel or pincushion distortion. You might have better luck carefully photographing your map with a camera and macro lens on a tripod. That can be done in sections too but they should be more "straight" and squared off correctly. Hope this helps or gives you more ideas. Good luck

BobHartung Loc: Bettendorf, IA

The map is 30"x60". The resultant stitches are huge: at the half way point I was over 1.4 Gig in size. My Mac Studio with the M1 Ultra chip was handling the data load okay if slowly. I may have to resort to the photography approach.

User ID

If I read you correctly, Row-1 represents an edge of the map and each row would progress toward the opposite edge.

Since you use six rows and trouble starts about halfway across, perhaps starting with the central rows and then working outward toward both edges will avoid the seemingly cumulative problem.


BTW, did you originally build each row by stitching from one end to the other end ? Or did you build each row from the center and work outwards toward both ends ?

I cannot fathom how a scanner can have barrel distortion, but if indeed it can, then starting all stitching from the center and working outward may be "least troubled" approach cuz youll be progressing fewer rows from your starting point ... three in each of two directions rather than all six rows in one direction.

lamiaceae Loc: San Luis Obispo County, CA

Actually, when I stitch together camera images to create one huge pano of a landscape I do it all at once, all 6 to 8 images. And, yes 1.4 GB PSD files. Going to PSB helped slightly back then. Final images converted to JPGs.
It works better that way on my current PC with 32 GB of RAM. With my old now long dead laptop with only 4 GB it was a challenge and I often had to do it piece meal.

BobHartung Loc: Bettendorf, IA

Each row is nearly a GB. That's nearly 7 GB in total.

lamiaceae Loc: San Luis Obispo County, CA

Oh, wow, I misread. That is HUGE! What res did you have your scanner set too? I don't usually go higher than what my Cameras could do. Though for High Quality scans of 35mm negatives I have to go rather high. Annoying how scanners show dpi like a printer but we really want ppi like camera res.

BobHartung Loc: Bettendorf, IA

Let's just say "very high" e.g. in excess of 1600 dpi

But I've decided to go with photographing the map in sections. That way lens distortion will be taken care of in LrC before merging. I think I'll try shooting it with my X2D (100 Mix) and the 120 XCD macro lens.

Thanks for your input though.

User ID

So youll be shooting about an 11x14" section per frame ?

larryepage Loc: North Texas area

I realize that you've decided to change your approach, but this sounds like it might be an issue with the choice of Projection (spherical, cylindrical, etc.). I've had problems before when moving my camera transversely instead of pivoting it around a point. That is just like trying to stitch these scans. Might be worth a double-check. And a work-around to try is to build columns first, then assemble across. Sometimes works, sometimes doesn't.

And as already suggested, adjustment to a lower resolution will retain all detail and significantly reduce processor load.

jcboy3

Try using the software Hugin for this.

BobHartung Loc: Bettendorf, IA

To All

It was the method. row by row and then trying to merge into row panos. I looked into the problem of the scanner lens as suggested by lamiaceae and I think that is probably the source. I worked backwards and ended up doing a single vertical pano with the camera at variable heights and perpendicular to the wall (where the map is hanging). There was just a bit of perspective correction but I got it done and the file is 1 GB. Shot with a Hasselblad X2D (100 Mix) and the XCD 120mm Macro lens. It looks good and I only wasted 3 days!

User ID

DirtFarmer Loc: Escaped from the NYC area, back to MA

Late to the discussion and I see you have found a solution but when I have had that problem I did the pano in steps. I made a pano with photoshop using as many elements as possible, then another with different elements, etc. Then I took the sectional panos and ran them through photoshop again and that worked.

BobHartung Loc: Bettendorf, IA

Tried that on a row by row basis and could not make it work. But I got it worked out in the end.