combining scans

[Espermaschine]

I have scanned an old recordcover with my DIN-A4 scanner.
The cover is bigger than the scanner so i had to scan it in two parts.

Now i try to make a composite of it, but the images are not exactly the same colour/value.
I tried eyeing it with the Curves or Levels Tool but i cant get it right.

Is it just a matter of value, or do i need to adjust all the RGB channels ?
My guess is, the differences come form the scanners light, so that seems to tell me its just the value that needs tweaking.


I also tried the method with sample points, but i can never get it right....

Any advice, or is there a way to automate this ?

EDIT:
added a link to the xcf

http://www.mediafire.com/file/v18vvsmy5p...can.xcf.gz

[Ofnuts]

The method works, but... there is a "grain" to the cover so you would have to pick the color on each layer for the exact same pixel (which is something that may not even exist if the scans are shifted by a non-integer number of pixels . You can instead forego the sample points, and use the color picker with some "sample average" radius to sample the colors in both layers. And then when you are very happy with the bottom of the cover, you find that this has made the top overlap a lot more obvious. I will have to figure out a workable method (but it's late here, so come back tomorrow).

[Espermaschine]

I noticed the paper gets slightly warped, while its under the lid being scanned.
Guess thats why i have the colour variations.

The various scans dont line up perfectly because of that (i tried in Difference Mode) too.

(02-28-2018, 01:42 AM)Ofnuts Wrote: The method works, but... there is a "grain" to the cover

Not sure what you mean with grain. The background has a speckled paper texure.
The lines are actually not on the original. Seems like they are some kind of artifact.

[rich2005]

You can go overboard with detail.

That is a good match between the scans, apart from the small areas bottom and top. Center looks ok to me.

A little cloning bottom layer to top layer registered mode and fuzzy brush merges the join.

   


Otherwise an application such as Hugin which incorporates colour matching.

[Espermaschine]

(02-28-2018, 09:16 AM)rich2005 Wrote: That is a good match between the scans, apart from the small areas bottom and top. Center looks ok to me.

I agree with you on this version, however i have scanned another, green, cover, which has more extreme variations, that wont go away with cloning seams.
And im also interested in the basic technique of adjusting two images to have the exact same colour.

Will have a look at Hugin as well. Thanks for the tip !

[Ofnuts]

The quick and efficient solution is to create a progressive blend on the two parts. Of course the colors at the left and right ends aren't exactly the same....

   

   

[Espermaschine]

(02-28-2018, 10:16 AM)Ofnuts Wrote: Of course the colors at the left and right ends aren't exactly the same....

But that is the whole problem im having


On a side note: i was able to remove the banding by using G'MIC smooth [bilateral] on the green channel.