I am trying to fix some really bad scanner work in some gaming cards that were done years ago. I would like to know which is the best way to adjust the brightness, levels, colors, etc, to be able to achieve (almost) the same level of brightness and sharpness to the card on the left as the card on the right:

lost brightness

As you can see, the scanner or whatever the way the left image was taken of is radically different from the right image. I would like to correct this brightness levels to match the white color of column in the background, as now it is almost grey-blueish.

This is another example:

lost brightness

Again, as you can see, the background image looks too damaged, almost black. Numbers on the image are as well too dark to be read easily.

Last example:

lost brightness

Foreground and background images are almost dark, and the text is almost unreadable.

Is there a way to properly recover from this quality loss? And, as I need to perform this fix for a lot (hundreds) of cards, is there a way to perform it in batches, instead of manually, with an specific software? I've tried with Gimp, and a couple of batch photo editors on Mac, but the results are not so great, or maybe I am using the wrong tools or settings.

Thanks in advance.


1 Answer 1


It would be possible to adjust these using levels adjustments, however the brightness and contrast seems to vary in different parts of these images, so this would involve making different selections and adjusting the levels separately. And because of this, I don't think you could automate this. There's too much variation and a single levels adjustment just won't work for this.

Here's an attempt I made on one of your images. I made the masks manually, and used them to apply different adjustments selectively to different regions of the image. I did this in Photoshop. Even after using five different adjustment layers (4 levels + 1 hue-saturation), it still isn't perfect.

enter image description hereClick to see larger

You could do something similar in GIMP, but it doesn't have adjustment layers. Instead you could make different selections, and copy and paste these pieces to separate layers, and make the adjustments to each layer as required.

To be perfectly honest, although I don't know how practical this would be for you in this particular case, but it might be better to just scan them again. There is nothing worse than trying to fix crappy scanned images. The amount of time spent fixing them will be much more than simply rescanning them.

Disclaimer: I note that these game cards are copyright, so I can't really condone copying these. This answer is for educational purposes only.

  • Thanks for taking the time to try. I wanted to at least know if this was technically possible, as it seems like we should be recovering long lost pixels on the way. Even if it is tedious, if I find a way to resolve it using Gimp I may be able to fix them eventually. Maybe there is some online AI that offers this kind of recovery.
    – fergardi
    Commented Apr 8 at 15:28
  • @fergardi Artificial Intelligence is certainly a possibility, but then with AI you are kind of limited to what the software "thinks" you want. Maybe with the right prompts. Sure. Give it a go. If you find something useful, you could even add it as an answer here. I've only dabbled with AI applications, no expert, so I wish I could give you more help with that.
    – Billy Kerr
    Commented Apr 8 at 16:03
  • I guess other option is just manually copying the "good" background card frame, and paste on top of it the "bad" title, text and image. It will be even more tedious to do this per card, but may lead to better results in the end.
    – fergardi
    Commented Apr 10 at 9:10
  • @fergardi - yes, this is more or less the same as I did with Photoshop. I used layer adjustments with masks instead, but in GIMP you could make selections and copy and paste the pieces onto separate layers and adjust each of them manually. I mentioned this in my answer already.
    – Billy Kerr
    Commented Apr 10 at 10:19

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.