3
\$\begingroup\$

I have some photos where I have processed the image to my liking but unfortunately only saved a low resolution (for instagram) version. I now would like a higher resolution version for printing. I have the original raw and would like to try and match my previous image as closely as possible. Is there any way to do this automatically?

The edits will have included cropping, possible perspective correction, exposure, shadow and highlight, colour etc.

I have access to the full Adobe suite, Gimp, Darktable, Photivo, and willing to try any other software or even some simple scripting with pyimage or similar.

Part of me thinks this is probably impossible but then it should be just some slightly complex maths to map one to the other.

Any suggestions?

\$\endgroup\$
2
  • \$\begingroup\$ What application did you use to process initially? \$\endgroup\$
    – Michael C
    Oct 11, 2019 at 0:58
  • \$\begingroup\$ @MichaelC - I did some with Darktable which should be recoverable as I think it saves the settings in a sidecar file but several were done on Photoshop Express on my phone while on the road. I think I possibly could have saved the settings but didn't. \$\endgroup\$
    – neil
    Oct 12, 2019 at 13:28

1 Answer 1

2
\$\begingroup\$

Sorry, but pragmatically the answer is no. There is no good way to go from a processed file back to the process used to get there. One can guess, but it's a very hard problem for a computer — it's just too complicated with too many possibilities.

(Of course, that's exactly the kind of thing machine learning is good at untangling, but I can pretty much guarantee that no one has trained a model your particular situation.)

\$\endgroup\$
1
  • \$\begingroup\$ I thought that was probably the answer. Dealing with crop/rotation/perspective corrections should be possible with something like hugin to match the features. The other corrections may be more challenging but I have some ideas that may work - or may fail dismally. I'll report back if I find a method. \$\endgroup\$
    – neil
    Oct 12, 2019 at 13:32

Your Answer

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

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