Gig works wonders on old greyscale ship photos and cityscapes, but it tends to mangle text, such as a ship’s nameplate or signs on buildings. Could Gig possibly be taught to recognize letterforms automatically within an image — and then treat them with a modified algorithm that yields clear, or at least, less mangled text?
I work on historical photos where preserving text within the image is as important as the image!
Thanks!
