There is a way to make something the original in terms of an NFT. Let's take an "original manuscript". What you, the writer, do is to print it out preferably on a not perfect printer, then scan it back in so as to accrue a lot of random 'edge' pixel values. Then destroy the paper original, NFT the scan data. Let this be the original. Now print out the scan, and scan that. Preferably this should be on a different printer and scanner. That is a copy that you post and show. But it will never get the same hash, and searching for the right hash will take conservatively, for a big enough manuscript, a zillion years. You have an original. Every copy then not only is provably a copy, but its provenance off the original can be, generally speaking, computed by the random digitization drift which can be estimated from the drift from the original to the first copy which only you know but can be more coarsely estimated by successive copy differences.

What I have never understood is why it is necessary to use a block chain for this stuff. A simple hash tree with no cost, as with ISO/ANSI X.509v3 certificates, gives the same functional assurance and has used hash trees a lot longer. All interbank transfers of money, trillions of dollars a day, use these hash trees. I don't see these NFT's dying anywhere. Just the stupid block chain ones with the proof of work (i.e., costing a huge electrical bill for guessing a random number). You could argue for decentralized $$ the proof of work is OK to limit supply of coin, but even that seems unnecessary. E.g., if you only want 2 bllion coins, make them all without proof of work and let people buy and sell them for whatever they can get knowing there can only be two billion. Same trick as above including destroying the root signing key (the original) after you have two billion.

Just say'n.