Loading…
3
Why does nobody talk about how fast AI image cleanup got good between 2021 and now?
I was going through my old drive last weekend and found shots from a trip to Austin in 2021. I ran one through a cleanup tool back then and it took about 40 seconds and left weird smeared spots on the leaves. I tried the same file on a new tool yesterday and it fixed the noise in maybe 3 seconds with almost no mush. So did the models just get way better at training, or did the newer chips and bigger data sets do most of the work here? One side says it is all about the model design and the other says raw compute is what really moved the needle. Either way the gap between 2021 and now is not small, it is like a different product. Where do you all land on which part deserves the credit?
0 comments
Log in to join the discussion
Log In0 Comments
No comments yet
Be the first to share your thoughts on this discussion.