In case you were wondering how the new unreleased (and unfinished) DeOldify model compares with the old- here's a few comparisons of best old (left) vs best new (right).pic.twitter.com/VcUG1rOvnv
-
That's absolutely amazing. Where on earth do you get the colour information from? I mean apart from the obvious "from training data" :)
Thanks! I just convert modern images to black and white and mess them up a bit to simulate old/damaged photos. That's your input for training. Then the target for training is just the original modern color photos. These are coming from several different image datasets.
-
-
Training against colour-removed Vs original makes total sense. Nice one. How are flowers looking in the new version?
-
Noticeably better- that was something I specifically addressed with more data.
End of conversation
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.