Sorry for spamming you with too many tweets today, but I think this one is worth it. The Solo trailer remade by my AI.pic.twitter.com/r509PBH5Ja
It does not work for me. It seems that my memory of movie scenes puts a heavier focus on semantic similarity than on visual similarity, so I don't recognize the scenes outside of a side-by-side context.
-
-
That is because semantically out of context makes no sense at all. If you want to do it the other way around (context parsing a sequence of images instead images parsing images) literally you are putting out of job every creator and the input can't be just a sequence of pictures
-
This is not a binary distinction. Of course we won't expect Quasimondo's net to extract the full semantic structure from his data, but there might be a level above "similar colors" and "similar pace of movement" and below "people acting on same intentions" that works well enough.
- 6 more replies
New conversation -
-
-
Yeah, I agree that it works better if you see the original and the remake side-by-side. But I also haven't started fiddling with the matching conditions yet.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
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.