For those afraid of deep fakes

this is interesting - actually, the methodology could be applied to some of our ideas eventually I presume… but in the meantime, let’s dream that the counterwar is on!

1 Like

I guess with such a known quantity: “human face”, being able to tell variations on this. I imagine some of their warp magnitude will be a product of the specific algorithm that that version of Photoshop is using. So adding some noise or variation to that process would probably jam this up some.

But it is an interesting thing.

Also curious to see (hear) how this starts applying to audio stuff (via fluid.flowprediction~ and fluid.bufflowprediction~ of course).

1 Like