people say an ai portrait photo lacks soul but rarely contend w their admittance that they are sensing ‘animation’ from the pixel sorting of a real portrait photo

i think it’s still the final 1% of uncanny valley. people can still (only just) discern that it’s not “real”. So I think it’s largely because the bar specifically for a portrait of a person is probably the highest of any other context. 

I think there are many examples of AI images that have soul and illicit a response. Even abstract/intentionally surreal ones featuring people. So it’s just a matter of closing that final 1% gap, which seems inevitable at this point.