![]() ![]() Bruit force steering away from those responses is kind of cheating the uncanny valley. There is a part of the brain located under the boundaries of the frontal and temporal lobes, that processes gustatory responses. You see part of our perception of beauty is the lack of contrasting information. That being said, I’m not sure that they are just a few tweaks away from nailing it. It’s high side of the uncanny valley because they are unrealistically attractive. There should also be randomly distributed flaws as well to create a more realistic crowd. It’s probably fixable with a final pass of mild randomization though. There is just too much symmetry in the individual examples. These are things that these AI axioms aren’t accounting for. For instance Cindy Crawford’s mole or Brook Shields’s eyebrows. Small flaws are often seen as attractive. We actually see the norm as attractive for the most part. Many followups of this study suggested that attractiveness was more or less based upon an average of human characteristics. The interesting result is that all of the attempts resulted in really attractive images. ![]() ![]() This sounds like BS but it’s part of Robert Sapolsky’s lectures on the Stanford You Tube page. They took the likenesses of a large number of criminals to average them into the “face of moral depravity”. ![]() Much later there was a more scientific study with the intention of testing the notion that came across an interesting result. There was a funny story about someone judging Neitzsche’s character by his homely face to which he replied “You know me?”. Back since before the time of Neitzsche it was thought that unattractive people were morally depraved. The reason that the faces are all attractive is probably because of something discovered in an interesting study concerning criminal profiling. Regarding the neural network generated faces: ![]()
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |