The human mind has evolved to be able to distinguish things in a split second, the most important being other humans. There is a disorder called prosopagnosia which is the inability to recognize human faces. There's a reason I can recognize a friend from two blocks away simply by the way they walk. Our minds are expecting a lot of exact detail to fit the pattern of a human face, or the pattern of physics, and if something is just slightly off, our mind says "wait, something is wrong. This is unusual" and we react primitively. The closer to photorealism that 3D graphics go, the riskier it is that you will pass the boundary where the mind goes from recognizing artifice and being comfortable and you will enter the realm of representation that fires those parts of the mind that are more about survival than luxury. Our mind will look for the cues that prove this is a real scene, and when they aren't there it ceases to be successful, and hence feels almost uncomfortable rather than uncanny. I've have always felt that 2D hand drawn cel animation had more "weight" to the characters because I wasn't expecting them to behave under real physical laws. For me, simply rotoscoping a human being in 2D animation removes all weight from the character and it isn't believable as something from the mind of Tex Avery. That's not to say that there hasn't been progress. Either through me getting accustomed to it or through advances in the physics rendering, I've noticed that quite a bit of 3d animation has developed the nuance, or at least illusion of nuance, that goes to convincing the mind that it is "real".