A lot of it is a 2D filter applied to 3D models, from the look of them. They're just too perfect in maintaining the coherence of the characters' form that it sinks into the uncanny valley.

I still remember while Naruto was airing, and people loved posting still frames of deformed characters. Yet in animation it actually conveys much more to the viewer than a consistent model will ever be able to.

A lot of it is a 2D filter applied to 3D models, from the look of them. They're just too perfect in maintaining the coherence of the characters' form that it sinks into the uncanny valley.