I suspect its just easier and faster (in terms of cpu and network throughput) to translate human facial expessions to a digital represebtation and apply it to lofi animated character than using something more hi def.