Seems like game developers have gotten car racing and general environmental rendering figured out. There has been times where I truly couldn't tell whether game play footage it was real video or rendered. That all ends when characters enter the picture. Anything "living" just seems off, whether it's a human or even a dog.
Why can't video game use actors to model movement and looks and then extrapolate from there? Maybe they do already, it's just not very convincing.
A least some of this is the characters movements have no context. They're making gestures that might look more natural if they were talking to another character but they're standing alone on a bridge.
There's also no realistic gravity between their feet/body and the ground or objects around them. They pop up and down if there's a step and appear to be floating.
Why can't video game use actors to model movement and looks and then extrapolate from there? Maybe they do already, it's just not very convincing.