Last week there was an Ask HN thread about live-rendered video games graphics still looking worse than 20 year old Hollywood CGI. A lot of the comments were about balancing performance and having less power, true, but there was something that wasn't being considered: what we expect CGI to look like.
I couldn't word the comment right so I didn't post it, but I was going to bring up the game Unrecord[1] as an example of fairly realistic looking game – it still has general limitations other live rendered work has, but the choice of using a high field of view to emulated a bodycam (alongside cold, high contrast lighting) sells a certain vision of reality.
I see it the same here. Maybe our idea of what CGI looks like clouds us, and something as simple as using a high FOV / fisheye lens throws us off. Also consider that due to the distortion of the lens, real life fisheye footage can have an almost unreal vibe to it.