It didn´t have to be this dumb. I mean, people talking about TLoU which has been out for years, still surprised by how its GI works. Yes its only for the flashlight, and yes, only on areas where you are expected to use it. It sure uses some precalculated (or hand-made) simplified representation of the scene, and it is very low frequency. All indicates some form of VPL solution, its been talked about to exaution already, if you were unaware, you are not qualified to give your shallow opinon. Sorry. QB had a presentation on siggraph about its rendering tech, and its far from perfect, and full of limitations that DO show in all videos I´ve seen. I love the final result, but its still full of artifacts. Voxel dynamic GI? That´s not in there, all the voxel GI is static, and all the dynamic GI is screenspace. That door knob pic looks amazing, but I´m guessing its more of an artifact that ended up looking right out of luck, than the top of the door knob actually bouncing that much light into the door. There´s simpy not enough coverage for that to be the case, specially at half res. Its AO is also very interesting. I had never seen other use of this Line-sweep AO, and I still don´t know exaclty how it works ( the paper referenced on the siggraph pres is behind a paywall) but it sure has a very diferent behaviour and set of artifacts at faliure cases, and so is their SSR.
There is a lot of stuff to be said about their solutions, even if we haven´t seen the final build yet (and event that would still be up to updates at this day and age) but people should at least know a minimum about what they are talking about if they wanna spill their views on it.
I might be sounding a little grumpy now, and its because I am. Sorry.