By using this site, you agree to our Privacy Policy and our Terms of Use. Close
SvennoJ said:
LegitHyperbole said:

I'm a fan of the choices, I particularly like Stellar blades solution where they off a balanced mode between the two and it works for me perfectly. I agree though, RT needs to be ignored until we have hardware that can actually handle it or do something significant with it. It's pitiful of bas ps5 and while better on pro it's still too much of a cost. Lazy devs think they can get away with this instead of working on the games graphics. 

I works in Spiderman 2 though. The default RT options that is. It managed a locked 30fps on base PS5 and now a locked 60fps on PS5 Pro. Just if you go for fidelity mode on Pro it throws in more hardly noticeable RT improvements which without a VRR display create terrible fps dips while swinging around.

That has also become a problem, relying on VRR. It's great for higher frame rates over 60 fps to maintain perceived stability, but it's a bad crutch for games that can't reach 60 fps. Big swings under 60 fps remain very distracting and don't help image reconstruction techniques.


Horizon looks amazing yet there running around water you notice it's still limited by screen space reflections, at the edges the reflections are missing or look weird. You only see it while running along water, not all that distracting. Certainly not like huge frame drops or upscaling artifacts.

I'm sure RT can put more 'realism' in scenes like this

Flood it with shadows and ray traced lighting. Yet for now it's not worth the performance cost. RT is great when you can switch to it and not have to do all the lighting work, shadow and reflection maps. But now it's just an extra, no time saver, more work. And thus not the priority for optimization.

RT might make this look better too, or worse

Running at a locked 60 FPS makes it look more impressive than adding RT reflections at 30 fps, imo.

It looks great without RT, not missing it.


Just like VR, RT is still in its experimental stage. Too early to switch to, can add depth but mostly adds performance woes.

As far as screen space reflections, I'd like to see more cheap solutions which don't rely on RT, whether it be blending them with cubemaps, or some sort of technique that uses info from previous frames to fill in what is being occluded from the current. The latter could be executed with AI maybe at the fraction of the costs of real RT