Dynamic Spotlights VR Performance

How bad is the performance hit? I’m not looking to cast shadows - only provide a sense of low visibility and atmosphere. I’ve considered a few alternative options as well - such as some sort of fake fog-darkness with a cutout that gets rotated with the player. Basically the pac-man shape. This could work pretty much only because it’s not only going to rotate on the Z axis and generally won’t move up or down all that much.

That said, I don’t know how well that will perform or look either - transparency has to be avoided or checkerboarding used but also I’d need to dither it to simulate a gradient of sorts.

I’m capable of measuring it myself but I’m sick at the moment and don’t want to put my contacts in and test this with proper performance capture - honestly sitting here in this chair is a little nauseating as is. That, and I’m actually not sure off the top of my head as to how I can understand what a 3GB GTX 1060 is capable of based on frame ms gathered from a 2070. I don’t want to make a game that only high end GPUs can run at non-masochistic frame-rates.

All that I can find online is “Don’t do it” but it doesn’t mention if they’re shadow casting, point lights / skylights, how many, and so on. I’m aware of and comfortable with the typical light faking techniques - emitters, textured cones / godrays, fake / precomputed lighting, and so forth - the methods by which you show that something casts light without actually needing the visibility to be altered. I think that in plenty, if not most cases I can actually do exactly this - make a semi-dark area (still able to see decently) and just have the “light” be a colored cone mesh for effect.

What I’m referring to, for example, would be things like the pitch dark where you want a sense of extreme vulnerability and anxiety about surprise. A reason to slow down. A reason to take a moment to study something before knowing for certain what it is or what it could be.

One final idea that’s coming to mind is using the depth buffer (assuming such a thing exists in forward rendering) to add in shadows based on depth. A sub-idea would be using distance from camera in the material to provide a “visibility lever” for certain objects - not a flashlight by any means, but achieves the end goal which is reduced visibility until the camera is near. Obviously culling distance could be used alongside a black texture somehow layered in to prevent seeing the skybox or whatever.