ZM vision has historically been a performance killer because of its massive light_dynamic, after some tests I suggest replacing it with a projected texture like the survivors' flashlight.
I've tested a red env_projectedtexture parented to the ZM, wide enough to cover the entire FOV, reached far enough to have similar range to current ZM vision and with shadows off. It had similar performance impact to the survivors' flashlight (AKA runs much better), provided better visibility as the ZM vision light_dynamic tends to be blinding near bright surfaces and also looked better IMO.
On that note, I've also found that, despite what I've read about projected textures being the most expensive light, in my tests even small light_dynamics, if they are set to affect lightmapped brushes, will run much worse than an equivalent env_projectedtexture. Although when the light_dynamic only affects models it's p. much free even if the model has a high polygon density.
ZM vision has historically been a performance killer because of its massive light_dynamic, after some tests I suggest replacing it with a projected texture like the survivors' flashlight.
I've tested a red env_projectedtexture parented to the ZM, wide enough to cover the entire FOV, reached far enough to have similar range to current ZM vision and with shadows off. It had similar performance impact to the survivors' flashlight (AKA runs much better), provided better visibility as the ZM vision light_dynamic tends to be blinding near bright surfaces and also looked better IMO.
On that note, I've also found that, despite what I've read about projected textures being the most expensive light, in my tests even small light_dynamics, if they are set to affect lightmapped brushes, will run much worse than an equivalent env_projectedtexture. Although when the light_dynamic only affects models it's p. much free even if the model has a high polygon density.