I’m trying to gain a few more ms by rendering at lower resolution the background and rezing down dynamically when the camera rotates rapidly.
I’d like to keep the foreground crisp though.
Back on the days of built in you could stack cameras but that required double the transform and double the culling and with many object that cost starts biting on mobile.
Is it better in URP? Any trick?
When you say multi-camera, are you using camera stacking feature in URP?
That was an example of how I would do it in built-in to drop resolution at a distance.
Can you clarify foreground and background? Foreground=UI and Background=3D Scene or do you mean something else?
Have you played with the ScalableBufferManager already?
3D scene both. Foreground = first 20 units and background the rest. They do that on Mario Odissey to keep fillrate within budget.
buffermanager I’ll do it later, for now I’m single camera and use urp.renderscale until I figure out the hardware. The bottleneck keeps changing. It was the cpu at 23ms and now it’s the fillrate. When I slide down renderscale to 0.5 the framerate bumps back up to 60, the target. Not sure where that 23ms cpu went.
I dug into that ScalableBufferManager, it doesn’t scale per camera
render scale is per pipeline, not per renderer (!) so… how can I scale down rendering on one camera?