Problem is, if you were to render the gun separately at native, then you'd need to add another step in there, upscaling the dynamic render target up to native before the downscale for q-native rendering. This is where you'd render the gun. So you'd take a memory hit, but also by having to render and write out another half million pixels per frame, you'd take a perf hit too. It doesn't sound like much but it adds up really fast. Also, anything behind the gun will be rendered redundantly (which would require quite some work to avoid - further adding overhead). So you'd likely end up in a situation where the extra overhead will negate the advantage of reducing the resolution, so you'd probably end up running slower.