How does Hi-Z Occlusion Culling work?
In occlusion culling (in general), the most important idea is to never cull visible objects. After this, the second-most important idea is to cull fast. GPUI's culling algorithm makes the camera generate a depth texture and use this to make culling decisions in the compute shaders. There are various advantages of this (e.g. you don't need to bake occlusion maps, can use culling with dynamic occluders, etc.) and it is extremely fast since all the operations are executed in the GPU. However, the culling accuracy is ultimately limited by the precision of the depth buffer. On this point, GPUI analyzes the depth texture, and decides how accurate it can be without compromising performance and culling actually visible instances.
As you can see in the following images, the depth texture is a grayscale representation of the camera view where white is close to the camera and black is away.
As the distance between the occluder and the instances become shorter with respect to the distance from the camera, their depth representation become closer to the same color:
In short, given the precision of the depth buffer, GPUI makes the best choice to cull instances for better performance - but also without any chance to cull any visible objects.