5bbcf646a7
# Objective Two more optimisations for UI extraction: * We only need to query for the camera's render entity when the target camera changes. If the target camera is the same as for the previous UI node we can use the previous render entity. * The cheap checks for visibility and zero size should be performed first before the camera queries. ## Solution Add a new system param `UiCameraMap` that resolves the correct render camera entity and only queries when necessary. <img width="506" alt="tracee" src="https://github.com/user-attachments/assets/f57d1e0d-f3a7-49ee-8287-4f01ffc8ba24" /> I don't like the `UiCameraMap` + `UiCameraMapper` implementation very much, maybe someone else can suggest a better construction. This is partly motivated by #16942 which adds further indirection and these changes would ameliorate that performance regression. |
||
|---|---|---|
| .. | ||
| src | ||
| Cargo.toml | ||
| README.md | ||
Bevy UI
