%% how one 2x2 quad of pixel-shader threads emulates coarse pixels on an M1 (Metal has no shading rate).
flowchart TD
    A["Quad of 4 pixel threads<br/>(Metal runs every pixel)"] --> B["Rate: base, SV_ShadingRate,<br/>16x16 image tile,<br/>combiners and clamps"]
    B --> C{"Rate<br/>1x1?"}
    C -- yes --> K["Shade normally"]
    C -- no --> D["Pick a 2x2 block<br/>of coarse pixels whose<br/>owners are covered"]
    D --> E{"Block<br/>found?"}
    E -- no --> K
    E -- yes --> F["Each thread shades one<br/>coarse pixel at its centre<br/>(interpolate_at_offset)"]
    F --> G["Coarse-sized<br/>ddx / ddy"]
    G --> H["quad_shuffle outputs<br/>and discard from the owner"]
    H --> I["Non-owners become<br/>helpers: no UAV stores"]