pub fn flat_styles_rgba8_from_geometry_columns(
geom_ids: &[u32],
geom_colors: &[f32],
resolved: &ResolvedPrepass,
decoder: &mut EntityDecoder<'_>,
) -> (Vec<u32>, Vec<u8>)Expand description
crate::prepass::flat_styles_rgba8 with the (dominant) geometry-style
source supplied as PRE-MERGED columns instead of a map — the sharded
finalize path. geom_ids are unique (the host merged shard results
first-wins) with geom_colors as rgba f32 quads; resolved carries the
support resolution (indexed colours, materials, element colours) and MUST
have an empty geometry_style_index. Output is byte-identical to the
serial flatten: same precedence (geometry > indexed colour > material >
element material), same id-ascending wire order, same rgba8 conversion —
without ever building the 4M-entry geometry hashmap.