Problem
GI is realtime-only (SSGI + IBL). There's no baked lightmap / irradiance-volume path for static geometry. It looks fine, but you lose a major perf lever and ground-truth bounce lighting for static-heavy scenes.
Why it matters (Tier 3)
A baked path gives high-quality, cheap indirect lighting for the static world (terrain, buildings, props), freeing the realtime budget for dynamic objects. Standard for shipping open-world titles.
Proposed approach (compute/offline)
- Offline lightmap bake (UV2 unwrap + path-traced or irradiance bake — can run in compute) for static meshes; runtime samples the lightmap.
- And/or baked irradiance volumes (probe grid) for dynamic objects sitting in static lighting.
- The bake step lives in the asset-cook pipeline (
AssetPackBuilder / BuildGamePanel).
Acceptance criteria
- A static scene shows baked indirect bounce; dynamic objects sample baked probes; runtime cost lower than full SSGI for the same look.
Tier 3 — visual quality / perf lever.
Score
capability: 5
craft: 13
stability: 1
decay: 1
effort: 8
confidence: 0.5
learning: 5
fun: 5
kano: delighter
blocked_by: []
blocks: []
Rated per issue-scoring · score = confidence × (capability + craft + stability + decay) / effort, derived by the picker.
Problem
GI is realtime-only (SSGI + IBL). There's no baked lightmap / irradiance-volume path for static geometry. It looks fine, but you lose a major perf lever and ground-truth bounce lighting for static-heavy scenes.
Why it matters (Tier 3)
A baked path gives high-quality, cheap indirect lighting for the static world (terrain, buildings, props), freeing the realtime budget for dynamic objects. Standard for shipping open-world titles.
Proposed approach (compute/offline)
AssetPackBuilder/BuildGamePanel).Acceptance criteria
Tier 3 — visual quality / perf lever.
Score
Rated per issue-scoring · score = confidence × (capability + craft + stability + decay) / effort, derived by the picker.