Lighting, GI and time of day

San Van’s time-of-day lighting combines VisualEnvironment bundles with local lights authored in lots. The 21 quality/ToD bundles (three tiers × seven ToDs) contain sun, sky, exposure and GI configuration, but no local lights. Local lights are overridden in place. Observed

GI and exposure

gibs_on selects GiRuntimeConfiguration.FallbackGi=1, ray-traced surfel GI; gibs_off selects 3, GPU probes. The fallback enum is None/Gibs/CpuProbes/GpuProbes/Null. Enlighten.Enable remains true in both tier settings. San Van uses Enlighten probes and a GI grid; the inspected meshes have no lightmap binding. Observed

Per-object GibsModeOverride values None 0 and Auto 1 do not provide an object-level RT-GI disable. Studio meshes carry Auto and EnlightenType 0, as do inspected stock props. Ray-traced GI is switched off globally instead, through Render.RaytraceEnable, GiRuntimeConfiguration.FallbackGi or the tier subsystem DingoTierManager.Subsystem gibs_off. Observed

High-tier environment	Minimum EV	Maximum EV	Confidence
Night	5.7	7.4	Observed
Morning	6.5	10.0	Observed
Noon	7.0	11.3	Observed
Afternoon	7.0	11.4	Observed
Evening	7.0	10.0	Observed
Level default	7.5	11.5	Observed

These VisualEnvironments enable automatic exposure and store nominal EV 12; the five ToD clamp intervals overlap only at EV 7.0–7.4. Every inspected ToD VisualEnvironment has ShaderParams NightAmount 0. Observed

Exposure and emission have console settings: PostProcess.ForceEVEnable, PostProcess.ForceEV, ForceEVCompensation and its enable, EVClampEnable, DrawExposureDebugInfo, and DingoShaderSettings.EmissiveMult and NightTodActive. PostProcess.ForceEVEnable 1 with PostProcess.ForceEV 9 locks exposure for measurement. Forcing an extreme EV (±1,000,000) turns the screen black for the rest of the session: neither disabling the force nor setting an ordinary value restores the picture. Observed

Emission and shader availability

Studio 20261012’s emission_texture, emission_color, emission_strength and emission_srgb fields clone the stock LED donor, program 1138. The compiled material writes RadiosityMaterial.EmissiveIntensity = 683 * emission_strength, and LED section vertices have Color0 zero. There is no ToD condition on those emission fields. Observed

Stock content barely uses EmissiveIntensity: of the 161,109 RadiosityMaterial objects in the EBX documents of every superbundle TOC, 161,085 hold 0. The other 24 are all in bam_levelroot and the dingolevel_sdm interiors: Glass_Opaque_01_2x2m_WSDark_M (200 and 1,200), Glass_Stained_01_2x2m_M_Interior (250 to 45,000 over several variants), lambert4 (30,000 and 45,000), Signage_Electronic_Wave_01_1x1_M (10,000) and DEV_ProxyGrid (1,500). Observed Stock emissive lamps, screens and signs therefore get their glow some other way, through their programs’ own emission and the blended passes described under decals. Inferred

Studio’s available map domains are surface, decal, foliage, ocean and glass_opaque. No unlit world program is identified in the inspected stock material/program set, and ReSkate’s merge retains base programs. Black base colour plus emission is an available material arrangement; a general unlit map-shader route is not established. Observed

Studio’s GI-grid bake is forcibly disabled inside compile-map in build 20261012: the compiler clears the GI option before baking, even when requested. --no-gi only changes the notice. Its separate --enlighten option does produce EnlightenDataAsset/DynamicEnlightenEntityData and a database; a measured Warehouse map’s database is 31.6 MB. Presence of a database alone does not establish useful interior illumination. Observed

Every inspected Studio map patches bam_levelroot/lighting_global without its DynamicEnlightenEntityData. Whether that removes useful stock-city Enlighten lighting while the mod is installed is not measured. Observed Inferred

Local lights

Only three light classes are used anywhere in the measured data: PbrSpot, PbrSphere and PbrRectangular. No PbrTube or Old* light is placed. Observed

San Van	Count	Measured values	Confidence
PbrSpot	20,686	OuterAngle in full degrees, median 125 (p10 90, p90 140); InnerAngle median 35; median 85,000 lm; attenuation radius median 13 m (p90 27); ReflectorEnable on 98%	Observed
PbrSphere	4,175		Observed
PbrRectangular	43	OuterAngle 40	Observed

The skate-park lots use spots of 85°/45° at 225,000 lm and 27 m, and of 135°/50° at 200,000 lm and 18 m, with spheres of 75,000 lm; the cmn_lgt_floodlight_* fixtures are 95–125°/20–45°, 80,000–130,000 lm and 10–16 m. San Van ties a fixture’s glow to its light through DingoLightComponentData (NightEnabled, Profile, MaterialData). Observed

A light entity’s fields include Intensity in lumens, AttenuationRadius, DiscRadius or SphereRadius, OuterAngle, InnerAngle and ReflectorEnable. Observed A spot with ReflectorEnable concentrates its lumens into its cone: with equal 162,813-lm spots 6 m up and inner angle 0, at locked exposure, the floor beside the pool centre is more than 15.6 times brighter under a 40° cone than under a 180° one, and about 3.9 times under a 120° one. The cone is therefore not a mask over a fixed distribution. Observed

Light retirement

An outgoing ToD subworld reports fully unloaded within one frame: its SubWorldReference handle at +0x80 is zero, the loaded bit +0x148 & 1 is clear and the instance at +0xb0 is zero. Its light ids return to their allocator 130–550 ms (8–32 frames) later, together in one frame, through the renderer’s per-frame pending-removal list (Skate+0x4ec3800 to +0x4ecdda0). Loading the next layer immediately after the subworld flag clears can therefore overlap both layers’ light allocations. Observed On a city of about 7,500 spot lights per ToD, a Noon-to-Afternoon switch at level load lost over 7,000 of the incoming layer’s spots in 3 of 5 loads. Observed

ReSkate’s unload-before-load barrier and per-class light guard address different parts of this pressure. Sourced · ReSkate See limits.

What emission strength remains visually stable across ToDs?

The exposure intervals suggest a narrow common range, but no calibrated emission ramp across all ToDs establishes a universal strength-to-appearance rule. Unknown