Skip to content
Engine Wiki

Time-of-day presets and light units

ForzaTechLighting and skyApplies toFH6

Each track has a time-of-day file in media\TimeOfDay\ (the Tokyo track Brio has TimeOfDay_Brio.ppr; TimeOfDay.ppr is the same kind of file for the older tracks) that lists its sky presets. A preset is a group of kind Sky, named by its label; it holds the sky and cloud files, the sun’s illuminance, the exposure settings and the scales that put authored brightness into the lighting units. Other groups select a preset for a stretch of the day, bind weather, fog and zones to presets, and hold the file’s globals. The files are plain (not encrypted) and read without keys. Observed

The file

Section titled “The file”

A .ppr file is a table of names followed by a flat list of groups, each a list of typed leaves. All values are little-endian. Proven · the six .ppr files of the install, TimeOfDay_Brio.ppr, TimeOfDay.ppr and Homespace.ppr in media\TimeOfDay and one each in media\DeformableTerrain, media\Water and media\Weather, each read to its last byte by a reader that refuses any value or length the layout below does not hold; none of the 9,076 zip archives of the install holds a .ppr entry

Part Layout Confidence
Header u32 version (3 in TimeOfDay_Brio.ppr, DeformableTerrain.ppr and WeatherManager.ppr; 2 in TimeOfDay.ppr, Homespace.ppr and Water.ppr), u32 name count Proven
Names per name a u8 length and that many printable ASCII bytes (2,954 names in TimeOfDay_Brio.ppr, 1,962 in TimeOfDay.ppr); three files have one empty name, which is the value of an empty string Proven
Root u16 name index (0 in every file: the file’s own name, TimeOfDay_Brio), u32 group count (2,350 in TimeOfDay_Brio.ppr, 1,337 in TimeOfDay.ppr, none in Homespace.ppr) Proven
Group u16 kind (a name), f32, f32, u16 label (a name), u32 leaf count, then that many leaves; groups do not nest Proven
Leaf u8 type, u16 name, u16 id, then the payload of the type Proven

The first group of TimeOfDay_Brio.ppr and of TimeOfDay.ppr is the globals, kind SkyGlobals (48 and 32 leaves). Proven A leaf’s name and id are table entries: the id is a GUID string (all 54,109 leaves of TimeOfDay_Brio.ppr), one for each pair of a group kind and a property name (452 pairs), so the same property in two kinds of group can have two ids. A name’s first use is followed in the table by its id (307 of 307 names; 50,966 of the 54,109 leaves have the id right after their name). Proven · TimeOfDay_Brio.ppr

Type Payload Confidence
0 i32, 0 or 1 (5,609 leaves of TimeOfDay_Brio.ppr) Proven
1 i32 Proven
2 f32 Proven
3 u16 name index, a string value Proven
4 f32 × 2 (21 leaves, each of a different property: ranges and the exposure boxes of the cameras) Proven
5 a scalar curve Proven
6 a colour curve Proven
7 f32 × 4 (one leaf, GroundPlaneColour) Proven

A curve (types 5 and 6) is a u32 component count and per component a u8 flag, a u32 key count and the keys as f32. A key starts with its time, then: for type 5 with flag 0 one value (8 bytes a key); for type 6 with flag 0 four values, red, green, blue and a fourth that is 0 in every key (20 bytes); for type 5 with flag 1 four values in version 2 files and five in version 3 files (20 and 24 bytes), whose order is not established. No other flag occurs. Proven · the six files; flag 1 occurs in 11 components of TimeOfDay.ppr and 8 of TimeOfDay_Brio.ppr, all on ExposureRange and Grading.Scales Most curves have one component; of the type 5 leaves of TimeOfDay_Brio.ppr, 34,125 have one, 732 two (the three ExposureRange properties), 277 four (Clouds.SunShadowChannelSelection) and 210 five (PostProcess.Grading.Scales); the 2,054 type 6 leaves have one. Proven · TimeOfDay_Brio.ppr Curve times are 0 at the first key of all 39,801 components of TimeOfDay_Brio.ppr and 1 at the last key of 39,645; the other 156 end between 0.9906 and 1, and one component has a key earlier than its predecessor. Proven · TimeOfDay_Brio.ppr

TimeOfDay_Brio.ppr has 2,350 groups of 22 kinds: SkyPreset 1,113, FogGlobalMaterial 270, Sky 203, SkyZoneBinding 162, FogGlobalPreset 127, FogGlobalShaping 101, SkyWeatherBinding 75, SkyTransition 69, SkyZoneFog 61, SkyWeatherPrecipitation1 to 3 25 each, SkyOverrideFog 24, SkySeasonBase and SkySeasonBinding 20 each, LocalFogWeatherBinding 12, AllowedScenarios 5, SkyZoneTunnel 4, SkyOverrideGrade 4, SkyZoneCity 3, SkyGlobals 1 and TransitionPropertySet 1. Proven · the file

  • The 203 Sky groups (the presets) each have 199 leaves, the same properties in the same order. A Sky’s label is its name (190 distinct: 13 labels are borne by two skies). The label Forte_Summer_01 is borne by one. Its leaf Files.NameOverride names the sky files, empty for Forte_Summer_01 itself, and Forte_Summer_01 for 12 other presets (11 named House_*_Forte_Summer_01, and InitialExperience_FreeRoam), which reuse its sky files with other scales. Proven · TimeOfDay_Brio.ppr
  • A SkyPreset group (1,113) has five leaves: SkyName, ForcedGrade, ForcedVolumetricFogPreset, ForcedVolumetricFogGlobalPreset and MoveToWholeFrame. Its label is the preset’s own name (BP_Summer01_Dawn_01) and its SkyName names a Sky label in 723 of the 1,113. Twenty-three name Forte_Summer_01, with group floats between 213.7 and 1,160; the one labelled Forte_Summer_01 has 720 and 780. Proven · TimeOfDay_Brio.ppr
  • A group’s two floats are stored in every group and mean something this page does not establish. SkyGlobals has 0 and 1,440 in TimeOfDay_Brio.ppr; the Sky labelled Forte_Summer_01 has 181.86 and 1,181.46; the 11 House_* presets of Forte_Summer_01’s files have 181.2 and 1,180.8. Unknown · stored values; the 1,440 is the minutes of a day

Physical light units

Section titled “Physical light units”

Over the 203 presets of TimeOfDay_Brio.ppr: Observed

Field Values
PhysicalLightUnits.Enabled 1 in 202, 0 in 1 (the one with 0 has both scales 1)
PhysicalLightUnits.LuminanceToLightBufferInvScale 6600 in 202, 1 in the one with Enabled 0
PhysicalLightUnits.SkyToLuminanceScale 6600 in 157; 13,200 in 14; 52,800 in 12; 26,400 and 25,000 in 6 each; 40,000 in 2; 1, 3,200, 32,000, 42,000, 45,000 and 62,000 in 1 each

The names read as: a sky texel is multiplied by SkyToLuminanceScale to give a luminance, and a luminance is divided by LuminanceToLightBufferInvScale to give the value held in the light buffer. Inferred[the field names; no constant buffer of the running game was read]

Sun and exposure

Section titled “Sun and exposure”
  • Sun.LightingIlluminanceLux is a curve in every preset. Its peak over the 203 presets is 0 in 48, from 0.05 to 0.32 in 33, from 1 to 300 in 26 (the night, studio and garage presets), and between 6,000 and 200,000 lux in the other 96: 72,000 (rounded to the thousand) in 22, 70,000 in 13, 80,000 to 82,000 in 12, 120,000 in 4, 200,000 in 1. Observed
  • PostProcess.AdaptiveExposure.ExposureBias is a curve of one component and .ExposureRange one of two components in every preset. In Forte_Summer_01 the bias is 0 at 0 and 0.063, −0.5 at 0.103, −0.35 at 0.18, 0 at 0.241, −0.25 from 0.284 to 0.291, 0 at 0.299 and 0.361, −0.7 at 0.379, −1 at 0.399 and 0 from 0.414 on; the first component of the range is −10 at 0, 2 at 0.077, 5 at 0.119, 0 from 0.18 to 0.899, 1.39 at 0.93, 2.5 at 0.96 and −3 at 1, the second 11 at 0, 14 at 0.08, 15.5 at 0.113, 20 at 0.4, 19.87 at 0.798, 18 at 0.899, 16 at 0.94, 12 at 0.96 and 10 at 1. The 11 House_* presets that reuse its files have a bias that is 0 except −0.2 at 0.376, −0.25 at 0.399 and 0 at 0.404, and the same range. Observed
  • EmissiveIntensityVsExposure (a curve, 1 at both of its two keys) and EmissiveIntensityExposureRange (−25 to 25) occur once in TimeOfDay_Brio.ppr, in the SkyGlobals group and in none of the presets: the emissive intensity does not vary with exposure there. Observed

Emissive scales

Section titled “Emissive scales”

Each preset has the scalars below as curves (constant ones have two keys). Counts are over the 203 presets of TimeOfDay_Brio.ppr. Observed

Field Values
EmissiveScales.TrackEmissiveEV constant 1 in 71; constant 6,600 in 50; a curve 1 to 50 in 30; constant 100 in 25; constant 50 in 8; 35, 1,200, 50,000 and 5 in 2 each; and 11 others once (3.5 to 30,000)
EmissiveScales.CarEmissiveEV constant 3,200 in 131; constant 155 in 17; a curve 825 to 38,720 in 17; 22,000 in 7; others
EmissiveScales.ForwardPlusEV constant 0 in 159; a curve from 0 up to 4.5 or 5 (night) in 23; others
EmissiveScales.LightGlowEV constant 0 in 196
EmissiveScales.VFXEmissiveEV constant 1 in 172; a curve from 1 up to 50 to 100 in 14
EmissiveScales.Cubemap constant 1 in 199
GlobalIllumination.GIEmissiveScale 1 in 139, 2 in 52, a curve 1 to 6.944 in 9, 10 in 2, 0 in 1

By the sun’s peak illuminance: of the 107 presets below 500 lux, TrackEmissiveEV is constant 1 in 55 and constant 6,600 in 43 (GIEmissiveScale 2 in 42 of them and CarEmissiveEV 3,200 in 42: 41 unnamed presets, Debug_Grey and Forte_Autoshow_01) and something else in 9; of the 96 presets above 500 lux it is a curve 1 to 50 in 30 (Forte_Summer_01 11, Forte_Spring_01 5, Forte_Overcast_01 4, Forte_Summer_02 3, Forte_Autumn_01 1 and 6 unnamed), constant 100 in 25 (BeautifulArea_Midday at 15,000 to 200,000 lux), constant 1 in 16, constant 50 in 8, constant 6,600 in 7 (sun 40,000 to 81,000 lux, GIEmissiveScale 2, SkyToLuminanceScale 6,600) and others in 10. The names are those of Files.NameOverride, the sky files a preset uses. Observed The preset labelled Forte_Summer_01 has TrackEmissiveEV 1 at 0, 3 at 0.029, 50 from 0.149 to 0.899, 3 at 0.989 and 1 at 1, so its peak is 50; CarEmissiveEV 225 at 0, 300 at 0.028, 825 at 0.06, 1,000 at 0.073, 10,000 at 0.108, 22,000 at 0.12, 38,720 at 0.54, 16,000 at 0.9 and 825 at 0.96; ForwardPlusEV 0, 4.5 at 0.057, 0 from 0.126 to 0.926, 3 at 0.946, 1.85 at 0.97 and 0 at 1; SkyToLuminanceScale 13,200; and a sun of 63,000 to 65,000 lux from 0.48 to 0.599, peaking at 65,000 at 0.54. The 11 House_*_Forte_Summer_01 presets that reuse its files have TrackEmissiveEV 1 at 0 and 0.059, 50 at 0.359 and 0.719, 1 at 0.959 and 1, CarEmissiveEV 825 at 0 and 0.06, 22,000 at 0.12, 38,720 at 0.54, 22,000 at 0.9 and 825 at 0.96, ForwardPlusEV 0 until 0.926, 4 at 0.946, 5 at 0.97, and a sun peak of 70,000 lux at 0.599. InitialExperience_FreeRoam, the twelfth, has TrackEmissiveEV 1 throughout. Observed

How the lighting passes consume the scales

Section titled “How the lighting passes consume the scales”
  • The world G-buffer pass writes the lit colour seed to RT1.xyz: forward local lights and the emissive term of Other world families. The local-light term is albedo / π times a light record’s intensity, the distance falloff and N·L, with no scene constant in the shader; the emissive term is the material’s emission times g(d) and carries no preset scale either. Observed · the projection_lod _E permutations read
  • FPlusPlus_CombineSeperateDiffuseSpec_Final_Tile0 loads the colour at the pixel from a read-write texture, adds the lit terms to it (in most material-flag branches the loaded colour is passed through unscaled), divides the sum by the scene constant cb[59].x, blends with a sky cube sample times cb[59].z by a fog factor, multiplies by cb[59].x and stores to the same texture. Observed · one compute shader of the 1,687 read The loaded colour is thus in the units of the lit terms and of the sky term cb[59].x · cb[59].z · cube; if cb[59].x · cb[59].z is SkyToLuminanceScale and the seed is in the same unit, the seed is a luminance in cd/m². Inferred
  • Neither that shader nor the deferred composite pixel shader FPlusPlus_Composite_NoDepthWrite_DirectLight_DXR_PixelClassify0 (which reads RT1.xyz only under two material flags, as a retro-reflective colour) multiplies the colour by the distance squared, so nothing there undoes g(d). Observed · two shaders

Open questions

Section titled “Open questions”
Where is TrackEmissiveEV applied?

No G-buffer shader reads multiplies an emissive term by a preset scale and the two lighting shaders read add it unscaled, so the scale reaches the emissive pixel through values the CPU uploads (a material constant, or a scene constant of a pass that was not read). Whether the lit unit is cd/m² or the light buffer’s (cd/m² / 6600) is not established by the files.

Is g(d) undone anywhere?

An emissive pixel at distance d holds e · g(d), 1/d² beyond 1 m. No shader read multiplies it back. A frame capture of one emissive pixel with known texel, distance and preset would fix scale · g(d) and the unit at once.

What does the 0 to 1 curve parameter mean?

Each curve’s time runs from 0 to 1 and the sun of Forte_Summer_01 is 0 until 0.075, rises from 0.091 and is 0 again from 0.926, peaking at 0.54. The preset’s two group floats are 181.86 and 1,181.46, and the SkyPreset labelled Forte_Summer_01 that selects it has 720 and 780. If the time maps linearly onto the two floats as minutes of the day (SkyGlobals has 0 and 1,440), the sun rises between minutes 257 and 273 (04:17 to 04:33), peaks at minute 722 (12:02), and is gone from minute 1,107 (18:27), and the peak lies in the hour of that SkyPreset. Inferred · the arithmetic; no running game was read Whether the mapping is linear and the floats are minutes is not established.