An extent read from the first or the last asteroid now fails the belt test (finding s3.belt-fixture-first-read). Builds.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The first fixture hid the big sun's weight in the barycentre (it stood at the origin) and tested reach with two roots only, where nearest and farthest coincide.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
WellsSuperpose is removed (its rule, roots superpose everywhere, is the one multi-root-zone-shape decides); HeightIsThePotentialOverOneK and FlowFieldTests probe inside each well's reach. ZoneWorlds gains Belt and AsteroidSize; fade and ring tolerances are relative to the Plummer pull. Not yet run on dotnet at this commit.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
One zone node at the massive roots' centre of mass owns the only global term (SubzoneReach is the one place the subzone shape is decided: isolated-subzones). WellNode gains Extent and Anchor; a belt orbit's extent is its outermost asteroid. GravityField.SeriesFrom = 0.4 is emitted into the shader template and Gravity.cginc. Tests: the spec's WellLawTests; the old seam test is replaced by TheBandPotentialIsExactInSinglePrecision. GravityFieldParityVerify runs the multi-root fixture (second sun, moon inside the planet's Roche limit, far window): PASS, largest drawn tau 1.228, max error 0.01028 u. The dotnet suite has not been run at this commit.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Batchmode on Starfire, Unity 6000.3.24f1: PASS, max error 0.01077 u (tolerance 0.03), gradient 0.07 degrees. The previous
frames teleported the planetoid away from its planet, which a compact tree rightly does not draw. WellTree.cs.meta added.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Full suite at the previous head: 2003 passed, 1 failed (the seam test, whose 3e-6 was tighter than the closed form's
single-precision floor, 1e-5 measured, 1.5e-5 in the research).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Hand-built zones move from OrbitsFromFieldTests into ZoneWorlds so WellLawTests reuses them. Not yet run.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Gravity.cginc is GravityShader.Source() output. OrbitsFromFieldTests spell periods from GravityField.Pull composed as the law
(binary members feel the partner's pull at the chord, the outer body feels c(total mass)). Fixtures whose bodies carried
mass 0 under constant curves (WellFixture, the mining scenes) now carry a mass: the tree conserves mass. DrawnPoseTests'
rails use the Plummer pull. Builds; not yet run.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
GravityField is now the band term U(eps, inner, outer) (C4 smoothstep fade, closed form and a series for tau > 1).
WellTree is built once per Zone from the orbit tree: node terms, reach, support, the pruned Force/Potential walk and
PullAround (rails). Zone.GetHeight/GetForce/PeriodAround delegate to it; WellReference, _wellOn, _strengths,
Strength() and DrawAhead.Well are gone; GravityLowering and the shader template take the flat term list;
HeightExaggeration halves to 2.0491803.
Does not pass yet: Gravity.cginc is not regenerated and PotentialOwnerTests/OrbitsFromFieldTests still spell the old law.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Generator rules read from the band law's reach (Roche margin, cusp bound, window ending
before the parent's fade, sibling spacing, capacity folding into the parent), applied as a
re-placement of today's zones; basin metrics for ruling wells-read-as-nested-basins (cusp,
blob, nesting, saddle curvature, dead ground, steep share, lanes) and a numpy port of
IsolinesUI.shader rendering four zones per candidate. Stable series branch for the band
potential when the softening is comparable to the band (tau > 1).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The belt fixture's PlanetSettings lacked AsteroidSize, AsteroidHitpoints and AsteroidRespawnTime, so building the zone
threw before the bound was read. Builds; the previous commit's belt test fails without this.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Fog Tint is B10G11R11_UFloatPack32. A quad sized to 1e-5 of the centre drew a plateau near 3.2e-5 from 1.75 to 2.1 P and
then zero: a step in the tail the RFloat verify could not see. AmbientFloor is now 2^-15 absolute and the quad ends where
the brightest channel reaches it (1.78 P), so the one step to dark is the format's last value, at most 1.5 floors. The
verify builds its target from the production render texture's descriptor, asserts the format, holds the profile to the
gaussian within the format's 2^-5 step where it is at least 8 floors, and bounds the rim and the step to dark by 2^-14.
Builds; AmbientEdgeVerify passes on Starfire.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A belt's asteroids circle its orbit's parent, so a belt reaches the chain's reach to that orbit plus its Radius
(AsteroidBelt.Radius, the largest asteroid Distance); the chain walk is one function for bodies and belts. The fixture
inserts a smaller body after the outermost so a last-wins bound fails, and a belt beyond the outermost body pins the
belt term. Builds; Unity-side files are in the next commit.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Both target cameras are deferred, so the AfterForwardOpaque buffer never drew. The render half of scene-stardust-uncullable now attaches-checks BeforeForwardAlpha, renders with cullingMask 1 (DrawProcedural is layer 0), and requires a near control (dust 100 below, size 5) to draw before the far probe is read, so a miss at 10000 can only be culling.
Builds: not compiled by this commit alone; verified in the Unity batchmode run reported with the cut.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
At size 5 the material's distance^.33 scaling made every quad sub-pixel, so the probe saw no dust with or without the cull. Builds.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Ambient Fog Light leaves Zone Brushes (root object, same pose) and Sector Tint.mat moves from
PowerBrush to Brushes/Ambient Brush: brightness falls as exp(-x^2 ln(1/_EdgeBrightness)), x in
units of Zone.PopulatedRadius. ZoneRenderer.FitAmbientLight, called by LoadZone, is the one
writer of the quad's size and _Reach (the radius where it is 1e-5 of the centre). AmbientEdgeVerify
renders the real scene object through that call. Builds in Unity 6000.3.24f1.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Deletes the bounded Graphics.DrawProcedural, its hand-inflated Bounds, the GravityCamera field (and its scene reference) and the dead OnRenderObject block. The camera replays the buffer at AfterForwardOpaque; it is attached on enable, detached on disable and destroy, and released. Unity compile not yet run for this commit.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
With them on, the effects wrote their own frame and the probe could not tell dust from noise. Builds; the check fails on the bounded draw (buffer count 0, no dust pixels, run against 5a788833 plus this probe).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The largest over bodies of the root orbit's offset from the origin, every orbit Distance
along the chain, and the body's Roche limit: a static bound on where any body can be,
computed once at construction and independent of Pack.Radius. Presenters read it.
Builds; the Unity side follows in the next commit.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Fails on the bounded Graphics.DrawProcedural (this commit builds; the check fails until the cut lands).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
VolumeRender and VolumeRenderCustom shaders, the four Fog Marching
materials, ZoneRenderer's FogMaterial/FogFarFadeFraction and its
_DepthCeiling/_DepthBlend writes, the ARPG scene values, and the
Volumetric.cginc _TintExponent, texFlow and tintGradient (with its
commented-out caller). The live nebula is VolumeCloudRenderer + CloudShader.
Unity compile not yet run on this commit.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Each side of a station grows outward from it: a turret sits innerReach + 2 turretReach beyond the body inside it,
so turrets of differing reach clear each other (same arcs as before for one turret hull at k = -1, +1).
StationPlacementTests: shared flank check (no off-ring turret escapes it, 1e-4 relative arc), a narrative
station with five authored turrets and a second, larger turret hull, a zone with no turret hull, and a sweep
that no station or turret reaches another orbital. Not yet run.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The generator picks from every non-rosette planet orbit, empty nodes
included, so the source is the one ring orbit no orbital entity rides.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
GameplaySettings carries ShipLayerHeight 3, OrbitalLayerHeight 15, FloaterLayerHeight 0 (authored in Settings.asset).
Entity.LayerHeight (Ship, OrbitalEntity) is the one offset RideHeight adds; DrawAhead.FloaterHeight is the floater
height ItemPickup and MineInstance draw at. HullData.GridOffset (Key 26) and the pickup/mine/prefab GridOffset fields
are gone with every reader. Fixtures (DrawnPose, FireControl*, ImpactFeedback) take heights from the layer owner;
LayerTests pin each kind's layer. FloatingItemsDrawnSmoke follows. ItemPickup/MineInstance.Bind take the settings from
ZoneRenderer. Builds as one commit with its tests; not compile-verified until the cut's head run.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
CreateLagrangeOrbit adds one sixth of a turn (was PI / 3 on a turns field).
PlaceTurret spaces turrets by k * (stationReach + 2 * turretReach) arcs,
k = -1, +1, -2, +2 (was t / 2 integer division putting two turrets on the
station, and radians-per-arc arithmetic on turns). Contacts.Radius moves to
HullData.Reach. Not yet verified on Yggdrasil.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Step pairs OfType<Ship>; stations and turrets sit on the orbital layer above the ship plane
(ruling aetheria-gravity:orbitals-above-floaters-below). Tests: a ship flies through a station and a turret;
two ships still bounce beside an orbital. Not yet compile-verified (verified once at the cut's head).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Operator flew 39ba73d5 on 2026-10-16: no jitter coasting beside bodies. Soul s1-s3; residuals in follow-up drawn-pose-residuals.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- ActionGameManager: the visible hostile/friendly indicators and MenuMap.Position read DrawAhead.Position at Clock.Lead.
- DrawAhead.Round drops its lead parameter: the target's lead is the drawn time less its zone's step time, derived inside,
so a caller cannot pass two times that disagree. Projectile and BallisticFlightTests updated; the hit tests derive their
lead from the drawn time; two assertions that only varied the now-absent lead are removed.
- Zone.RailAt logs "Requested orbit is not part of this zone!" for a missing parent again (DrawnPoseTests.AMissingParentOrbitIsLoggedAtADrawnTime).
Builds: not yet compiled at this commit (verified by the run that follows).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
HUD target, lock, incoming-lock, bearing and sweep markers anchor on DrawAhead.Position / AtBearing at Clock.Lead.
GravityLowering's record fill is DrawAhead.Well (pure, pinned at non-zero lead). Rounds take DrawAhead.Time
instead of float Zone.Time + lead. Sensor ping, shield hit, compass icons, wormhole bearing, loot ViewOrigin and
the guided-projectile aim read the drawn pose. Not built in Unity yet.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
GetOrbitPosition's step-time fill and RailAt each decided the rail chain (root at
FixedPosition, child at parent plus offset). The fill is now RailAt(id, _time).
DrawnPoseTests builds the sun on an off-centre FixedPosition (the ZoneGenerator
cluster-root shape) and adds a root test at step and drawn time, and an unset-orbit
OrbitalEntity fact for the DrawAhead guard.
All commits build, unverified until the job in the report.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
