Subway: enemy spawns
A large abandoned subway station — platforms, concourses, shops and back-of-house rooms.
- Level id
subway- Difficulty tier
- 0 — Explicit difficulty tier of the level (0 = the starting level). Metadata only: no automatic scaling of counts, damage or loot, and no enemy family is enabled by it.
LEVELS.subway.tierin game/src/levels/registry.ts:53 (repository access required) - Spawn profile
- Subway —
SPAWN_PROFILES.subwayin game/src/ai/spawnTables.ts:869 (repository access required) - Spawns tried per level
- 22–32 (E = 27), over 2 generation contexts
- Room modifiers
- All 45 room keys resolve to 1× for every family (neutral); the profile sets no room modifiers
Generation contexts
A generation context is one run of the enemy placement code. Each run draws one count, uniformly from its range, and tries to place that many spawns. The contexts below are parts of the one Subway level and are generated with it; they are not separate levels. A context that runs more than once per level (for example once per storey) draws a new count each run.
| Context | Storeys one run covers | Runs per level | Count per run | E[count] per run | Per level | E per level |
|---|---|---|---|---|---|---|
| Stationstation | the main level, then the storey moves onto the upper / lower levels | 1 | 22–32U{22..32} | 27 | 22–32 | 27 |
| Service annex (per annex)service | one seeded service annex off a track tunnel (its rooms and corridors only) | 0–2one per zone (seeded) | 5–6U{5..6} | 5.5 | 0–12zone budget, not in the total | 5.5per zone |
| Whole level | all core contexts (zones reported separately) | 22–32 | 27 |
Spawns the level's context runs try to place (Σ runs × count). The family roll replaces, never adds; placed totals can fall short where eligible, spaced cells run out. Not included: the zone context 'service' (a per-zone budget of 5-6 per zone, 0-12 per level), reported separately.
Service annex (per annex): A PER-ANNEX ZONE BUDGET: one run per seeded service annex (0 to TRACK_EXT.annex.max per level, a generation census quantity — see docs/enemy-spawns.md "Service annexes"), on the annex's own fork, appended after every core spawn and tagged 'service'. It is additive to the core contexts and NOT in the whole-level total above; the level's zombie types apply; its spawns draw from the annexes' OWN family pool (SERVICE_FAMILY_POOL: a configured 20 % rat/dog selection share, Blightbelly 5 %, Strainer 8 % (the only pool that enables it: STRAINER_CONTEXTS), no Siren — per-spawn probabilities, not quotas). The seeded roster pass then runs on the finished level (see the seeded roster): it fixes the level's Strainers at EXACTLY ONE on the maps whose seed rolled the Strainer present, and at none on the others, and it decides per level whether Blightbellies and rat/dogs are there at all (the slot rules of the roster-only Screacher and BB list the core contexts, not the annex). The sampled census shows the outcome with the pass applied. Killed annex enemies stay dead for the run (per-index memory).
Service annex (per annex): its own family pool (configured selection share per spawn) — zombie 67 % · blightbelly 5 % · siren 0 % · ratdog 20 % · strainer 8 % · screacher 0 % · bb 0 %. This context's OWN pool: its spawns draw their family from it (the same single roll per spawn, room modifiers and eligibility on top) instead of the level pool. The shares are CONFIGURED selection probabilities per spawn, not quotas: a 5-6-enemy annex may hold none or several; the sampled census shows the placed outcome.
These are targets: the generator can place fewer when it runs out of eligible, spaced floor (see placement rules). The game has no per-area or per-square-metre spawn rate; the count is spread over the eligible areas of the context.
Generator: generateStation in game/src/gen/station.ts:78 (repository access required). One station per level seed: placeSpawns (fork "spawns", after the loose floor finds) places the station context on the main level, placeLevelSpawns (fork "levelSpawns") moves a weighted share onto the upper / lower storeys; then (Track 2) placeServiceSpawns (gen/annexContent.ts, fork "track-ext-content") runs the service context once per seeded service annex, appended last; assignEnemyArchetypes rolls every family once at the end. Last of all the level registry runs the SEEDED ROSTER pass on the finished layout (ai/enemyRoster.ts applyEnemyRoster, roster version 4): from the level seed alone it decides which special families this map holds at all, and it only ever changes the family of ordinary ambient zombie slots — never a count, a position, an index, a horde member or an outfit zombie (see the seeded roster).
Enemy families
Every placed spawn is one enemy family. A family's conditional share = weight / Σ weights (denominator 1) is the chance that the per-spawn roll of one placed spawn draws that family. It changes the mix, not how many spawns are placed. The seeded roster then decides whether this level holds each special family at all.
Every placed spawn draws its family independently: roll = enemyRoll(level seed, spawn index) ∈ [0, 1), walked over the pool weights of the spawn's room (the level pool when the room has no modifier), in registry order (the rat/dog keeps the top band; a family at weight 0 — the roster-only Screacher / BB — has no band). A drawn family REPLACES a zombie; it never adds to the count. Last, the seeded roster pass (roster version 4) runs once on the finished level: a family the level's seed rolled absent has exactly none there, a present one is placed by its policy (see the seeded roster).
| Family | Weight | Conditional share (per-spawn roll) | Expected per level by the per-spawn roll alone (before the roster) | Seeded roster: chance the level's roll includes it | Per level with the roster (sampled) |
|---|---|---|---|---|---|
| Infectedzombie | 0.9 | 0.9 / 1 | 24.3 | alwaysthe baseline: never rolled | in 20 of 20 levels · 24.05 per level sampled |
| Blightbellyblightbelly | 0.05 | 0.05 / 1 | 1.35 | 60%of level map seeds · share top-up | in 16 of 20 levels · 2.1 per level · 2.63 per level that holds it sampled |
| Sirensiren | 0 | 0 / 1 | 0 | nevernot rolled on this level | in 0 of 20 levels · 0 per level sampled |
| Tunnel scavenger (rat/dog)ratdog | 0.05 | 0.05 / 1 | 1.35 | 60%of level map seeds · share top-up | in 13 of 20 levels · 1.7 per level · 2.62 per level that holds it sampled |
| Strainerstrainer | 0 | 0 / 1level pool; its weight is in a context's own pool: Service annex (per annex) 8% | 0 | 35%of the map seeds whose map has a spawn in its contexts · exactly one, in its contexts | in 1 of 20 levels · 0.05 per level · 1 per level that holds it sampled |
| Screacherscreacher | 0 | 0 / 1roster-only: no per-spawn roll draws it (placed by the seeded roster alone) | 0 | 40%of level map seeds · a count in place of ordinary zombies | in 5 of 20 levels · 0.45 per level · 1.8 per level that holds it sampled |
| BBbb | 0 | 0 / 1roster-only: no per-spawn roll draws it (placed by the seeded roster alone) | 0 | 40%of level map seeds · a count in place of ordinary zombies | in 5 of 20 levels · 0.5 per level · 2 per level that holds it sampled |
Expected per level by the per-spawn roll alone = share × spawns tried per level (E = 27), with neutral room
modifiers and before any placement shortfall.
That is the stage before the roster pass. The last two columns are the seeded roster: a level either
holds a family or has none of it, so the per-level figure of a level that holds it is higher than the long-run mean.
Defined at ENEMY_POOL in game/src/ai/spawnTables.ts:125 (repository access required).
Seeded roster
The seeded roster pass (roster version 4) runs once on the finished Subway layout, after every generation context, the per-spawn family roll, the eligibility rules and the hordes. One roll per family per level map seed; every storey of the level shares it. It never changes how many enemies the level has or where they stand — only which family an ordinary ambient zombie slot is.
| Family | Can this level hold it | Chance the roll includes it | On this level | The roll itself over level seeds 1 to 10000 | Sample levels with the roster (sampled) |
|---|---|---|---|---|---|
| Infectedzombie | always: the baseline, never rolled | alwaysthe baseline: never rolled | Always: the baseline family, never rolled. The other families only ever take ordinary ambient zombie slots. | not rolled | in 20 of 20 levels · 24.05 per level sampled |
| Blightbellyblightbelly | yes: every map of the level is rolled | 60%of level map seeds · share top-up | The roll includes it in 60 % of this level's map seeds: there the per-spawn roll's placements stay and are topped up, and such a map holds at least one. Every other map holds none. | 5991 of 10000 include it59.9% of those seeds | in 16 of 20 levels · 2.1 per level · 2.63 per level that holds it sampled |
| Sirensiren | no: never rolled on this level | nevernot rolled on this level | Never rolled: this level's profile gives the family no weight and no eligibility rule (it is opt-in per level). | not rolled | in 0 of 20 levels · 0 per level sampled |
| Tunnel scavenger (rat/dog)ratdog | yes: every map of the level is rolled | 60%of level map seeds · share top-up | The roll includes it in 60 % of this level's map seeds: there the per-spawn roll's placements stay and are topped up, and such a map holds at least one. Every other map holds none. | 6054 of 10000 include it60.5% of those seeds | in 13 of 20 levels · 1.7 per level · 2.62 per level that holds it sampled |
| Strainerstrainer | depends on the map: only a map with a spawn in its contexts is rolled | 35%of the map seeds whose map has a spawn in its contexts · exactly one, in its contexts | The roll includes it in 35 % of the map seeds whose map has an ambient spawn in its context here (service): such a map holds exactly one, there. A map without such a spawn is never rolled, and every other map holds none. | 3553 of 10000 include it35.5% of those seeds | in 1 of 20 levels · 0.05 per level · 1 per level that holds it sampled |
| Screacherscreacher roster-only | yes: every map of the level is rolled | 40%of level map seeds · a count in place of ordinary zombies | The roll includes it in 40 % of this level's map seeds: such a map holds its count. Every other map holds none. | 3932 of 10000 include it39.3% of those seeds | in 5 of 20 levels · 0.45 per level · 1.8 per level that holds it sampled |
| BBbb roster-only | yes: every map of the level is rolled | 40%of level map seeds · a count in place of ordinary zombies | The roll includes it in 40 % of this level's map seeds: such a map holds its count. Every other map holds none. | 4096 of 10000 include it41% of those seeds | in 5 of 20 levels · 0.5 per level · 2 per level that holds it sampled |
How a family this level holds is placed
- Blightbelly (share top-up). Share top-up ("option B"), at least one. What the per-spawn roll placed stays; then every remaining ordinary ambient zombie slot whose pool gives the family weight converts with chance share × (1 / rate − 1) × (ambient slots whose pool gives it weight / the ordinary zombie slots among them). A level that holds the family so has about (the per-spawn roll's number) / rate of it, and over many levels the family total stays about what the per-spawn roll alone placed — concentrated on fewer levels. At rate 1 nothing is topped up. PRESENT MEANS AT LEAST ONE: if the per-spawn roll and the top-ups together left none, one ordinary ambient zombie slot whose pool gives the family weight becomes it (picked by weight: a slot with a larger share is more likely), also at rate 1. Only a level with no such slot at all holds none (the pass reports "no slot"). Its rule.
- Tunnel scavenger (rat/dog) (share top-up). Share top-up ("option B"), at least one. What the per-spawn roll placed stays; then every remaining ordinary ambient zombie slot whose pool gives the family weight converts with chance share × (1 / rate − 1) × (ambient slots whose pool gives it weight / the ordinary zombie slots among them). A level that holds the family so has about (the per-spawn roll's number) / rate of it, and over many levels the family total stays about what the per-spawn roll alone placed — concentrated on fewer levels. At rate 1 nothing is topped up. PRESENT MEANS AT LEAST ONE: if the per-spawn roll and the top-ups together left none, one ordinary ambient zombie slot whose pool gives the family weight becomes it (picked by weight: a slot with a larger share is more likely), also at rate 1. Only a level with no such slot at all holds none (the pass reports "no slot"). Its rule.
- Strainer (exactly one, in its contexts). Exactly one, in its contexts (service, damp). Only a map that has an ambient spawn in one of them is rolled. Present: one the per-spawn roll placed there is kept and any others become ordinary zombies; with none, an ordinary ambient zombie slot there converts — else the slot of a share-top-up family's member there, which is exchanged to an ordinary zombie slot elsewhere. A level whose roll includes the family always holds exactly one. Its rule.
- Screacher (a count in place of ordinary zombies). Counted: placed by the roster alone (no per-spawn roll can draw it). One count per level map — 1 in 40 %, 2 in 40 %, 3 in 20 % of the levels that hold it — and that many ordinary ambient zombie slots of the contexts station, street, floor, roof, pit, ring become the family: only in an area at least 6 cells (3 m) on its shorter side and 80 cells (20 m²) large, never area kind closet and never a room of type / purpose exit-stair, washroom, office-stair, stair-core, office-elevator, supply-closet, janitor-closet, electrical-closet; never a horde member, an outfit zombie or a spawn of another context. Slots are taken in a seeded order, each at least 14 m (straight 3D line) from the ones already taken: where the level cannot space them out the level holds as many as fit, at least one. A level with no such slot at all relaxes the rule to any ordinary ambient zombie of those contexts, then to any ordinary ambient zombie. Each one takes a model variant (sc01, sc02, sc03, sc04), dealt in turn from a seeded start. Its rule.
- BB (a count in place of ordinary zombies). Counted: placed by the roster alone (no per-spawn roll can draw it). One count per level map — 1 in 35 %, 2 in 40 %, 3 in 25 % of the levels that hold it — and that many ordinary ambient zombie slots of the contexts station, street, floor, pit, ring become the family: only in an area at least 5 cells (2.5 m) on its shorter side and 60 cells (15 m²) large, never area kind closet and never a room of type / purpose exit-stair, washroom, office-stair, stair-core, office-elevator, supply-closet, janitor-closet, electrical-closet; never a horde member, an outfit zombie or a spawn of another context. Slots are taken in a seeded order, each at least 9 m (straight 3D line) from the ones already taken: where the level cannot space them out the level holds as many as fit, at least one. A level with no such slot at all relaxes the rule to any ordinary ambient zombie of those contexts, then to any ordinary ambient zombie. Each one takes a model variant (lullaby, binky, teddy, changeling), dealt in turn from a seeded start. PLACED LAST (its rule applies from roster version 3): after every family of an earlier roster version has its slots, whatever their policies. It takes only ordinary ambient zombie slots those families left, so each of them stands exactly where the earlier version put it. Its rule.
Order: Strainer, then Screacher, then Blightbelly, then Tunnel scavenger (rat/dog), then BB — the most constrained first (BB last: added in a later roster version, it takes only slots the others left), each on the ordinary zombie slots still left.
Sampled roster census
sampled Observed, not a rule. SAMPLED: the same 20 generated levels of LEVELS.subway.generate (seeds 5003..155464, 20 seeds, shipped start weapon), each followed by the seeded roster pass at roster version 4 — what a level holds. "Before" = what the per-spawn roll alone had placed; "after" = with the pass applied. The pass never changes the 28.85 ambient spawns per level of this sample, only their families. A SMALL sample: the long-run share of levels that hold a family is its roster rate, not this count. The roll depends on the level seed and the family only, so these seeds roll the same families present on every level. Ambient spawns only (horde members are the separate horde pool). Not exhaustive.
- Blightbelly: rolled present in 16 of the 20 sample levels; 16 of them hold it. Long-run rate: 60% of level map seeds.
- Tunnel scavenger (rat/dog): rolled present in 13 of the 20 sample levels; 13 of them hold it. Long-run rate: 60% of level map seeds.
- Strainer: 5 of the 20 sample levels could hold it (a map with a spawn in its contexts); 1 of those rolled it present and 1 hold it. Long-run rate: 35% of such maps.
- Screacher: rolled present in 5 of the 20 sample levels; 5 of them hold it. Long-run rate: 40% of level map seeds.
- BB: rolled present in 5 of the 20 sample levels; 5 of them hold it. Long-run rate: 40% of level map seeds.
| Family | Levels that could hold it | Levels whose roll included it | Levels that hold it | Placed by the per-spawn roll alone (before the pass) | Slots the pass took from it | Slots the pass gave it | With the roster (after the pass) | Mean per level | Mean per level that holds it | Most in one level | Share of the ambient enemies | Present without a legal slot | Levels by number held (number: levels) |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Infectedzombie | 20 | 20 | 20 | 515 | 49 | 15 | 481 | 24.05 | 24.05 | 33 | 0 | — | |
| Blightbellyblightbelly | 20 | 16 | 16 | 31 | 7 | 18 | 42 | 2.1 | 2.63 | 4 | 0 | 0: 4 · 1: 2 · 2: 6 · 3: 4 · 4: 4 | |
| Sirensiren | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | — | 0 | 0 | 0: 20 | |
| Tunnel scavenger (rat/dog)ratdog | 20 | 13 | 13 | 31 | 8 | 11 | 34 | 1.7 | 2.62 | 5 | 0 | 0: 7 · 1: 4 · 2: 3 · 3: 2 · 4: 2 · 5: 2 | |
| Strainerstrainer | 5 | 1 | 1 | 0 | 0 | 1 | 1 | 0.05 | 1 | 1 | 0 | 0: 19 · 1: 1 | |
| Screacherscreacher | 20 | 5 | 5 | 0 | 0 | 9 | 9 | 0.45 | 1.8 | 3 | 0 | 0: 15 · 1: 2 · 2: 2 · 3: 1 | |
| BBbb | 20 | 5 | 5 | 0 | 0 | 10 | 10 | 0.5 | 2 | 3 | 0 | 0: 15 · 1: 1 · 2: 3 · 3: 1 |
“Slots the pass took from it”: spawns the per-spawn roll made this family that are another family after the pass (an absent family's all become ordinary zombies). “Slots the pass gave it”: spawns that became this family in the pass. For the baseline family the two are the slots the other families took and gave back.
Roster version 4; the rules of the pass and every family's rule: the seeded roster.
Defined at applyEnemyRoster in game/src/ai/enemyRoster.ts:285 (repository access required).
Zombie types
Not separate enemy families. The types (coated, shambler) split the Infected family only; the family shares above do not change with them. Wardrobe outfits are a further appearance choice, also not a family.
| Type | Chance per spawn |
|---|---|
| coated | |
| shambler |
Each spawn of a run: coated with the coated chance, else shambler; then, if the run has no coated, its first spawn becomes coated, and (count > 1) if it has no shambler, its second becomes shambler. Ratdogs keep the rolled type but ignore it.
| Context | Spawns in the run (n) | Expected coated share |
|---|---|---|
| Station | 22 | |
| Station | 32 | |
| Service annex (per annex) | 5 | |
| Service annex (per annex) | 6 |
Shambler / coated type (the profile's zombie type roll) and the wardrobe outfit chosen at spawn time from (level seed, spawn index, type) by ai/wardrobe.ts pickZombieLook — sub-distributions of the zombie family, not separate families.
Defined at ZOMBIE_TYPES in game/src/ai/spawnTables.ts:379 (repository access required).
Baseline placement rules
These are the generator's own rules for where the counted spawns go. Room modifiers at 1× leave every one of them unchanged. They are also why a generated level can hold fewer spawns than its count drew.
Station
One run covers: the main level, then the storey moves onto the upper / lower levels. Count per run uniform integer U{22..32} (rng.int, inclusive); E[count] = 27.
Which cells can hold a spawn
- a FLOOR cell of the context's storeys (not the outer border row)
- walking distance from the start ≥ 36 cells (18 m)
- every cell within 1 cell(s) walkable
- the area's effective weight (area weight × room intensity) > 0
| Rule | Value | Unit | In metres |
|---|---|---|---|
| walking distance from the player start | 36 | cells | 18 |
| clear walkable cells around the candidate (radius) | 1 | cells | 0.5 |
Which areas are visited first
First matching rule wins (back-of-house rooms and service corridors otherwise). A relative weight in the weighted area order, multiplied by the room intensity D of its room key (1 when the room has no modifier). Weight 0 = never a candidate.
| Areas matching | Area weight | Why |
|---|---|---|
| kind platform | 2.2 | platforms: the long public edges by the tracks |
| type exit-stair / washroom | 0.5 | closed street exits and washrooms: dead-end pockets |
| type eatery / shop | 0.5 | storefront units |
| public space | 1 | other public halls, passages and concourses |
| kind closet | 0.6 | closets |
| any other area | 1.4 |
How many spawns one area may hold
The placement walks the area order once per pass (caps 1, 2, 3); an area takes one more spawn per visit while it is below its cap for that pass. Room modifiers never raise a cap.
| Areas matching | Cap | Why |
|---|---|---|
| kind platform | pass cap + 1 | platforms hold one more per pass |
| kind closet or type exit-stair or storefront unit | at most 1 | closets, closed exits and storefront units: at most one |
| any other area | the pass cap |
- Tries per area visit
- 12 random eligible cells
- Spacing
- at least 4 m between spawns (straight line)
Placement defined at SUBWAY_PLACEMENT in game/src/ai/spawnTables.ts:370 (repository access required).
How a run places its count, and why it can fall short
- Draw count = U{22..32} (one draw per run), then the zombie types of that many spawns.
- Collect every eligible cell, grouped by area; order the areas by a weighted shuffle (Efraimidis–Spirakis keys u^(1/weight)).
- For each cap pass (1, 2, 3), walk the order: an area below its cap gets up to 12 random cell picks; the first pick ≥ 4 m from every enemy already placed takes one spawn.
- Stop when the count is reached. Spacing, caps and eligible floor can leave a run short of its count (placed < count); validation regenerates the attempt where the profile requires it.
- Each placed spawn is then a family by its own roll (see the family pool) — independent of the area unless a room modifier says otherwise.
Checks that reject a whole generation attempt
- fewer than 8 placed in one run → the whole attempt is regenerated
- an enemy closer than 36 walking cells to the start, on an unreachable cell or intersecting geometry → the attempt is regenerated
Context defined at SPAWN_PROFILES.subway.contexts.station in game/src/ai/spawnTables.ts:872 (repository access required).
Service annex (per annex)
One run covers: one seeded service annex off a track tunnel (its rooms and corridors only). Count per run uniform integer U{5..6} (rng.int, inclusive); E[count] = 5.5.
Which cells can hold a spawn
- a FLOOR cell of the context's storeys (not the outer border row)
- walking distance from the start ≥ 36 cells (18 m)
- every cell within 1 cell(s) walkable
- the area's effective weight (area weight × room intensity) > 0
- an area of the zone (one service annex), ≥ 6 m from the annex's entry door (its keep point)
| Rule | Value | Unit | In metres |
|---|---|---|---|
| walking distance from the player start | 36 | cells | 18 |
| clear walkable cells around the candidate (radius) | 1 | cells | 0.5 |
| straight-line keep-off from stair / elevator landings, the roof hatch and the ladder foot | 6 | m | 6 |
Which areas are visited first
First matching rule wins (annex rooms (pump / electrical rooms, stores, workshops, custodial rooms) otherwise). A relative weight in the weighted area order, multiplied by the room intensity D of its room key (1 when the room has no modifier). Weight 0 = never a candidate.
| Areas matching | Area weight | Why |
|---|---|---|
| kind closet | 0.6 | annex closets (janitor / electrical / supply) |
| kind service | 1 | annex service corridors (1.5-2 m wide) |
| any other area | 1.4 |
How many spawns one area may hold
The placement walks the area order once per pass (caps 1, 2); an area takes one more spawn per visit while it is below its cap for that pass. Room modifiers never raise a cap.
| Areas matching | Cap | Why |
|---|---|---|
| kind closet | at most 1 | closets: at most one |
| any other area | the pass cap |
- Tries per area visit
- 12 random eligible cells
- Spacing
- at least 5 m between spawns (straight line)
Placement defined at SERVICE_PLACEMENT in game/src/ai/spawnTables.ts:405 (repository access required).
How a run places its count, and why it can fall short
- Draw count = U{5..6} (one draw per run), then the zombie types of that many spawns.
- Collect every eligible cell, grouped by area; order the areas by a weighted shuffle (Efraimidis–Spirakis keys u^(1/weight)).
- For each cap pass (1, 2), walk the order: an area below its cap gets up to 12 random cell picks; the first pick ≥ 5 m from every enemy already placed takes one spawn.
- Stop when the count is reached. Spacing, caps and eligible floor can leave a run short of its count (placed < count); validation regenerates the attempt where the profile requires it.
- Each placed spawn is then a family by its own roll (see the family pool) — independent of the area unless a room modifier says otherwise.
Checks that reject a whole generation attempt
- more than 6 placed in one run → the whole attempt is regenerated
- an enemy closer than 36 walking cells to the start, on an unreachable cell or intersecting geometry → the attempt is regenerated
Context defined at SPAWN_PROFILES.subway.contexts.service in game/src/ai/spawnTables.ts:874 (repository access required).
Storey moves
After the station context, when every enemy is still on the main level, a weighted share moves onto the upper / lower storeys (spawns on stair flights always move). Totals are unchanged.
- How many move
- moved = min(max(placed − 6, 0), max(storeys with candidates, round(placed × W_levels / (W_levels + W_main)))), W = Σ effective area weights; one area per storey first, then by the weighted order, 3 passes, 12 tries per area visit
- Kept on the main level
- at least 6
- Area weights
- same area weights as the station context
- Passes / tries
- 3 passes, 12 tries per area visit
- Spacing
- 4 m horizontally, 2 m vertically
- Keep-away
- 36 cells of walking distance from the start
- Defined at
SPAWN_PROFILES.subway.levelMovesin game/src/ai/spawnTables.ts:876 (repository access required)
Moves only relocate spawns that were already counted: the level total and the family shares are unchanged.
Room modifiers
All 45 room keys resolve to 1× for every family (neutral); the profile sets no room modifiers. A room key or family without an entry resolves to 1×, which leaves the baseline placement unchanged.
Room key = the area's purpose id (gen/purposes.ts) when it has one, else its room type. Each field resolves: purpose entry → room-type entry → 1 (missing = neutral). Two modifiers exist per room key and family. Density changes local intensity: how strongly the room's areas are preferred when spawns are spread. Weight changes conditional composition: which family a spawn placed in that room is. Neither changes the count range, and caps and spacing still apply. No explicit entries: every other room / family resolves to 1×. Rows list the room keys seen in the sampled census plus any explicit entry; a key not listed resolves to 1× as well.
| Modifier | Changes | How | At 0 | At 2 |
|---|---|---|---|---|
| Density (local intensity) | local intensity | area weight × D, D = all.density × Σ_f p_f·w_f·d_f / Σ_f p_f·w_f (p = pool weight, w = weight modifier, d = density modifier); the room's family mix uses p_f·w_f·d_f | D = 0: the room is never a candidate (no spawns there); the count roll and the other rooms are unchanged, so its spawns go elsewhere when capacity allows. | The room is visited earlier in every cap pass, so it tends to fill its caps before the count runs out: more spawns there, fewer elsewhere, the same total. |
| Weight (conditional composition) | conditional composition | P(family f | spawn lands in the room) = p_f·w_f·d_f / Σ_g p_g·w_g·d_g | That family never spawns in that room (its spawns there become the other families). | Twice its relative odds in that room before normalisation; the room still attracts the same number of spawns. |
The profile sets no explicit room modifier entries: every room key below resolves to the neutral 1×.
| Room key | Key kind | Sampled room types | Room intensity D(room) | Infected: density | Infected: weight | Blightbelly: density | Blightbelly: weight | Siren: density | Siren: weight | Tunnel scavenger (rat/dog): density | Tunnel scavenger (rat/dog): weight | Strainer: density | Strainer: weight | Screacher: density | Screacher: weight | BB: density | BB: weight |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Station agent officeagent-office | Room purpose | Station agent office agent-office | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
| Bakerybakery | Room purpose | Food counter eatery | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
| Burger counterburger | Room purpose | Food counter eatery | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
| Coffee barcoffee | Room purpose | Food counter eatery | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
| Concourseconcourse | Room type | Concourse concourse | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
| Convenience storeconvenience | Room purpose | Shop shop | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
| Custodial & waste roomcustodial-room | Room purpose | Storage room storage | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
| Deli counterdeli | Room purpose | Food counter eatery | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
| Donut shopdonut | Room purpose | Food counter eatery | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
| Electrical closetelectrical-closet | Room purpose | Janitor closet janitor-closet, Supply closet supply-closet | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
| Electrical roomelectrical-room | Room purpose | Utility room utility | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
| Street exit (closed)exit-stair | Room type | Street exit (closed) exit-stair | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
| Ticket hallfare-hall | Room type | Ticket hall fare-hall | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
| First-aid roomfirst-aid-room | Room purpose | Crew room crew-room | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
| Floristflorist | Room purpose | Shop shop | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
| Food courtfood-court | Room type | Food court food-court | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
| Hot-dog standhotdog | Room purpose | Food counter eatery | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
| Janitor closetjanitor-closet | Room purpose | Janitor closet janitor-closet | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
| Juice barjuice | Room purpose | Food counter eatery | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
| Key cutting & shoe repairkeys | Room purpose | Shop shop | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
| Crew locker roomlocker-room | Room purpose | Crew room crew-room, Storage room storage | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
| Lost property officelost-property | Room purpose | Crew room crew-room, Storage room storage | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
| Newsstandnewsstand | Room purpose | Shop shop | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
| Noodle counternoodles | Room purpose | Food counter eatery | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
| Stair passagepassage | Room type | Stair passage passage | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
| Pharmacypharmacy | Room purpose | Shop shop | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
| Phone repairphone | Room purpose | Shop shop | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
| Pizza counterpizza | Room purpose | Food counter eatery | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
| Platformplatform | Room type | Platform platform | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
| Pump roompump-room | Room purpose | Utility room utility | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
| Retail arcaderetail-arcade | Room type | Retail arcade retail-arcade | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
| Fare revenue roomrevenue-room | Room purpose | Crew room crew-room, Storage room storage, Utility room utility | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
| Transit security officesecurity-office | Room purpose | Utility room utility | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
| Service corridorservice-corridor | Room purpose | Service corridor service-corridor | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
| Shawarma grillshawarma | Room purpose | Food counter eatery | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
| Signal relay roomsignal-room | Room purpose | Utility room utility | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
| Staff roomstaff-room | Room purpose | Crew room crew-room, Storage room storage, Utility room utility | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
| Stairsstair-core | Room type | Stairs stair-core | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
| Stock roomstock-room | Room purpose | Stock room stock-room | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
| Maintenance storagestorage | Room purpose | Storage room storage, Utility room utility | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
| Supply closetsupply-closet | Room purpose | Janitor closet janitor-closet, Supply closet supply-closet | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
| Sushi countersushi | Room purpose | Food counter eatery | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
| Electronics storetech | Room purpose | Shop shop | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
| Washroomwashroom | Room purpose | Washroom washroom | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
| Maintenance workshopworkshop | Room purpose | Storage room storage, Utility room utility | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× | 1× |
Room intensity D(room) multiplies the room's baseline area weight (see placement rules). Defined
at roomIntensity in game/src/ai/spawnTables.ts:992 (repository access required). Meaning of each field: room modifiers.
Sampled census
sampled Observed, not a rule. These figures come from running the real Subway generator on a small fixed corpus: 4 generated layouts from seeds 11, 222, 3333, 44444, pistol start. Eligible floor is the floor the generator's own candidate filter accepted (74322 cells of 0.25 m², 18580.5 m² in total), not the map's bounding box. Densities are placed spawns per 100 m² of that eligible floor over this sample only. Families are counted with the seeded roster applied: what these levels hold. The roster changes families only, so the totals are the generator's.
SAMPLED: 4 sample levels of the registered generator (seeds 11, 222, 3333, 44444, shipped start weapon) WITH THE SEEDED ROSTER APPLIED (roster version 4: what a level holds; the roster only changes families, so counts, positions and eligible floor are the generator's). Eligible floor = cells the generator's own candidate filter accepted (walking distance, clearance, storey, stair / landing keep-off, weight > 0) x 0.25 m² — never bounding boxes. Not exhaustive; not a runtime rule.
| Family | Placed over the sample (with the roster) | Mean per level | Sample levels that hold it | Per 100 m² eligible floor (sampled) | Placed by the per-spawn roll alone (before the roster pass) |
|---|---|---|---|---|---|
| Infectedzombie | 99 | 24.75 | 4 of 4 | 0.533 | 103 |
| Blightbellyblightbelly | 6 | 1.5 | 4 of 4 | 0.032 | 2 |
| Sirensiren | 0 | 0 | 0 of 4 | 0 | 0 |
| Tunnel scavenger (rat/dog)ratdog | 1 | 0.25 | 1 of 4 | 0.005 | 4 |
| Strainerstrainer | 0 | 0 | 0 of 4 | 0 | 0 |
| Screacherscreacher | 2 | 0.5 | 1 of 4 | 0.011 | 0 |
| BBbb | 1 | 0.25 | 1 of 4 | 0.005 | 0 |
| Context | Eligible floor m² | Placed over the sample | Mean per level | Per 100 m² eligible floor (sampled) | Placed by family (sampled outcome) |
|---|---|---|---|---|---|
| Station | 18580.5 | 109 | 27.25 | 0.587 | zombie 99 (90.8 %) · blightbelly 6 (5.5 %) · ratdog 1 (0.9 %) · screacher 2 (1.8 %) · bb 1 (0.9 %) |
| Area kind | Eligible floor m² | Infected placed | Blightbelly placed | Siren placed | Tunnel scavenger (rat/dog) placed | Strainer placed | Screacher placed | BB placed | Per 100 m² eligible floor (sampled) |
|---|---|---|---|---|---|---|---|---|---|
| closet | 35.75 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 2.8 |
| concourse | 9299 | 22 | 0 | 0 | 0 | 0 | 1 | 1 | 0.258 |
| passage | 731.5 | 8 | 0 | 0 | 1 | 0 | 0 | 0 | 1.23 |
| platform | 5542.3 | 4 | 0 | 0 | 0 | 0 | 1 | 0 | 0.09 |
| room | 1796.3 | 48 | 4 | 0 | 0 | 0 | 0 | 0 | 2.89 |
| service | 1175.8 | 16 | 2 | 0 | 0 | 0 | 0 | 0 | 1.53 |
| Type | Placed over the sample |
|---|---|
| coated | 42 |
| shambler | 57 |
| Room key | Baseline area weight(s) | Layouts with it | Areas | Eligible floor m² | Infected placed | Blightbelly placed | Siren placed | Tunnel scavenger (rat/dog) placed | Strainer placed | Screacher placed | BB placed | Per 100 m² eligible floor (sampled) |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Station agent officeagent-office | 1.4 | 4 | 4 | 47 | 4 | 0 | 0 | 0 | 0 | 0 | 0 | 8.51 |
| Bakerybakery | 0.5 | 3 | 4 | 23.25 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| Burger counterburger | 0.5 | 2 | 2 | 14.75 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| Coffee barcoffee | 0.5 | 4 | 7 | 39.25 | 1 | 1 | 0 | 0 | 0 | 0 | 0 | 5.1 |
| Concourseconcourse | 1 | 4 | 33 | 4936.8 | 13 | 0 | 0 | 0 | 0 | 0 | 1 | 0.284 |
| Convenience storeconvenience | 0.5 | 4 | 5 | 19 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 5.26 |
| Custodial & waste roomcustodial-room | 1.4 | 2 | 2 | 43.25 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 2.31 |
| Deli counterdeli | 0.5 | 3 | 6 | 57.25 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 1.75 |
| Donut shopdonut | 0.5 | 4 | 6 | 30.75 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| Electrical closetelectrical-closet | 0.6 | 4 | 5 | 10 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| Electrical roomelectrical-room | 1.4 | 4 | 7 | 85.75 | 3 | 0 | 0 | 0 | 0 | 0 | 0 | 3.5 |
| Street exit (closed)exit-stair | 0.5 | 4 | 12 | 144 | 4 | 0 | 0 | 0 | 0 | 0 | 0 | 2.78 |
| Ticket hallfare-hall | 1 | 4 | 7 | 1949.8 | 5 | 0 | 0 | 0 | 0 | 0 | 0 | 0.256 |
| First-aid roomfirst-aid-room | 1.4 | 3 | 3 | 46 | 0 | 1 | 0 | 0 | 0 | 0 | 0 | 2.17 |
| Floristflorist | 0.5 | 3 | 5 | 7.75 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 12.9 |
| Food courtfood-court | 1 | 4 | 4 | 988.5 | 2 | 0 | 0 | 0 | 0 | 0 | 0 | 0.202 |
| Hot-dog standhotdog | 0.5 | 4 | 5 | 17.75 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 5.63 |
| Janitor closetjanitor-closet | 0.6 | 4 | 9 | 10.75 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| Juice barjuice | 0.5 | 4 | 4 | 18 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 5.56 |
| Key cutting & shoe repairkeys | 0.5 | 4 | 5 | 11.25 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| Crew locker roomlocker-room | 1.4 | 3 | 3 | 56.75 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| Lost property officelost-property | 1.4 | 4 | 4 | 50.25 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| Newsstandnewsstand | 0.5 | 4 | 5 | 15 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 6.67 |
| Noodle counternoodles | 0.5 | 2 | 2 | 18.25 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| Stair passagepassage | 1 | 4 | 14 | 381.5 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0.262 |
| Pharmacypharmacy | 0.5 | 4 | 5 | 37.25 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 2.68 |
| Phone repairphone | 0.5 | 4 | 6 | 5.25 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| Pizza counterpizza | 0.5 | 3 | 3 | 22.75 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 4.4 |
| Platformplatform | 2.2 | 4 | 8 | 5542.3 | 4 | 0 | 0 | 0 | 0 | 1 | 0 | 0.09 |
| Pump roompump-room | 1.4 | 4 | 4 | 53.75 | 2 | 1 | 0 | 0 | 0 | 0 | 0 | 5.58 |
| Retail arcaderetail-arcade | 1 | 4 | 7 | 1424 | 2 | 0 | 0 | 0 | 0 | 1 | 0 | 0.211 |
| Fare revenue roomrevenue-room | 1.4 | 4 | 4 | 58.5 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 1.71 |
| Transit security officesecurity-office | 1.4 | 4 | 4 | 78 | 2 | 0 | 0 | 0 | 0 | 0 | 0 | 2.56 |
| Service corridorservice-corridor | 1.4 | 4 | 34 | 1175.8 | 16 | 2 | 0 | 0 | 0 | 0 | 0 | 1.53 |
| Shawarma grillshawarma | 0.5 | 4 | 4 | 38.25 | 2 | 0 | 0 | 0 | 0 | 0 | 0 | 5.23 |
| Signal relay roomsignal-room | 1.4 | 4 | 4 | 38 | 2 | 0 | 0 | 0 | 0 | 0 | 0 | 5.26 |
| Staff roomstaff-room | 1.4 | 4 | 8 | 98.25 | 3 | 0 | 0 | 0 | 0 | 0 | 0 | 3.05 |
| Stairsstair-core | 1 | 4 | 30 | 206 | 3 | 0 | 0 | 1 | 0 | 0 | 0 | 1.94 |
| Stock roomstock-room | 1.4 | 4 | 15 | 423.3 | 8 | 0 | 0 | 0 | 0 | 0 | 0 | 1.89 |
| Maintenance storagestorage | 1.4 | 4 | 8 | 116 | 4 | 0 | 0 | 0 | 0 | 0 | 0 | 3.45 |
| Supply closetsupply-closet | 0.6 | 4 | 7 | 15 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 6.67 |
| Sushi countersushi | 0.5 | 3 | 3 | 21.25 | 2 | 0 | 0 | 0 | 0 | 0 | 0 | 9.41 |
| Electronics storetech | 0.5 | 2 | 4 | 29 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 3.45 |
| Washroomwashroom | 0.5 | 4 | 8 | 87.25 | 1 | 1 | 0 | 0 | 0 | 0 | 0 | 2.29 |
| Maintenance workshopworkshop | 1.4 | 4 | 4 | 88.25 | 3 | 0 | 0 | 0 | 0 | 0 | 0 | 3.4 |
How to read these numbers
- Count range = intensity: how many spawns one run of a context tries to place, drawn uniformly. E[count] = (lo + hi) / 2.
- Conditional share = weight / Σ weights: the chance that one placed spawn draws that family. It sets the mix, not the number. A family with an eligibility rule places fewer than it draws (measured, sampled).
- Seeded roster = the chance that this level holds a special family at all, rolled once per family from the level seed, after the per-spawn roll. A level has the family or has none of it; the enemy count never changes.
- Area weights and caps decide where spawns go and are not chances. Eligibility, spacing, caps and tries can leave a level with fewer spawns than the count drew.
- Per 100 m² figures are sampled over eligible floor of a named corpus, never a runtime rule.
More: How to read enemy numbers.
Where to tune it
Each value names the source file and export where it lives (the game repository is private).
| Value | Current | Where it lives |
|---|---|---|
| count range (station) | 22–32 | SPAWN_PROFILES.subway.contexts.station.count in game/src/ai/spawnTables.ts:872 (repository access required) |
| count range (service) | 5–6 | SPAWN_PROFILES.subway.contexts.service.count in game/src/ai/spawnTables.ts:882 (repository access required) |
| family pool weights | zombie 0.9, blightbelly 0.05, siren 0, ratdog 0.05, strainer 0, screacher 0, bb 0 | ENEMY_POOL in game/src/ai/spawnTables.ts:125 (repository access required) |
| coated chance | 0.4 | ZOMBIE_TYPES in game/src/ai/spawnTables.ts:379 (repository access required) |
| area weights | 5 rules, otherwise 1.4 | SUBWAY_AREA_WEIGHTS in game/src/ai/spawnTables.ts:327 (repository access required) |
| caps | passes 1 / 2 / 3; 2 rules | SUBWAY_CAPS in game/src/ai/spawnTables.ts:339 (repository access required) |
| tries per area visit | 12 | SUBWAY_PLACEMENT.tries in game/src/ai/spawnTables.ts:371 (repository access required) |
| spacing (m) | 4 | SUBWAY_PLACEMENT.spacingM in game/src/ai/spawnTables.ts:371 (repository access required) |
| keep-away from the start (cells)18 m walking | 36 | SUBWAY_PLACEMENT.startCells in game/src/ai/spawnTables.ts:371 (repository access required) |
| area weights | 2 rules, otherwise 1.4 | SERVICE_AREA_WEIGHTS in game/src/ai/spawnTables.ts:387 (repository access required) |
| caps | passes 1 / 2; 1 rules | SERVICE_CAPS in game/src/ai/spawnTables.ts:396 (repository access required) |
| tries per area visit | 12 | SERVICE_PLACEMENT.tries in game/src/ai/spawnTables.ts:406 (repository access required) |
| spacing (m) | 5 | SERVICE_PLACEMENT.spacingM in game/src/ai/spawnTables.ts:406 (repository access required) |
| keep-away from the start (cells)18 m walking | 36 | SERVICE_PLACEMENT.startCells in game/src/ai/spawnTables.ts:406 (repository access required) |
| landing keep-off (m) | 6 | SERVICE_PLACEMENT.landingKeepM in game/src/ai/spawnTables.ts:406 (repository access required) |
| storey moves (keep on main, passes, tries, spacing) | 6, 3, 12, 4 m / 2 m | SPAWN_PROFILES.subway.levelMoves in game/src/ai/spawnTables.ts:876 (repository access required) |
| room modifiersroom key → family (or all) → {density, weight} | none (all 1×) | SPAWN_PROFILES.subway.rooms in game/src/ai/spawnTables.ts:877 (repository access required) |
| difficulty tiermetadata only (no scaling) | 0 | LEVELS.subway.tier in game/src/levels/registry.ts:53 (repository access required) |
| seeded roster rate (blightbelly)chance a level map holds the family at all; one roll per level seed | 0.6 | FAMILY_ROSTER.blightbelly in game/src/ai/enemyRoster.ts:154 (repository access required) |
| seeded roster rate (ratdog)chance a level map holds the family at all; one roll per level seed | 0.6 | FAMILY_ROSTER.ratdog in game/src/ai/enemyRoster.ts:158 (repository access required) |
| seeded roster rate (strainer)chance a level map holds the family at all; one roll per level seed | 0.35 | FAMILY_ROSTER.strainer in game/src/ai/enemyRoster.ts:160 (repository access required) |
| seeded roster rate (screacher)chance a level map holds the family at all; one roll per level seed | 0.4 | FAMILY_ROSTER.screacher in game/src/ai/enemyRoster.ts:162 (repository access required) |
| seeded roster rate (bb)chance a level map holds the family at all; one roll per level seed | 0.4 | FAMILY_ROSTER.bb in game/src/ai/enemyRoster.ts:175 (repository access required) |