IRON LITANY

Iron Litany — the narrow hull and its unwelcome inhabitants

Version 0.3.0 candidate · 5 October 2026. This records implemented rules, design experiments and the next creative work separately. These particular rites, tank pattern and encounters are original campaign fiction, not quoted Warhammer canon.

A compact hull with a longer journey inside

The same body, surveyed deck, machines and crew are used by both cameras. Seventeen short bulkheads leave real doorways and force turns around the gun, listening cell, chapel and engine. The sloping bow and bevelled hull now read more clearly as armour. There is no second, enlarged first-person map.

Local areaDeck height relative to the main passagePurpose
Raised command cell+0.80 mCommander's driving position, reached through a narrow forward approach
Vox stair and listening cell+0.38 mRecessed receiver above the traverse passage
Forward doglegMain deckTurns and short sightlines beneath the command cell
Port loading galleryMain deckLoader, shell rack and side service approach
Starboard traverse passageMain deckGunner and access to the raised receiver
Raised gun cradle+0.34 mCentral weapon structure, with side aisles around it
Votive recess+0.24 mSmall shrine behind its own rib wall
Return-pipe elbowMain deckCoolant service and a short corner toward the engine
Aft stowage blind−0.18 mA cramped side pocket beside the motive well
Sunken motive well−0.42 mEngine service below the rest of the hull
Custodian's narrow spineMain deckConnects the working rounds

The height field uses continuous ramps at these local changes. Body height, cameras, floor picking, crew, wounds and projectile obstruction all read it. Navigation sweeps the body's width against bulkheads and machines, so a diagonal shortcut cannot clip a doorway corner. Saved bodies from the old layout may move to the nearest valid point only when a newly fitted rib swallowed their old position; invalid new-layout saves are rejected.

In overhead, the current area receives full light, adjoining areas retain 36%, and distant areas retain 13%. Light eases between rooms. Labels show the current working area; the bottom station strip remains a route planner. First person retains the actual lamps and full armour. This is an interior visibility aid, not a claim of exact optical darkness or hidden simulation.

More outside context, still through small slits

The latest creator request explicitly asks for more of the ports to contribute nearly all the time. Overhead now shows short cones of the real shared outdoor scene from all eight open ports, independently of the priest's facing. No pictures float beside the tank and no alternate world is generated for a window.

Each cone extends at most 4.9 m beyond its port, widens by 0.22 m per metre, and fades over its last 0.5 m. Black gaps remain between cones and the world beyond them remains black. Shutter closure narrows the view and grime reduces its brightness. This is a declared cutaway convention: it relaxes the overhead eye-height test so actual ground tops and passing silhouettes remain legible. It does not grant a full external camera.

First person still uses the priest's real eye and facing, both faces of each 0.45 m-deep armour tunnel, its 0.45 × 0.18 m opening, and interior obstruction. Port eye height follows the local deck. Turn and approach the physical aperture to inspect it. Ground, ruins, infantry, armour and wrecks are the same moving geometry in both modes. The bounded outdoor pool has 27 original procedural pieces, with nearer lanes and track-rutted terrain to make incomplete glimpses useful at the actual framebuffer size.

The earlier eye-image captures and foldout helpers remain as a tested historical experiment; they are not instantiated by the current runtime. Current diagnostics report shared-world-cones, zero capture targets and zero offscreen captures. See VISION_DESIGN.md for the current contract and historical evidence.

Office of the Loaded Bearing

After protected instruction, the Motive Altar replaces its ordinary repeated responses with a live weighted lever, oil cup, running belt and sight glass. Preparation costs one oil and one incense. The work is available when engine health is 94 or below, or a relevant incident needs attention.

1. Unload. Bring the weighted clutch below one-quarter load and let it settle. The lever has mass and slews at a bounded rate; setting a target does not instantly move the mechanism.

2. Anoint. Regulate the cup until the oil film stays between the two marks, 0.42–0.66, for 2.2 seconds. Heat and remaining load drain it. An open cup left too long floods the bearing and causes wear.

3. Return. Close the cup. Restore the weight through small detents while retaining the witnessed film. Large jumps in the actual return impose wear, heat and an affront to the spirit. Full return must seat with a retained film before completion.

A slack belt genuinely slows the tank and reduces generation while easing engine demand. The task therefore changes the other duties' timetable. It cannot be solved by rapidly repeating a button, and holding the cup open is not the solution. Completion repairs the engine and restores accord, but does not purge a foreign sender.

Paid preparation, oil film, clutch target and wear survive saving. Pausing freezes the entire watch; restoration releases hands so tools must be deliberately regripped. Two actual-input completions incurred zero wear. One included a save/reload during the return, with the exact film restored. The feedback now explicitly says to close the cup and bring back the weighted lever.

Chaos has a body and an entry point

Small deck wounds appear regularly after instruction: the first at 34 mission seconds, then approximately every 26–41 seconds as the crossing and corruption demand. After a repair halt, the next watch's first seep is due after 24 mission seconds. Existing religious failures can influence the wound location and accelerate growth; low corruption does not make the tank immune.

A fresh wound is visible before its first creature emerges seven seconds later. Early inspection and a three-second binding can prevent that birth entirely. An open wound keeps raising corruption. Its small hunched intruder has two hits of resistance; spreading blight eventually enlarges a surviving body or admits a larger four-hit intruder. Physical pursuit follows the same doorways and staggered deck as the priest.

Switch with F, drag to aim and click or press Space to fire. The six-round consecrated pistol has finite reserve ammunition; R or its visible charging control reloads in 1.5 seconds. The camera initially turns toward a nearby unoccluded threat, but shots still require a real aimed ray. Armour, machines and raised floors block fire. Misses spend ammunition. Service tools must be laid down before firing or reloading.

Shooting kills the body, not its source. After a kill, the wound waits at least 26 seconds before admitting another body. Approach the empty wound and hold E or the fitted cautery contact continuously for three seconds. The first contact pays one seal; interruptions lose progress but retain that paid seal on the wound. Walking away, releasing contact, pausing or losing browser focus releases the hand safely. Only current or adjacent area's blight markers are shown; their walk command takes the priest to a valid point beside the wound, not into it.

If seals run out, three aimed shots can burn an empty wound closed. This spends ammunition, costs four spirit and two trust, and records an affront. It is a material expedient, not a free source of seals or a substitute for every religious duty. Neither purification route silently resolves unrelated crew, vox or personal corruption.

The original pooled manifestation models have cracked bone, torn flesh, horns, rib cages and claws, with separate small and large proportions. Spatial cues use mechanical/material sounds, and the pistol and charging relay have new procedural responses. These are runtime art and sound candidates; procedural scars do not replace authored Diablo I-quality anatomy, motion or a listening review.

What the experiments say about priorities

The complete experiment record compares two controllers on seeds 1, 41, 1984 and 4096, with 0.05-second simulation steps and a 900-second cap. Both walk real routes, pay service and ammunition costs, replace three crew, and reach all three repair halts. The protected lesson is omitted from their timing.

PolicyResult across four seedsPhysical wounds sealedCreatures killedPersonal shots
Attend breaches promptly, then serviceFour simulated victories11 per run5–619–21
Service first, defer breaches until close or numerousFour simulated victories6–816–1947–54

The maintenance-first policy finished about six seconds earlier, but required considerably more ammunition and interruptions. Finishing time alone is therefore a poor quality measure. The useful design reward is preventing repeat fights and keeping the working round intact.

These controllers have ideal aim only when the actual armour admits a shot, and covenant decisions use the visible plume and impressed pins. Ordinary legacy maintenance still reads its next step from the simulation. The experiments do not render the camera, interpret low-resolution art, hear cues or represent a human attention budget. They are balance evidence, not proof that the new campaign is easy, fully balanced or human-playtest approved. Failed preparations also remain common in the synthetic policies; surface availability and player discovery deserve further review.

PLAYTHROUGH_V7.md records actual browser controls separately: an intentionally extended halted watch ended in ward loss; a recovery-focused run reached its first repair halt, performed three replacements and survived saving/reloading.

Next mechanisms to develop and test

These are proposals, not features promised by this build. Keep each tied to a physical consequence and prototype one at a time.

The next visual priority is authored adult crew anatomy and genuine shell/tool hand contact, followed by an approved room exemplar with less uniform amber lighting. Walkable religious refuges, teachers, space transit and campaign politics remain later work. Cinematic and narrated gameplay production still waits for the creator's demo approval.