Fitted offices, captured aim and the moving battlefield — v0.4.0
This revision is a set of playable interaction experiments, not final acceptance of the art or campaign balance. The aim is to make the priest read and handle a material witness instead of repeatedly selecting a prescribed answer from a menu. The same simulation runs underneath both cameras.
The first seconds
The tank crawls under covering fire during the protected opening lesson. Tracks, ground and the shared battlefield already move. Three covering rounds use actual ammunition and animate the loader; the campaign clock, enemy incursions and unattended losses remain suspended until the engine answers. The first lesson now uses the real weighted motive lever and oil cup.
The optional Practice the offices watch opens all instruments with an ample, finite reagent allotment. It keeps the same costs, tool dynamics, wear and completion conditions, without new intrusions or campaign losses. Returning to the campaign restores its separate save. Practice never overwrites it.
Experiments to compare
| Station | What the hand does | What the eye and ear must judge | Consequence and question for playtesting |
| Engine | Slack a weighted drive, meter the unction cup, close the cup and return the weight in small detents. | Moving oil film, loaded bearing and material contact. | Mishandling wears the bearing; clutch position changes tank speed. Does tending two coupled controls feel like caring for a living mechanism? |
| Shrine / personal tether | Swing a damped censer beneath five wax witnesses, then bring the chain to rest. | Wax gradually turns ivory; smoke, momentum and seated-witness sounds. | Rushing spills coals. Only the presence caught when work began is bound; a later sender survives. Does this feel contemplative under pressure, or too much like painting a progress bar? |
| Gunner | Turn a correction wheel that controls sight velocity, then centre it to stop. | Keep the copper crosshair inside a moving pale reference. | The firing circuit waits while his responses are corrected. Is steering momentum enjoyable enough to keep? |
| Loader | Slide the heavy shell along its cradle, then turn its locking collar. | Shell lag, collar position and the ivory witness band. | Forcing the shell damages the witness; the crew cannot fire during handling. Does the two-stage tactile handoff read clearly? |
| Cannon | Move a spring-loaded recoil counterweight. | Settle both position and motion inside the ivory marks. | The cannon is interlocked during work. Is the relation between spring motion and cannon recoil clear enough? |
| Cooling | At low pressure, turn the return-filter bypass to follow its drifting band. At high pressure, rotate the relief wheel against a live gauge. | Pressure, discharge, seal wear and the returning pipe knocks. | Over-bleeding causes material wear; closing the relief allows recovery. Does this create worthwhile judgment rather than busywork? |
| Commander | Bring his governor to the ordered marching position and hold its mechanical response steady. | A marked order and a settling weight. | Governor position affects tank speed while worked; his discipline recovers but temperament remains. This is deliberately the simplest experiment and may need more negotiation. |
| Vox | Regulate smoke, challenge actual witness pins and trace their impressed route with a weighted stylus. | A false audible answer without the matching raised pin must be distrusted. | A wrong branch scars the circuit. This earlier, more elaborate experiment remains available for comparison. |
The fitted plate itself accepts dragging or turning. Keyboard arrows adjust, C centres, and Home/End reach the stops. This is an accessibility alternative to the same physical values, not a faster completion path. The pressure wheel retains fine adjustments in a disclosure. Emergency repair, inspection, orders and quarantine remain explicit decisions.
The plates are experimental instruments in the station panel. They are not yet fully modelled first-person hand animations. Several instruments share position-and-settling logic; they are six contrasting service experiments, not a claim of six wholly unrelated game systems. Playtesting should retain the ones with the strongest identity and replace the weak ones.
Personal combat
Q draws the pistol, enters first person and requests native mouse capture. Relative mouse motion aims the camera around a centred crosshair. Q holsters and releases the cursor immediately. A first click after a denied capture requests aiming without spending a bullet. Click or Space fires, R reloads, and F returns overhead with free hands.
An actual close strike from an intruder records the attacker, switches to first person, draws and faces that attacker, and requests capture. A strike cannot pass through a bulkhead. Already armed players retain their own aim. Escape releases capture and pauses; lost focus pauses the watch. Resume makes a fresh capture request when the pistol is still drawn. After a browser refusal or Escape, a fresh click may be required by browser policy; the UI reports the actual capture state.
The desktop permission policy allows pointer capture only for the bundled main document. All other requested permissions remain denied. Source: Electron session permission handling and browser pointer-lock behavior.
Constructed tank and shared exterior
The bore, chamber, barrel bands, moving recoil assembly and muzzle light share one axis. The commander and his controls sit beside that axis, with a separate reachable priest position. Turret teeth, trunnions and ammunition supports give the gun a visible mounting. Rear engine shafts, transmission, final drives, brakes, radiator grilles and exhausts make a continuous power train. Track skirts, towing eyes and edge-following pipes finish the silhouette without closing the surveyed corridors.
Eleven cramped compartments, short rib walls and local raised/sunken platforms remain one map. Current-room lighting preserves the sense of hidden adjoining spaces. The overhead convention now projects eight narrow cones about 8.8 metres beyond the armour, rather than 4.9. First person still uses the actual eye, small opening and tunnel; closed shutters and armour occlude the view.
A bounded shared exterior contains 240 instanced corpse assemblies, 18 braced riflemen and the existing pools of ruins, armour and other battlefield pieces. Three reusable effect slots carry rifle flashes, tracers, smoke and artillery. Nothing creates a separate scene per window. Geometry and sound consume the same deterministic shot timeline; tracers terminate at the armour.
Exterior blasts pass through a muffled bus. Interior cannon fire retains its heavy impact and rattling return; footsteps, working tools, claws and seated-witness sounds have their own material character. The exterior drops under machine-spirit speech and under intricate witness work. Seven existing original AudioX samples are reused; no new generated audio or licensed game samples were added. A running audio graph and objective tests do not constitute listening approval.
Art boundary
Keep adult proportions, assembled iron mechanisms, bone and parchment, rust, olive and selective warm lamps. The low-resolution scene stays separate from readable interface type. Added troops, corpses and mechanisms are original procedural candidates. Diablo I remains the target; this revision does not establish final art fidelity or authorize the cinematic.