# Prompt for Fable 5: Voxel Thanjavur, c. 1010 CE

Build a voxel recreation of **Thanjavur under Raja Raja Chola I**, set in the year his temple was consecrated (1010 CE), rendered in **three.js**. Fine-grained voxels — small enough to render curved gopuram profiles and domed kumbams cleanly, not blocky Minecraft-scale cubes.

## Scope

Full walled city: temple complex, palace precinct, market district, residential quarters, fortification wall and moat. This is a city, not a diorama of one building.

## Historical accuracy tier (mixed — this matters, follow it precisely)

**Strict — model these as close to the archaeological record as voxels allow:**
- **Brihadeeswarar Temple ("Periya Kovil")**: rectangular complex ~240.79m (E-W) × 121.92m (N-S), surrounded by a moat (unfilled in this era — do not depict the later filled-in state). Vimana 66m tall, 13 tiers, square base ~30m per side, tapering pyramidal profile (not curvilinear — that's a later Chola style). Granite construction throughout — Thanjavur has no local granite, so the material read should feel imported/precision-cut, not rough-quarried-on-site. Octagonal granite kumbam capstone (~80 tons) at the apex. Nandi mandapam facing the sanctum. Two entrance gopurams on the east axis — outer (Keralantakan tiruvasal) and inner (Rajarajan tiruvasal), inner gopuram ~30m, smaller than the vimana. Subsidiary shrines aligned axially around the main shrine (Parvati, Murugan, Ganesha, Chandeshvara, Nataraja). Since this is 1010 CE, the temple is *newly consecrated* — crisp edges, fresh carving, no centuries of weathering, no later Nayak/Maratha-era fortified outer walls (those came after the 16th century — anachronistic here).
- **Minor/subsidiary temples** in the city: base these on known Chola-period Dravidian temple grammar (square sanctum, pyramidal vimana, mandapas, east-facing) at reduced scale — smaller vimanas, single-tier gopurams, no requirement to match specific named structures.

**Speculative but period-grounded — flag internally as reconstruction, not record:**
- **The royal palace**: important caveat — Raja Raja I's actual palace has never been archaeologically located; nothing of it survives. Do **not** model it on the existing Thanjavur Maratha Palace (Nayak-era, 16th century — 500+ years too late, wrong dynasty, wrong construction). Instead, build a plausible Chola-era royal precinct: brick-and-timber construction (not granite — granite was reserved for temples), pillared durbar hall, courtyards, following the same vastu-purusha-mandala logic as the temple but at a domestic/administrative scale. Treat it as an informed reconstruction, and it's fine for this to read a little more "generic period-appropriate" than the temple. Place it inside the city's fortification walls — take full liberty with its design, but its location is within the city limits, not a separate/outer complex.
- **City fabric**: streets radiating from the temple's east axis, market street(s), residential quarters graded by class (nobility near the palace/temple, artisans and merchants further out, with the Kaveri riverside used for goods transport), fortification wall with gates. Stylize freely here — no record survives, so prioritize a coherent, legible city over invented precision.

## Camera / interaction

Toggle between two modes:
- **First-person**: walk the streets at eye level.
- **Orbit**: free-fly cinematic camera for overview shots.

## Population & economy — rich simulation

This should read as a living imperial capital, not a museum diorama:
- Scheduled routines: priests conducting rituals at temple opening/closing times, market stalls opening/restocking through the day, processions (temple festival processions with musicians are appropriate to the period).
- Elephants and bullock carts moving goods (elephants were literally used to haul temple construction materials in this era — appropriate detail).
- Boats on the Kaveri riverside for trade.
- Market stalls with distinguishable goods — textiles, grain, spices, pearls and gems (Chola-controlled the Gulf of Mannar pearl fisheries and Kolar gold — the market should feel wealthy, not subsistence-level).
- Crowd density should visibly differ between the temple courtyard (pilgrims, priests), the market (merchants, buyers, haggling clusters), and residential quarters (quieter, domestic).

## Secret mode: Tunnel Mode

A toggle (keybind, e.g. `T`) that:
- Fades all surface buildings to translucent/wireframe.
- Reveals an underground tunnel network as glowing paths connecting key buildings (palace, temple, treasury/granary, at least one gate) and extending out beyond the city walls.
- Frame this as a speculative/legendary feature (escape and supply tunnels are a recurring motif in South Indian fort/palace lore) — not a claimed historical fact. The prompt to Fable 5 should say: build this as an atmospheric "secret" layer, clearly a game-like reveal rather than an archaeological claim.
- Transition should be a smooth fade (surface opacity down, tunnel glow up), reversible on toggle-off.

## Audio

Ambient soundscape: temple bells, nagaswaram and oduvar-style devotional singing near the temple, market din (voices, cart wheels, animals) in the market district, quieter ambient tone in residential areas. Spatialize by proximity if the engine supports it; otherwise a layered ambient loop that crossfades by zone is fine.

## Technical constraints (flagged, not resolved — your call)

Fine-grained voxels × full walled city × rich crowd/economy simulation is a heavy combination for a browser three.js target. Use whatever combination of chunking, LOD, instancing, and crowd simplification-at-distance gets this running acceptably — but don't sacrifice the "living capital" feeling by defaulting to a static, empty city. If a tradeoff has to be made, make it explicit in your response rather than silently thinning out the population.

## Stack

Three.js. Structure the build so the tunnel-mode toggle and camera-mode toggle are simple state flags, not separate scenes.