The World's Most Advanced Procedural ║ ║ D&D Dungeon Generator for Grasshopper
============================================================
ADVENTURERS WANTED! — COMPLETE COLLECTION
Wickerson Studios · 2026
Powered by Claude AI · www.wickersonstudios.com
============================================================
"Where Mathematics Becomes Myth.
Where Algorithms Become Legend."
------------------------------------------------------------
WHAT IS THIS?
------------------------------------------------------------
A complete series of Grasshopper C# Script components for
Rhino 7/8 that procedurally generate fully-simulated D&D
dungeons in 3D, animated in real-time, powered by a full
stack of artificial intelligence and optimisation systems.
Each version builds on the last. The collection spans from
a simple 2D dungeon generator all the way to a nine-
dimensional AI engine with neural networks, evolutionary
algorithms, and organic cave geometry.
All scripts are original creations by Wickerson Studios,
generated with Claude AI.
------------------------------------------------------------
THE FILES — VERSION HISTORY
------------------------------------------------------------
v1 WickersonStudios_AdventurersWanted.cs
─────────────────────────────────────
The original. BSP room placement, party pentagons,
monster markers, treasure diamonds, corridor lines.
8 inputs / 8 outputs. OD&D 1974 rules faithful.
v2 WickersonStudios_AdventurersWanted_v2.cs
─────────────────────────────────────────
Full THAC0 tables, Vancian spell slots, 35 monsters,
11 character classes, 18 trap types, 30 room types,
multi-floor generation, animated party march,
torch flicker, monster eye pulse, treasure rotation.
v3 WickersonStudios_AdventurersWanted_v3.cs
─────────────────────────────────────────
D3 = Z — TRUE 3D GEOMETRY
Extruded wall Breps, NURBS floor + ceiling surfaces,
vault rib arches, 3D furniture (pillars, altars,
coffins, thrones, tables, bookshelves, pools),
3D staircases with treads + handrail, 3D door frames,
3D monster geometry, animated chest lid, fog of war.
v4 WickersonStudios_AdventurersWanted_v4.cs
─────────────────────────────────────────
D4 = T — TIME ENGINE
Static state persists between solves. HP depletes
per turn. Spell slots deplete. Torches burn down
every 40 turns — light cones shrink. XP accumulates
→ level-up flash. Monster patrol AI with waypoints.
Noise propagation through rooms. Faction warfare
matrix. Box Brep solid walls. Initiative tracker 3D.
v5 WickersonStudios_AdventurersWanted_v5.cs
─────────────────────────────────────────
D5 = M — METADATA
MaterialRecord (stone type, age, weathering, resonance,
sound attenuation). MagicResonance height-map mesh.
NarrativeThread catenary curves between rooms.
AtmosphereRecord (fog, cold, smoke, humidity).
ItemProvenance (previous owners, history, curses).
MonsterMemory (alert states, aggression learning).
EconomicIndex (treasure value inflation). CharacterTrauma
(permanent injuries from near-death). 14 outputs.
v6 WickersonStudios_AdventurersWanted_v6.cs
─────────────────────────────────────────
D6 = Φ — TOPOLOGY / ORGANIC CAVE FORM
Fourier-series wall profiles: f(θ) = R + Σ[Aₙ·cos(n·θ+φₙ)]
6 cave morphology types with unique harmonic coefficients.
fBm Perlin noise (quintic fade, multi-octave).
Lamé super-ellipse barrel vault ribs.
Catenary stalactites/stalagmites.
Rhodonea crystal clusters: r = cos(k·θ).
Trochoid underground rivers.
Cubic Bézier + Lissajous corridor cross-sections.
Lamé fossil profiles with ammonite spirals.
Geothermal vents. CaveLOD input (1→5).
v7 WickersonStudios_AdventurersWanted_v7.cs
─────────────────────────────────────────
D7 = Ψ — INTELLIGENCE
A* pathfinding: f(n) = g(n) + h(n), Euclidean heuristic.
Dijkstra SSSP priority queue — party finds safest route.
Behaviour Tree: Selector/Sequence nodes, 6 states
(SLEEP/PATROL/ALERT/HUNT/AMBUSH/FLEE).
Utility scoring via sigmoid σ(noise−2).
Genetic Algorithm: tournament selection, crossover,
swap mutation, fitness = treasure/(danger+1).
Ant Colony Optimisation: τ(t+1)=(1-ρ)τ(t)+Δτ, ρ=0.3.
Monte Carlo combat: 50-trial outcome distribution.
Minimax target selection: depth-2 tree.
Utility AI: MU/Cleric spell casting selection.
Gradient descent difficulty auto-balance.
Reinforcement learning: monster aggression accumulation.
v8 WickersonStudios_AdventurersWanted_v8.cs
─────────────────────────────────────────
D8 = AI∞ — DEEP LEARNING
Neural Network: 8→12→8→1 feedforward, Xavier init,
ReLU/ReLU/Sigmoid, predicts encounter difficulty.
MCTS: UCB1 = Q/N + C√(lnN_parent/N), 4-iter rollout.
Q-Learning: 125 states × 5 actions, α=0.1, γ=0.9.
PSO: v(t+1) = ω·v + φ_p·rp·(pbest−x) + φ_g·rg·(gbest−x).
k-Means Clustering: Lloyd's, k++ seeding, Voronoi.
Bayesian Optimisation: GP with RBF kernel, UCB acq.
HTN Planning: hierarchical task decomposition.
Max-Flow Network: Edmonds-Karp BFS, min-cut corridors.
All v7 systems retained. 19 inputs / 15 outputs.
v9 WickersonStudios_AdventurersWanted_v9.cs
─────────────────────────────────────────
D9 = ∞ — THE Nth DIMENSION
CMA-ES: full covariance matrix adaptation, σ step-size
control, rank-μ update. Gold-standard black-box opt.
DQN: Deep Q-Network with experience replay (cap 512),
target network sync every 20 turns, ε-decay.
Policy Gradient (REINFORCE): ∇J(θ)=E[∇logπ·G],
party casters learn optimal spell casting policy.
NSGA-II: non-dominated sorting, crowding distance,
Pareto front of danger vs reward objectives.
Cellular Automata: B3/S12345 Moore neighbourhood,
cave morphology evolves each turn.
Spectral Clustering: normalised graph Laplacian,
Fiedler vector via power iteration, bipartitions dungeon.
Self-Organising Map: Kohonen 8×8, h(d)=exp(−d²/2σ²).
Tabu Search: |T|=7, aspiration criterion, trap placement.
L-System: F→F[+F]F[-F]F, branching corridor grammar.
Flood Erosion: BFS wave, proportional stone wear.
All v8 + v7 systems retained. 20 inputs / 16 outputs.
BONUS WickersonStudios_JoyOfLife_v2.cs
─────────────────────────────────────────
Parametric art script — 5 world systems animated:
GARDEN (Fibonacci phyllotaxis sunflower, rhodonea roses),
COSMOS (logarithmic spirals), OCEAN, FOREST,
CELEBRATION (fireworks). Speed + AnimSpeed controls.
------------------------------------------------------------
ANIMATION SETUP (v2 through v9)
------------------------------------------------------------
1. Params > Util > Timer — set interval to 80ms
2. Connect Timer output → the script component
3. Boolean Toggle → Animate input (TRUE = run)
4. Button → Reset input (rewinds state)
5. Number Slider 0.005→0.2 → AnimSpeed input
Recommended AnimSpeed: 0.04
Recommended CaveLOD: 3 (v6–v9 only)
------------------------------------------------------------
RECOMMENDED INPUT VALUES
------------------------------------------------------------
PartySize 4 (classic D&D party)
DungeonLevel 3–7 (scales danger, monster HD)
Seed 42 (change for new dungeon)
RoomCount 15 (5–40 range)
TreasureRate 0.4 (0=none, 1=every room)
TrapRate 0.3
NumFloors 2 (1–3)
CaveLOD 3 (v6–v9: 1=fast, 5=max detail)
AI TOGGLES (v7–v9):
ShowAI true A* paths + Dijkstra routes
ShowACO true ant colony pheromone trails
ShowGA true GA fitness bars + PSO particles
ShowNN true neural network activation rings
ShowClusters true Voronoi territory boundaries
ShowPareto true NSGA-II Pareto frontier (v9)
ShowLSystem true L-system branching corridors (v9)
------------------------------------------------------------
D&D RULES IMPLEMENTED
------------------------------------------------------------
· OD&D (1974) + Holmes Basic (1977) + B/X (1981)
· THAC0 tables by class group (Fighter/Cleric/MU)
· 5 saving throw categories per class
· Vancian spell slots — MU + Cleric levels 1–6
· Encumbrance + movement rate calculation
· STR/CON attack + HP bonuses
· Morale checks (2d6 vs morale score)
· Surprise rolls (1–2 on d6)
· Reaction rolls (2d6 table)
· Initiative (d6 + DEX modifier)
· Critical hits (natural 20 = double damage)
· 35+ monsters with HD, damage, XP, morale, faction
· Treasure types A–H with magic item tables
· 11 character classes with race integration
· Faction hostility matrix (Chaos vs Chaos, etc.)
------------------------------------------------------------
OUTPUT LAYERS BY VERSION
------------------------------------------------------------
v1–v2 2D curves: walls, rooms, party, monsters, FX
v3 SolidWalls / FloorCeiling / Furniture3D (Brep)
v4 + TurnState (live dashboard) / CombatLog
v5 + ResonanceField / AtmosphereGeo / NarrativeWeb
/ SoundMesh / MetaReport
v6 CaveWalls (Fourier) / FloorCeiling (Perlin noise)
/ Furniture3D (crystals, stalactites, rivers)
v7 + AIOverlay (A*, Dijkstra, ACO) / GAFitnessMap
/ MCDistribution
v8 + NNOverlay / ClusterMap / FlowNetwork
/ PSOParticles
v9 + LSystemGeo / ParetoFront / full v9 AI stack
------------------------------------------------------------
TECHNICAL NOTES
------------------------------------------------------------
· Rhino 7 or 8 required
· Grasshopper C# Script component
· No external dependencies — pure .NET standard library
· Static fields maintain state between Grasshopper solves
· Timer component drives animation loop (80ms interval)
· PseudoRand() uses unchecked{} for intentional overflow
· Dijkstra uses SortedSet<(double d, int i)> with named
tuple Comparer to avoid C# inference ambiguity
· CMA-ES covariance update uses diagonal approximation
for computational efficiency in script context
------------------------------------------------------------
CREDITS
------------------------------------------------------------
Scripts Wickerson Studios · 2026
AI Engine Claude (Anthropic)
Website www.wickersonstudios.com
Rules OD&D / Holmes / B/X (1974–1981)
Inspired by Gary Gygax · Dave Arneson · B. Holmes
"The dungeon remembers. The monsters learn.
Every wall is a polynomial. Enter at your own risk."
============================================================
www.wickersonstudios.com · Enter at your own risk. ☠
============================================================