Brainstorming

Synthesis from Motivation & Inspiration

• Focus: Single-player basketball RPG with world exploration, player progression, and story-driven elements.
• Inspirations: NBA2K MyPlayer, Pokémon (exploration/collection), RPGs (customization, story), Captain Tsubasa (unique backgrounds), city builders (discovery), roguelites (replayability).
• Avoids: Grind, pay-to-win, pure team management sim.

Game World Ideas

• Semi-Open World: World consists of different regions, countries, and a global area. Exploration is rewarded (mentors, items, story events).
• World Progression: Player starts in one region and must succeed to unlock access to others, eventually reaching a global stage.
• Dynamic World: Regions have unique cultural traits and influencing factors. Street tournaments, scouting, and rival encounters occur dynamically.
• Rivalries & Relationships: Rivalries can form between players, squads, and even regions, influencing story and matches. Long-lasting rivalries can evolve into nemesis relationships.
• Fictional League: Creative freedom for rules, characters, and locations. (TO BE DECIDED: Real-world countries or a fantasy world?)
• Fog-of-War Exploration: Each region begins hidden; courts, shops, gyms, and points of interest are revealed through player movement (Civilization-like uncovering) with seed-based procedural variation per run.
• Region Unlock Criteria: “Domination” of current region (win key tournaments / prestige threshold) unlocks travel to a new region; regions offer unique training buffs (e.g., stamina gains x2) and mentor pools.
• Procedural Seeding: World, local players, rival archetypes, and location buffs generated from a season/run seed ensuring replayable variety while enabling reproducibility.

Core Mechanics

Player & Squad Progression

• Player Attributes: Skills level up via play, training, and rare mentors/items.
• Player Traits: Unique backgrounds (inspired by Captain Tsubasa, NBA 2K Badges and Project Zomboid) give special abilities or challenges.
• Player Knowledge: Playing with or against others reveals their attributes, skills, traits, and tendencies over time.
• Collectibles: Gear (like shoes and other wearables), skills, or perks found in the world affect play style and attributes. (TO BE SPECIFIED FURTHER)
• AI Adaptation: After a loss, AI players might train or recruit to improve their capabilities and strategies.

On-Court Gameplay

• Game Modes: Games range from 1v1, 2v2, to 3v3, with the player always participating.
• Arcade-Sim Hybrid: Accessible but deep controls.
• Signature Moves: Unlock/equip special moves or animations.
• Team Chemistry & Familiarity:
  • Relationships with teammates affect performance. Chemistry is influenced by winning, losing, or impressive plays.
  • High familiarity with squad members unlocks set plays and advanced strategies (e.g., pick and roll, post-ups). (TO BE SPECIFIED FURTHER)
• Customizable Strategies: A building-block engine allows for creating custom plays. Availability of blocks depends on player IQ, skills, chemistry, and familiarity. (TO BE SPECIFIED FURTHER)
• Teammate Control Model: Possession-centric selection: when any squad member has the ball the player can choose from context-aware action cards (dribble variants, pass lanes, shot types, special) with success chances derived from attributes, traits, fatigue, chemistry. Off-ball teammates follow AI behaviors influenced by familiarity & assigned development goals.
• Knowledge Reveal: Repeated encounters gradually surface hidden rival stats/traits, lowering uncertainty in success probability ranges.

Off-Court Gameplay

• Squad Building:
  • Build a squad of up to 4 players (3 starters, 1 rotation for 3v3).
  • If the squad is incomplete, random local players are available for selection.
  • Recruiting requires providing incentives (e.g., money, playing time, chance of success). (TO BE SPECIFIED FURTHER)
  • Strategic decision: Evolve the current squad or recruit better players at the cost of chemistry and familiarity.
• Training & Mentorship: Choose how to spend time (train, explore, build relationships).
• Story Choices: Decisions affect path, unlock/close opportunities.
• Indirect Teammate Development: Player sets wearables and dialog-driven growth focuses (e.g., “improve perimeter defense”), but teammates autonomously gain XP and evolve; direct micro-training is reserved for the user’s avatar to preserve differentiation.
• Mentor & Location Synergy: Visiting specialty courts or gyms applies temporary training modifiers; mentors may require prior exploration knowledge or rivalry prestige thresholds.

Roguelite Elements

• Run-based Progression: Each season/run is unique (mentors, rivals, events).
• Permanent Progression: Some unlocks/knowledge persist between runs (backgrounds, skills, starting bonuses).
• Seed Persistence: Player can optionally reuse a prior world seed to attempt optimized routing with accumulated meta-knowledge.
• Adaptive Opposition: Post-loss AI squads may recruit or intensify training focus areas, generating emerging rival build paths.

Failure Handling

• Regular Match Loss: Grants reduced XP and minimal chemistry gain (may decay if performance poor); prestige & reputation gains withheld; knowledge still accumulates.
• Tournament Elimination: Ends progression in that bracket for the season; player can pivot to side events, training arcs, or exploration to recover advantage.
• Strategic Recovery Loop: Losses increase probability of encountering scouting events or mentor opportunities offering comeback pathways (balancing morale without trivializing failure).

Session Length & Accessibility

• Flexible Session Design: Core loops (explore tile -> trigger encounter/event OR play short-sided match) support 5–10 minute micro sessions while longer arcs (tournament run + exploration sweep) scale to 45–60 minutes.
• Touch-Friendly Interaction: Action selection panels and exploration movement simplified to tap/drag; success probability & trait effect overlays presented as concise icon stacks.
• Return-Friendly State: Mid-region exploration, pending mentor meeting, or queued tournament round each form natural save points; quick re-entry surfaces “next 3 actionable options” UI.
• Diegetic UI Scaling: High IQ/vision traits widen vignette & highlight pass lanes—also functions as an accessibility aid for clarity on small screens.

Art Style

High-Level Direction

• Core Split: Top-down exploration uses a lightweight, modular sprite system; match view uses stylized manga/comic action panels plus minimal 3D/2.5D staging for positioning and camera context.
• Tone: Energetic, aspirational shonen-sports feel (Captain Tsubasa influence) but restrained enough to allow fast production.

Exploration Layer (Top-Down)

• Base Assets: Pick an existing permissive CC0 or low-cost RPG sprite pack with 3–5 body bases (short, tall, lanky, strong).
• Customization: Implement a JSON-driven layer map (body -> skin tone palette swap -> hair overlay -> wearable overlays -> trait icon).
• Tech: Simple palette swap (nearest-color mapping) plus optional tint masks for team colors.
• Output: Real-time composited sprites (no need for AI generation here initially).

Match Layer (Perspective + Panels)

• Spatial Context: Lightweight 2.5D court (orthographic or shallow perspective) with placeholder capsule rigs for positioning.
• Action Resolution: After player selects an action, generate 2–3 manga panels: (Setup) -> (Clash/Motion) -> (Outcome).
• Panel Styles: Monochrome ink with selective color pops (team color, ball, special effects) to emphasize tactical choices.
• Speed: Cache common poses (dribble, pass, jump, block) as ControlNet pose templates or static silhouette layers.
• Generative Path: Use a stable pipeline (e.g. Stable Diffusion or Flux + LoRA per recurring character for consistent face/hair).
• Fallback: If generation exceeds time budget (e.g. >1.0s), show pre-composited template panels with dynamic text overlays.

Dynamic Character Consistency

• Data Model: Each player has a style seed (hash of name + trait IDs) used for: palette selection, prompt embedding token (e.g. <RIVAL_07>).
• LoRA Strategy: Batch-generate a small set (4–6) of high-value rivals/squad members; generic players use base model + descriptive prompt tags.
• Wearables: Maintain a vector/flat color layer library (SVG or PNG with alpha) that can be composited pre-generation (via ControlNet mask) or post-generation (overlay if alignment is stable).
• Body Types: Provide silhouette masks to upstream model via inpainting to control proportions.
• Versioning: Store generated panel metadata (prompt, seed, RNG) for reproducibility/debug.

Pipeline Phases

1. Prototype (No AI): Hand-assemble panels using layered static art + effect overlays.
2. Hybrid: Introduce AI only for “hero” moments (critical passes, clutch shots).
3. Full: Expand AI panel generation to most actions; refine caching/performance.
4. Enhancement: Add trait-specific visual motifs (e.g. spark trails, aura lines) via post-process shader or overlay library.

Performance & Tooling

• Panel Cache: Key on (actionType, playerTraitCluster, successTier).
• Async Gen: Spawn generation in background; display placeholder composition if not ready; swap in final panel.
• Asset Build Script: Generates body-type palettes and wearable composite previews for QA.

Risk Mitigation

• Consistency Issues: Limit AI usage early; invest in a robust manual base pack.
• Time Sink: Fix a hard budget per action (e.g. 800ms target, fallback at 1000ms).
• Style Drift: Central style guide (brush weight, contrast, color pop rules).

Immediate Next Steps (if you want to move forward)

• Choose base top-down sprite pack candidates.
• Define 5–8 canonical action types (pass, drive, shot, block, steal, special).
• Draft style guide (panel layout grid, font for SFX, color accent rules).
• Decide initial AI vs manual split (e.g. only “shot” and “special” use AI in v0).

Core Gameplay Loop

Overview

The loop layers micro decisions (tile exploration, short matches, training, recruitment) inside region progression toward season domination, all governed by a reproducible procedural seed to enable roguelite variety and meta learning.

Create a Player

Before starting a season, a player has to be created.

Season (Run) Loop

1. Seed Initialization (world, regions, rival archetypes, mentor pools, wearable distribution).
2. Start in first region; pursue prestige and tournament wins to meet domination thresholds.
3. Unlock subsequent regions sequentially; each introduces new buffs, mentors, rivals.
4. Reach global stage OR conclude upon defined end condition (final tournament, time limit, narrative milestone).
5. Season Wrap-Up: Persist meta unlocks (backgrounds, select skills, style seeds) and optionally reuse seed for optimized routing in next run.

Region Progression Sub-Loop

1. Enter region (fog-of-war grid hidden).
2. Explore tiles to reveal courts, gyms, shops, mentors, events.
3. Trigger dynamic events: rival encounter, scouting, item discovery, side match, mentor lead.
4. Accrue prestige, chemistry, knowledge, buffs; pursue tournaments / key courts.
5. Dominate region (threshold of prestige or tournament victory) to unlock travel option.

Exploration Micro-Loop

1. Move to adjacent tile (reveals terrain + rolls event table).
2. Surface actionable options (Match, Training, Recruit, Story, Shop, Skip).
3. Player chooses one; unchosen opportunities may decay or persist per rules.
4. Resolve choice; apply progression deltas (attributes, fatigue, chemistry, knowledge, buffs).
5. Present “Next 3 Suggestions” (context-sensitive recommended actions) for continuity and quick re-entry.

Match Loop (Short-Sided Basketball 1v1/2v2/3v3)

1. Initialize possession with action card set (dribble variants, pass lanes, shot types, specials) derived from attributes, traits, fatigue, chemistry, familiarity, knowledge certainty.
2. Player selects an action; success probability displayed as range (narrows with rival knowledge).
3. Resolve via panel sequence (Setup → Clash/Motion → Outcome) affecting score, fatigue, chemistry, trait triggers.
4. Update stats: XP ticks based on participation in match, knowledge reveal increments, possible unlock of new play blocks at familiarity thresholds.
   • Decision: Have user’s player taking most actions (+ player XP & - team chemistry), or distribute them (+ other players XP & + team chemistry)
5. Alternate possession or chain fast-follow actions (rebounds, steals) until win/loss condition.
6. Post-Match screen: summarize XP, prestige, chemistry delta, rival adaptations queued, suggestions.

Training & Mentorship Loop

1. Select location/mentor (location modifiers: e.g. stamina gain x2, trait synergy boosts).
2. Choose focus (primary attribute, secondary tweak, trait progress, signature move unlock attempt).
3. Resolve with seed-based variance; apply fatigue and attribute XP; possible card / trait unlock.
4. Return to exploration decision point with updated suggestions.

Recruitment & Relationship Loop

1. Initiate dialog with prospective player (local or scouted rival).
2. Negotiate incentives (playing time, prestige trajectory, wearables, chemistry impact).
3. Accept / Decline adjusts roster composition, chemistry baseline, familiarity build timers, potential future play blocks.

Knowledge & Fog Interaction

• Spatial Uncertainty: Exploring removes fog, revealing strategic assets and opportunities.
• Mechanical Uncertainty: Repeated rival encounters narrow stat probability ranges (confidence improves action selection).
• Combined effect: higher clarity yields expanded action card sophistication (e.g. precision pass lanes, advanced set plays).

Chemistry & Familiarity

• Chemistry shifts instantly after impactful plays or match outcomes (+ for teamwork success, − for selfish failures).
• Familiarity accrues per shared possessions and minutes; threshold unlocks advanced play blocks (pick and roll, post-up chains, coordinated specials).

Adaptive Opposition

• Losses or rapid player advancement enqueue rival adaptation jobs (focused training, recruitment, trait development).
• Rival build path changes feed back into future probability ranges and may introduce new rival action cards.

Failure & Recovery

• Regular match loss: reduced rewards (XP dampened, prestige withheld), but knowledge still increases.
• Tournament elimination: bracket progress halts; exploration and side activities offer comeback pathways.
• Recovery Events: Loss boosts odds of mentor encounters, rare wearables, specialty court reveals.

Session Modularity

• Micro decision (tile + choice) ~1–2 minutes.
• Short match 5–10 minutes.
• Region cycle spans multiple sessions; save points at exploration state, pending mentor meeting, queued tournament.
• Resume flow always presents “Next 3 Suggestions” based on last snapshot.

Core Loop Pseudo-Flow

SeedInit()
while !SeasonComplete:
    RegionSelectOrContinue()
    while !RegionDominated:
        tile = ExploreStep()
        events = RollEvents(tile)
        choice = PlayerSelect(events)
        Resolve(choice)        # Match | Training | Recruit | Story | Shop | Skip
        ApplyProgression()
        RivalAdaptationTick()
        OfferNextSuggestions()
    UnlockNextRegion()
SeasonWrapUp(); PersistMetaProgress()

State Model (Simplified)

• Global: Seed, SeasonMeta, RegionList
• Region: FogGrid, Courts[], Gyms[], Shops[], Mentors[], Domination
• Player: Attributes, Traits, Fatigue, Wearables, KnowledgeMap (PlayerID→Reveal%), ChemistryMap, FamiliarityTimers, Prestige
• Squad: Members[], ChemistryAggregate, PlayBlocksUnlocked[], PrestigeAggregate
• Rivals: Archetype, BuildPathHistory[], AdaptationQueue[]
• Buffs: ActiveLocationModifiers[], MentorSessionEffects[]
• ActionCards: Derived each possession from (Attributes, Traits, Fatigue, Chemistry, Familiarity, KnowledgeCertainty, ContextState)

Metrics & Tuning Hooks

• Engagement: Average actions per session (target 3–6 micro decisions).
• Variety: Distinct event types per region (>5 baseline).
• Progress Clarity: Sessions ending with suggestions panel (~100%).
• Replayability: Seed variation (trait distribution, mentors, rivals).
• Fair Challenge: Recovery event rate balanced to avoid rubber-band.

Rationale Alignment

Supports exploration reward (fog), tactical depth (action cards + probabilities), roguelite variability (seed), accessible sessions (modular micro loops), meaningful failure recovery (mentor/item opportunities) and persistent meta growth.

Next Steps

• Prototype: Fog-of-war region reveal + possession action card selection + chemistry/familiarity modifiers.
• Draft style guide (panel grid, SFX font, ink weight, color pop rules) and fallback panel templates.
• Implement seed initialization & serialization (world, player style seeds, rival archetypes).
• Outline mentor & location buff system data schema (JSON layer mapping + training modifiers).
• Begin UX wireframes for touch vs desktop input flows.
• JSON schemas for entities.
• Prototype state machine implementation.