Ania AI Storying Tech Paper
Ania AI Storying: A Browser-Based Rich-media Narrative Experience with Decision Analysis
Authors: Chris Harvey, Ania Temasek
Abstract
Ania AI Storying is a browser-based narrative game that integrates rich-media storytelling with interactive decision-making. Stories are divided into scenes, each presented through a song (hosted on streaming platforms, and used as video soundtracks) and a video with narration (hosted on platforms such as YouTube). Players predict the protagonist's next action via multiple-choice questions on the ania.ai website, using contextual clues, a character profile, and decision feedback. This white paper outlines the logic, design, and technical framework, emphasizing external asset hosting and browser-based logic.
1. Introduction
Ania AI Storying blends passive rich-media consumption with active decision analysis, leveraging external platforms for content delivery and ania.ai for game logic. It:
-
Delivers songs via streaming platforms and videos with narration via hosting platforms.
-
Hosts game logic, profiles, and choices on ania.ai.
-
Engages players in predicting protagonist actions using clues and feedback. The game targets fans of narrative-driven experiences, accessible via modern browsers.
2. Core Concept
Ania AI Storying divides a story into scenes, each comprising:
-
Rich-media Presentation:
-
Song: Streamed from platforms (e.g., Spotify, SoundCloud), embedded as the video soundtrack.
-
Video with Narration: 2D animated, hosted on YouTube, with voiceover narration, embedded on ania.ai.
-
-
Decision Point: A multiple-choice question (“What does the protagonist do next?”) with a number of options, hosted on ania.ai.
-
Analysis Mechanics:
-
Contextual Clues: Embedded in song lyrics, video visuals, and narration.
-
Character Profile: A dynamic record of traits and decisions on ania.ai.
-
Decision Feedback: Post-choice analysis on ania.ai showing alignment and outcomes.
-
Objective: Players analyse the protagonist's decisions to predict actions, earning points or unlocking story branches, with seamless integration of external assets and ania.ai logic.
3. Game Mechanics
3.1 Rich-media Clues (Option 1)
-
Purpose: Provide hints about the protagonist's motivations via song, video, and narration.
-
Implementation:
-
Song: Streamed via Spotify/SoundCloud API, embedded in YouTube video as soundtrack. Lyrics encode cues (e.g., “fearful steps” for caution). Tense tone suggests danger.
-
Video with Narration: YouTube-hosted MP4, embedded on ania.ai. Video animation shows body language (e.g., trembling hands) and environment (e.g., glowing eyes). Narration voiceover reveals thoughts (e.g., “mentor's warning”).
-
Pre-Choice Prompt: On ania.ai, JavaScript-driven question (e.g., “What emotion drives the protagonist?”) unlocks hints if correct.
-
-
Logic: Clues align with protagonist traits, balancing clarity and ambiguity to encourage analysis.
3.2 Character Profile or Journal (Option 2)
-
Purpose: Track protagonist traits and decisions to aid prediction.
-
Implementation:
-
Profile Structure: JSON object on ania.ai with attributes (e.g., “Caution: High”) and journal entries (e.g., “I can't trust the forest”).
-
Dynamic Updates: JavaScript updates profile after scenes (e.g., “Chose to hide, Caution +10”).
-
Player Interaction: HTML/CSS interface on ania.ai lets players select a trait before choosing (e.g., “Caution”).
-
-
Logic: The profile scaffolds reasoning, tracking the protagonist's arc to guide choices.
3.3 Decision Breakdown and Feedback (Option 3)
-
Purpose: Validate choices with feedback, deepening decision understanding.
-
Implementation:
-
Immediate Feedback: HTML text on ania.ai explains alignment (e.g., “Hiding fits Lila's caution, seen in her trembling hand”).
-
Consistency Score: JavaScript calculates score (e.g., 90% for canonical choices).
-
Narrative Consequence: Text or brief YouTube clip shows outcome (e.g., “Creature retreats”).
-
Analysis Mode: Replay YouTube video/song on ania.ai with CSS annotations (e.g., text overlays) and a reflection question (“Why this choice?”).
-
Community Stats: Fetch player choices via ania.ai backend API (e.g., “65% chose 'Hide'”).
-
-
Logic: Feedback connects choices to clues, encouraging refined analysis. Analysis Mode highlights missed hints.
3.4 Integrated Flow
-
Scene Presentation: On ania.ai, stream song and embed YouTube video with narration.
-
Analysis Prompt: On ania.ai, “What emotion drives the protagonist?” (Unlocks hint).
-
Profile Review: Select trait on ania.ai.
-
Choice: Pick from 4 options on ania.ai.
-
Feedback: Show alignment, score, outcome on ania.ai.
-
Analysis Mode: Replay video/song with annotations, answer reflection question.
-
Profile Update: Save choice on ania.ai.
-
Progress: Load next scene on ania.ai.
4. Technical Architecture
4.1 Platform
-
Target: Web browser (Chrome, Firefox, Safari, Edge).
-
Framework: JavaScript, HTML5 for text/imagery, Custom UI.
-
Backend: Node.js with MongoDB for player data (profiles, choices), analytics, and community stats.
4.2 Rich-media Delivery
-
Song: Streamed, incorporated in YouTube video.
-
Video with Narration: YouTube-hosted MP4s, embedded on ania.ai. Narration as voiceover, synced with visuals.
-
Storage: YouTube for videos, Sudo for songs, MongoDB Storage for metadata or fallback assets.
4.3 Decision System
-
Data Structure:
-
Scene Object: {ID, Song_URL, Video_URL, Clues[], Choices[]}.
-
Choice Object: {ID, Text, Trait_Alignment, Outcome_Text, Consequence_ID}.
-
Profile Object: {Traits{Cautious, Curious}, Journal[], Past_Choices[]}.
-
-
Logic Flow:
-
Load scene → Stream multimedia → Prompt analysis → Update profile → Present choices → Process choice → Deliver feedback → Save state.
-
-
Branching: Limited branching per scene, converging at key scenes. JavaScript state machine tracks paths.
4.4 Analysis Mechanics
-
Clue Integration: Tag clues in assets (e.g., lyrics with “fear,” video frames with “trembling hand”). JavaScript parser matches clues to choices.
-
Profile Updates: Rule-based (e.g., “If choice = Hide, Caution +10”). Stored in Firebase Realtime Database.
-
Feedback Algorithm:
-
Consistency score: Sum(Trait_Alignment * Clue_Matches) / Max_Score.
-
Feedback text: Template-based (e.g., “You chose [Choice], aligning with [Trait] due to [Clue]”).
-
Analysis Mode: Replay YouTube video with CSS text overlays.
-
5. Design Rationale
5.1 Why Rich-media?
-
Immersion: Songs and videos with narration create a cinematic feel.
-
External Hosting: Reduce ania.ai's storage load, leveraging reliable platforms.
-
Clue Variety: Lyrics, visuals, and narration offer diverse hints, enriching analysis.
5.2 Why Decision Analysis?
-
Engagement: Predicting actions makes players active participants.
-
Critical Thinking: Analysing clues and traits fosters inference, appealing to narrative fans.
-
Replayability: Feedback and profiles encourage revisiting scenes.
5.3 Why Limited Branching?
-
Feasibility: Full branching is resource-heavy. Limited branches balance variety and scope.
-
Cohesion: Converging paths maintain narrative focus.
5.4 Why External Platforms?
-
Scalability: Streaming and hosting handle high traffic, reducing ania.ai's server load.
-
Ease of Updates: New songs/videos can be uploaded to platforms without changing ania.ai's core logic.
-
User Familiarity: Players know YouTube/Suno, easing adoption.
6. Challenges and Mitigations
-
Platform Dependency:
-
Challenge: YouTube/Suno outages or API changes could disrupt assets.
-
Mitigation: Cache assets locally, maintain fallback MP3/MP4s on MongoDB Storage.
-
-
Production Complexity:
-
Challenge: Creating songs and videos is resource-intensive.
-
Mitigation: Use AI tools (e.g., Suno for music, Runway for video) for prototyping and production.
-
-
Clue Ambiguity:
-
Challenge: Clues must guide without being obvious.
-
Mitigation: Test with players, adjust clue prominence (e.g., clearer visuals, louder narration).
-
7. Potential Enhancements
-
Dynamic Audio: Adjust song intensity via JavaScript based on choices.
-
Lore Collectibles: Clickable video objects (YouTube annotations or ania.ai overlays) unlock codex entries.
-
Social Features: Share choices via X, Instagram, TikTok.
-
AI Personalization: Adapt clues based on player behaviour.
8. Implementation Roadmap
-
Prototype:
-
Develop 1 scene (song on Suno, video with narration on YouTube, logic on ania.ai).
-
Use AI-generated assets.
-
Implement clues, profile, feedback for 1 path.
-
Test with 10-20 players.
-
-
Launch:
-
Polish assets, fix bugs.
-
Deploy via Firebase Hosting on ania.ai.
-
Market via X, Instagram, TikTok and narrative game forums.
-
9. Conclusion
Ania AI Storying redefines browser-based narrative gaming by integrating externally hosted multimedia (Sudo songs, YouTube videos with narration) with decision analysis on ania.ai. Its clues, profiles, and feedback create an engaging, replayable experience. A focused prototype can evolve into a unique game.