Orchid-Creator: An Authoring Tool Supporting LLM-Driven Interactive Narrative Creation

Human-LLM CollaborationInteractive Narrative & Immersive StorytellingAI-Assisted Creative WritingCreative Collaboration & Feedback SystemsGame Developers & DesignersContent Creators (YouTubers, Podcasters)AI/ML Researchers & Engineers

Paper Title

Orchid-Creator: An Authoring Tool Supporting LLM-Driven Interactive Narrative Creation

Publication Info

  • Topic area: Interactive Digital Narratives (IDNs) and LLM-driven authoring tools
  • Keywords: Interactive Digital Narratives, Large Language Models, Authoring Tools, Story Graphs, Player Co-Creation, Narrative Structuring, AI Creativity, IDN Testing, Branching Structures

Background and Problem

  • Problem / challenge: Existing tools for creating Interactive Digital Narratives (IDNs) struggle with balancing narrative complexity, authorial control, and player interactivity. LLM-driven systems often lack support for complex branching mechanisms, leading to narrative drift and undermining authorial intent.
  • Significance: Addressing these challenges is critical for enabling richer, more coherent, and interactive storytelling experiences that leverage the creative potential of LLMs while preserving authorial agency.
  • Motivation and related work: Prior tools like Twine and AI Dungeon either focus on static branching or open-ended LLM-driven narratives but fail to integrate both effectively. Existing research highlights the need for tools that combine structured planning with LLM creativity, but many remain conceptual frameworks without practical implementation or evaluation.

Solution

  • Proposed approach: Orchid-Creator, a web-based authoring tool that uses nodes, graphs, and LLMs to design and test branching IDNs. It employs a card-based interface for defining narrative elements and a story graph for structuring transitions.
  • Novelty:
    1. Combines LLM-driven generation with structured branching using nodes and graphs.
    2. Introduces Global Variables for precise control over narrative transitions.
    3. Supports real-time testing and iterative design with retrospective debugging.
    4. Encourages player co-creation by integrating diverse input types (text, number, boolean).
  • Procedure and key techniques:
    1. Phase 1: Define narrative components (world settings, characters, events, inputs, global variables) using cards.
    2. Phase 2: Assemble a story graph by connecting nodes and defining transitions based on Global Variables.
    3. Phase 3: Test the IDN by simulating player inputs, observing transitions, and debugging paths.

Results

  • Concrete findings:
    • Usability ratings: Card interface (4.0/5), story graph (4.38/5), variable setup (4.5/5).
    • Structuring narratives was rated easier (M = 6.0/7, p < .01) and produced better-structured stories (M = 5.3/7, p < .05) compared to Twine and AI Dungeon.
    • Balanced author control (M = 5.5/7) with outcome diversity (M = 5/7, p < .01).
  • Advantage over baselines:
    • Outperformed Twine in idea exploration (M = 5.53, p = .0111) and interactivity authoring (p = .0021).
    • Outperformed AI Dungeon in structural experimentation (p = .0103) and narrative coherence.
  • Experiments / evaluation:
    • Study 1: Usability study with 8 experienced IDN creators who created diverse narrative structures and highlighted the tool’s strengths in narrative decomposition and iterative testing.
    • Study 2: Comparative study with 20 participants showing Orchid-Creator’s superiority in creative support, authorial agency, and usability compared to Twine and AI Dungeon.
    • Study 3: Case study with a narrative artist who used Orchid-Creator to create a six-stage IDN, demonstrating its practical application and player engagement.
  • Limitations and future work:
    • LLM limitations: Coherence issues, generation speed, and memory constraints.
    • Limited support for multimedia and physical interaction inputs.
    • Need for integration with game engines and genre-specific optimizations.

Summary

Orchid-Creator is an innovative authoring tool that integrates LLM-driven narrative generation with structured branching using nodes and graphs. It enables authors to design complex IDNs while balancing authorial control and player co-creation. Empirical studies demonstrate its advantages over existing tools like Twine and AI Dungeon in facilitating narrative structuring, interactivity, and creative exploration. A case study highlights its practical utility and adaptability in artistic projects. While challenges remain in LLM coherence and multimedia integration, Orchid-Creator represents a significant advancement in interactive narrative design, fostering richer and more dynamic storytelling experiences.

Quick Actions

Share

Share this page

ios_share

https://hci.top/en/papers/chi/223280/2026

AdRecommended

Learn AI Coding at CodeNow

open_in_newOpen DOI Link
DOI: https://doi.org/10.1145/3772318.3791426
At a Glance

Paper Snapshot

fact_check
dataset
Source
CHI
calendar_month
Year
2026
emoji_events
Award
No award tagged
group
Authors
4 authors
sell
Subtopics
Human-LLM Collaboration, Interactive Narrative & Immersive Storytelling, AI-Assisted Creative Writing, Creative Collaboration & Feedback Systems
work
Professions
Game Developers & Designers, Content Creators (YouTubers, Podcasters), AI/ML Researchers & Engineers
article
Content Status
Full text indexed
hub
Related Papers
0 related papers