Creative writing is a juggling act. Authors must manage an intricate web of character arcs, plot points, locations, and timelines. Maintaining consistency across all these interconnected elements is a monumental task—especially when experimenting with new ideas. A seemingly small change, like moving a character to a different location, can trigger a cascade of edits, forcing the writer to hunt down every related sentence to preserve narrative coherence.

To cope, many writers build external aids—scribbled timelines, relationship charts, or spreadsheets—to track their story worlds. But these tools are disconnected from the actual text. What if you could bridge this gap? What if you could rearrange a timeline or move a character on a map and watch those changes update your manuscript automatically?

This is the central idea behind a recent paper from researchers at the Université de Montréal and the University of Toronto. They introduce visual story-writing—a new paradigm for writing support that turns story elements into interactive visualizations. Instead of only writing with words, you can now write by manipulating visuals.

An infographic showing the visual story-writing workflow. A narrative text is used to generate visuals (Interactions, Locations, Timeline), and manipulating these visuals suggests edits back to the text.

As illustrated above, the system parses a narrative text to generate interactive graphs of character interactions, spatial maps of locations, and a timeline of events. The magic happens when a writer engages with these visuals: connecting two characters in the graph can create a new scene where they interact; dragging a character icon to a different spot on the map rewrites the text to narrate their new journey. By aligning the writer’s reasoning process with visual representations, story revision becomes intuitive, playful, and far less error-prone—laying a foundation for creativity supported by both words and visuals.

Background: From Analysis to Interaction

Visualizing stories is not a new practice. For decades, writers and literary scholars have used diagrams to understand narrative structure—for example, Freytag’s pyramid, which charts the dramatic arc, or popular “storyline” visualizations mapping character relationships over time. These tools are valuable for analysis, offering a bird’s-eye view of completed narratives. Yet, they are typically static outputs. You can look at them, but you can’t use them to actively rewrite the story.

Other fields—like computer programming—have embraced alternative representations for editing. Developers often use color pickers instead of typing hexadecimal codes, or drag-and-drop UI builders instead of handcrafting layout code. The principle is simple: some tasks are easier with spatial or visual manipulation than with text alone.

The authors argue that creative writing can benefit from this same principle. While some existing tools have explored visual interfaces for tasks like outlining or generating text from images, the concept of a fully bi-directional system—where text informs visuals and visuals edit text—has remained largely unexplored. This work formalizes visual story-writing and presents a theoretical framework for building such tools.

A Framework for Visualizing Stories

To create a system that can visually represent any story, you need a formal description of what makes up a story. The researchers turned to narratology—the study of narrative structure—to design a framework for story visualizations. This framework identifies fundamental story elements and defines operations to combine them into complex, meaningful visual constructs.

Story Elements: The Building Blocks

Narrative theory distinguishes between the fabula (the raw chronological sequence of events in the fictional world) and the syuzhet (the way those events are presented to the reader, potentially including flashbacks, foreshadowing, or narrative gaps). This distinction yields eight fundamental elements:

A table listing the eight story elements, categorized under Fabula (Actor, Location, Time, Event) and Syuzhet (Character, Space, Temporality, Focalization).

For example:

  • “Alice” is a Character (syuzhet) who fulfills the role of Actor (fabula), such as the hero.
  • Time refers to the chronological timeline, while Temporality captures the narrated order, which may skip or reorder events.
  • Location is the physical place (“Alice’s house”), while Space refers to the narrated setting or mood (“home,” “eerie”).

This structured vocabulary lets the system reason about different layers of narrative.

Operators: Building Visual Constructs

With these building blocks, the framework defines four operators—the PACU model—that combine elements to form visualizations.

An infographic illustrating the PACU framework with four operators: Position, Associate, Connect, and Unfold, showing how they can be chained to build complex story visualizations.

  1. POSITION (P): Place elements according to another element.
    Example: POSITION(Characters BY Locations) maps characters onto their locations.
  2. ASSOCIATE (A): Link elements with additional metadata.
    Example: ASSOCIATE(Time WITH Focalization) creates a timeline showing POV shifts.
  3. CONNECT (C): Draw edges between elements based on another element.
    Example: CONNECT(Characters BY Events) generates an interaction graph.
  4. UNFOLD (U): Duplicate and organize elements based on another element.
    Example: UNFOLD(Locations BY Characters) lists locations visited by each character.

By chaining these simple operators, the framework can describe existing visualizations and generate new ones. For instance, the famous xkcd “Storyline” visualization is: start with Time, UNFOLD by Characters, CONNECT by Events.

The Visual Story-Writing Prototype

Grounded in this framework, the researchers developed a prototype system featuring a text editor alongside three interconnected visual panels, automatically generated from Alice’s Adventures in Wonderland.

The user interface of the visual story-writing prototype. It shows a text editor on the left and three visual panels on the right: an event timeline, an actions and entities graph, and a locations map.

1. Entities & Actions View

This view shows characters and key objects (“entities”) as nodes, with directed edges labeled by actions (e.g., Alice → run after → White Rabbit).

Editing options:

  • Edit Traits: Click “Alice” to adjust traits like “curious” or “sleepy”; the AI rewrites the story to reflect these changes.
  • Add/Remove Entities: Double-click to create new characters; delete nodes to remove them.
  • Add/Remove Actions: Connect two entities to add a new action or delete existing ones to remove an interaction.

Three examples of editing in the entity view: (a) adjusting sliders for Alice’s traits, (b) deleting the ‘Book’ entity, and (c) adding a new ‘Brother’ entity and connecting him to Alice and her Sister.

2. Locations & Entities View

A spatial representation of the story world showing location nodes and where each character resides.

Functions:

  • Add Location: Double-click to create a location (“The Field”).
  • Move Entity: Drag a character icon to a different location; the text updates to reflect the move.

A three-panel sequence showing how to move entities in the location view. First a new location ‘field’ is created, then the ‘book’ is moved to it, and finally the ‘sister’ is moved to the field.

3. Timeline of Events View

A temporal view showing events in the narrated order, with emoji markers for involved entities.

Features:

  • Find Scenes: Hover over an event to highlight corresponding sentences in the text.
  • Targeted Edits: Select events to scope visual edits to those specific scenes.
  • Reorder Events: Drag events to a new position; the system restructures the text accordingly.

The timeline view in action. A user can hover over an event to highlight text, or drag an event to reposition it earlier or later in the story, which reorganizes the narrative.

All edits are bi-directional: visual changes update the text, and manual text edits sync back to visuals. A history tree tracks changes, allowing undo/redo and branching exploration.

Putting It to the Test: User Studies

The team ran two studies to evaluate the approach.

Study 1: Planning & Reviewing

Participants answered high-level revision questions (e.g., “Could any characters be combined?”) using either a text-only interface or a read-only version of the prototype.

Results:

  • Confirm Intuitions: Visuals provided reassurance and boosted confidence.
  • Quick Search: Clicking an entity or event replaced lengthy skimming.
  • New Insights: Location and timeline views prompted reflection on character motivation and spatial logic.

Some found visuals added cognitive load or didn’t match their creative habits, but most valued the visual complement to text.

Study 2: Editing & Creative Exploration

Eight experienced writers used the full prototype for structured edits and free-form writing.

Bar charts showing participants’ positive ratings for the entity, location, and event timeline views across statements like ’easy to understand’ and ’easy to use’.

Findings:

  • Effortless Editing: Successful manipulation of characters, locations, and events; AI handled inconsistencies automatically.
  • Exploration Support: Writers used visuals to brainstorm, reorder events, and spark inspiration.
  • High Creativity Support: Creativity Support Index (CSI) average score: 71.5, with high ratings for Enjoyment and Results Worth Effort.

Table showing the results of the Creativity Support Index (CSI). ‘Enjoyment’ and ‘Results Worth Effort’ received the highest scores.

Critiques included a desire for more control over rewrites and alternative visualization styles. Yet engagement remained high, suggesting the approach’s adaptability.

Conclusion & Future Directions

This work demonstrates that visualizations can be interactive mediums for creating stories, not just analytical aids. By defining a framework of story elements and operators and implementing a bi-directional prototype, visual story-writing helps track elements, express complex edits through simple interactions, and explore creative variations playfully.

Future expansions could incorporate:

  • Abstract elements like emotions or motivations.
  • Style editing through visual controls for sentence length or tone.
  • Custom visualization builders tailored to individual writers’ workflows.

Ultimately, visual story-writing is about augmenting creativity—empowering writers with tools that match how we imagine narratives: as a rich, interconnected web of characters, places, and events, not just as a line of text.