The Editor

Dialogue

YarnSaga has a built-in dialogue system that extracts spoken lines from your scene descriptions and stores them separately. Dialogue is never sent to the image AI — keeping it out of the illustration prompt and preserving your characters' words for display and export.

How dialogue extraction works

When you write a scene description and click Generate, the AI validator reads the text. If it detects dialogue — quoted speech, character lines, or spoken content — it extracts it before the image prompt is built.

You'll see a confirmation step: the drawer shows the cleaned scene (description without the dialogue) and the extracted dialogue lines separately. You can:

  • Confirm — save both the cleaned scene and the extracted dialogue, then generate the image from the cleaned scene
  • Dismiss — keep the original scene text as-is and generate without extracting

Why dialogue is kept separate

Including spoken lines in the image prompt confuses the image model — it tries to render text literally or distorts character expressions trying to "show" what they're saying. Keeping dialogue out of the image prompt produces cleaner, more accurate panel illustrations.

The extracted dialogue is stored in panel.dialogue in the database, available for display, export, and future use without affecting image generation.

Writing dialogue in scene descriptions

You can write dialogue naturally in your scene description — quotation marks, speaker labels, or conversational format. The AI will detect and extract it.

Scene description with dialogue

Elena faces Captain Volt across the burning bridge. She says, "You were supposed to protect them." He looks away. "I know," he replies quietly.

After extraction, the scene sent to the image AI becomes:

Cleaned scene (what the image AI receives)

Elena faces Captain Volt across the burning bridge. She confronts him. He looks away.

And the dialogue is stored separately as:

Extracted dialogue (stored in panel)

Elena: "You were supposed to protect them."
Captain Volt: "I know."

Dialogue from AI narrative generation

When you use the AI Story Generation feature to generate panel scenes from a full narrative, dialogue is extracted automatically for each panel — no confirmation step needed. The AI returns both scene and dialogue as separate fields simultaneously.

Adding speech bubbles manually

Dialogue stored in a panel is not automatically rendered as speech bubbles on the canvas. YarnSaga keeps this manual so you have full creative control over layout, positioning, and style.

To display your character's words in the panel:

  1. Place a speech bubble from the left toolbar (see Speech Bubbles & FX)
  2. Double-click the bubble and type (or paste) the dialogue line
  3. Position and style the bubble to point at the correct character

The stored dialogue in the panel serves as your reference — you can see it in the Scene drawer while placing bubbles.

Keep scene and dialogue separate from the start

For the cleanest workflow, write your scene description as pure visual direction — setting, action, mood, camera angle — and keep the spoken words for after the image is generated. Then add speech bubbles manually once you're happy with the illustration.