Skip to content

AI Creation Studio

The AI Creation Studio is a dedicated workspace for generating complete character cards from scratch using AI. Describe your concept or select tags, and the AI generates a name, description, first message, and example dialogue — all in one flow.

TIP

AI Creation Studio requires an AI provider to be configured. See AI Setup to get connected first.

Opening the Studio

Access the AI Creation Studio from the main vault view:

  1. Click the ✨ AI Creation Studio button in the vault header.
  2. The studio opens in a full-screen workspace with a clean, focused interface.

Input Modes

The studio supports two ways to define your character concept:

Write Mode

Type a free-form description of your character idea. The AI uses your text as the creative foundation.

Example concepts:

  • "A cynical dwarven blacksmith with a secret past, living in a mountain fortress who speaks in riddles"
  • "A cheerful android barista who dreams of becoming human"
  • "A mysterious vampire librarian who collects forbidden knowledge"

Requirements:

  • Minimum 3 words
  • Choose a generation style before generating
  • The more detail you provide, the better the AI can capture your vision

Tags Mode

Select from curated tag categories to build your concept visually. The Generation category appears first, followed by character tags:

CategoryPurposeExamples
GenerationWriting perspective and tenseFirst Person, Third Person, Present Tense
IdentityCore identity traitsFemale, Male, Elf, Android, Vampire
PersonalityCharacter traitsTsundere, Yandere, Kuudere, Cheerful, Cynical
RoleRelationship or occupationTeacher, Knight, Barista, Mother, Rival
GenreSetting or themeFantasy, Sci-Fi, Modern, Historical
ToneInteraction styleRomantic, Dark, Comedic, Wholesome
AppearancePhysical traitsTall, Short, Muscular, Petite

Tag Selection:

  • Click tags to select or deselect them
  • Select multiple tags per category
  • Choose one perspective and one tense from Generation
  • Tags automatically convert to a readable concept (e.g., female_knight → "Female, Knight")
  • Conflicting tags are automatically excluded (e.g., selecting "female" disables "male" and related tags)

Switching Modes

When switching from Tags to Write mode, your selected tags are automatically converted to text and populate the concept field. You can then edit or expand on them.

Generation Style

Every character needs a generation style before the AI can start. Choose:

  • Perspective — first person, second person, third person, or first person with
  • Tense — present tense or past tense

Only one perspective and one tense can be selected at a time.

Descriptions are written like character-card reference material. If you choose First Person with , descriptions use instead of talking directly to "you." First messages and example dialogue still follow the style you selected.

Feeling Lucky

The 🎲 Feeling Lucky button (Tags mode only) randomly selects character tags and starts generation immediately. Choose a generation style first.

How it works:

  • Core categories (Identity, Role, Personality): 1–2 tags each
  • Supporting categories (Genre, Appearance, Tone): 0–2 tags each
  • Respects tag exclusion rules (no conflicting combinations)
  • Shows a visual "tag vortex" animation before generation starts (can be disabled in Settings)

Disabling the Vortex:

  1. Open Settings from the studio header
  2. Go to the UI tab
  3. Toggle Show Lucky Vortex off
  4. When disabled, Feeling Lucky generates immediately without the animation

Generation Process

When you click Generate Character, the AI creates four fields in sequence:

  1. Name — Character's name
  2. Description — Detailed character description
  3. First Message — Opening greeting
  4. Example Dialogue — Sample conversation demonstrating personality and speech patterns

During generation:

  • Each field streams in real-time as the AI writes
  • A progress indicator shows which field is currently generating
  • You can Stop generation at any time
  • The concept or tags used are displayed for reference

After generation:

  • Review the generated content in the preview panel
  • Edit any field directly in the preview
  • Retry individual fields if you're not satisfied
  • Regenerate fields to get completely new content
  • Continue generation if it was stopped early

Preview & Editing

The right panel shows a live preview of your character card as it's generated.

Preview features:

  • Real-time updates as fields are generated
  • Click any field to edit it directly
  • Changes are saved automatically
  • Character count shown for each field

Field Actions:

Each completed field has action buttons:

ButtonActionWhat It Does
🔄 RetryRetryRe-generate the field if generation failed
RegenerateRegenerateGenerate completely new content for this field

TIP

Retry is for fixing errors. Regenerate is for getting a fresh take on a field you don't like.

Saving to Vault

Once generation is complete (or partially complete with at least a name), you can save the character:

  1. Click Save to Vault in the bottom-right corner
  2. The character is added to your vault immediately
  3. A success message appears with three options:
    • Open Character — Jump to the editor to continue working
    • Create Another — Start a new character from scratch
    • Library — Return to the vault view

Partial Saves

You can save a character even if not all fields are generated, as long as it has a name. Missing fields will be empty in the editor.

Generation Settings

The studio uses your configured AI provider and sampler settings. To adjust:

  1. Click Settings in the studio header
  2. Configure your AI provider, model, and sampler settings
  3. Changes apply immediately to new generations

See AI Setup and Sampler Settings for details.

API Usage

Each character generation uses a minimum of 4 API calls — one per field (Name, Description, First Message, Example Dialogue).

Additional calls occur when:

  • Retrying a failed field
  • Regenerating a field
  • Continuing a stopped generation

Cost Awareness

Generation can consume significant tokens depending on your model and sampler settings. Monitor your API usage if you're on a metered plan.

Tips for Best Results

Write Mode:

  • Be specific about personality, setting, and tone
  • Include unique quirks or speech patterns
  • Mention relationships or background if relevant

Tags Mode:

  • Mix categories for richer characters (e.g., Identity + Personality + Genre)
  • Use Appearance and Tone tags to fine-tune the vibe
  • Experiment with Feeling Lucky for unexpected combinations

After Generation:

  • Edit fields directly in the preview to refine the output
  • Regenerate individual fields if they don't match your vision
  • Save early and continue editing in the main editor for advanced features

Troubleshooting

"AI Provider Not Configured"

  • You need to set up your AI provider first
  • Click Configure AI in the warning banner
  • See AI Setup for instructions

Generation fails or returns empty content

  • Check your API key and model configuration
  • Verify your model supports chat completions
  • Try a different model or adjust sampler settings

Generation is too slow

  • Reduce max_tokens in sampler settings
  • Use a faster model
  • Check your network connection

Fields don't match my concept

  • Try regenerating individual fields
  • Edit the concept to be more specific
  • Manually edit fields in the preview

Next Steps

Privacy