Mito Director: Advanced AI Assistant

Feature Snapshot

Field	Value
Feature name	Mito Director
Primary goal	AI-powered creative assistant for generating images, videos, and breaking down scripts
Availability	All users

Key Facts the Model Must Know

• Mito Director is the AI assistant integrated throughout the platform
• The prompt bar at the bottom of the canvas has three tabs: Image, Video, and AI Director
• Image tab: Generate AI images with model selection, aspect ratio, resolution, count, @references, and /tools
• Video tab: Generate AI videos with model selection, duration, aspect ratio, source frames, multi-shot, and audio
• AI Director tab: Chat with Director for creative guidance, attach files, project-specific conversation history
• Break down scripts into structured storyboard scenes automatically
• Can reference existing canvas assets using @mentions (up to 6 references for images)
• Supports /tools for camera, lighting, and style presets (Image tab only)
• Two prompt modes: Fine-tune (enhanced, default) and Raw (sent as-is)

Prompt Bar Tabs

The prompt bar sits at the bottom center of the canvas with three tabs:

Tab	Icon	Purpose
Image	Image icon	Generate AI images on the canvas
Video	Video icon	Generate AI videos on the canvas
AI Director	Sparkles icon	Chat with Director for creative assistance

Each tab changes the prompt bar's layout, options, and behavior.

Core Capabilities

Image Generation (Image tab)

Generate AI images using the prompt bar:

1. Click the Image tab in the prompt bar
2. Choose a model from the model selector at the top
3. Add image references using @mentions or the reference area (if the model supports them)
4. Write a prompt in the prompt input — use @ for references and / for tools
5. Adjust options: aspect ratio, resolution (1K/2K/4K), image count (1–10)
6. Click generate — cost is shown on the button before submitting

Available aspect ratios: 1:1, 2:3, 3:2, 3:4, 4:3, 5:4, 4:5, 9:16, 16:9, 21:9.

Placeholder text: "Describe an image. Use @ to reference images or groups. Use / to use tools."

Video Generation (Video tab)

Generate AI videos using the prompt bar:

1. Click the Video tab in the prompt bar
2. Choose a video model and duration
3. Set the aspect ratio (model-dependent: 1:1, 3:4, 4:3, 9:16, 16:9, 21:9)
4. Add source inputs: start frame, end frame, video source, or references (model-dependent)
5. Write a prompt — each model has a character limit
6. Toggle audio generation and custom voice if supported
7. Use multi-shot mode for complex scenes (model-dependent)
8. Click generate — cost is shown on the button

Video model capabilities vary. The interface adapts automatically — unsupported options are hidden.

Chat Assistant (AI Director tab)

Provide creative support through conversation:

• Click AI Director tab at the bottom of the prompt bar
• Project-specific conversation history and context
• Canvas awareness and asset understanding
• File upload support via paperclip button (images, videos, documents)
• Copy responses for reuse
• Suggestions appear on first message: "Generate an image", "Create a storyboard", "What can you do?"

Placeholder text on first message: animated rotation of "bring images to life", "craft cinematic videos", "write compelling scripts", "visualize storyboards". After first message: "Ask director".

Script Breakdown (Storyboard)

Turn scripts into scenes automatically:

1. Add script text to canvas
2. Go to Storyboard mode
3. Drag script to Mito Director area
4. Director analyzes and creates scene breakdowns
5. User reviews and edits generated scenes

Asset References and Tools

@mentions (Image tab)

• Type @ in the prompt to reference existing assets
• Autocomplete searches across all project assets
• Click the add button in the reference area to browse project assets
• Double-click a reference thumbnail to insert it as a mention in the prompt
• References appear as thumbnails in the reference area (max depends on model, up to 6)
• Some models don't support references — the reference area and mentions are hidden automatically
• Per-model constraints: max file size (MB), min resolution (pixels)

/tools (Image tab)

Type / to access creative parameter tools:

• Inspiration, lens, shot type, angle, lighting, light source, film stock, camera movement, depth of field, motion

Video model capabilities

Each video model exposes different capabilities that control what options appear:

• Start/end frame: Provide images for the first or last frame
• Image/video references: Add reference media for style guidance
• Multi-shot: Write separate prompts per shot segment
• Audio generation: Generate a matching audio track
• Custom voice: Choose from available voice options
• Video editing: Edit or transform existing videos
• Motion video/image: Motion-based generation modes

Home Page Prompt Bar

The home/landing page has its own prompt bar with:

• Animated rotating placeholder text: "Ask MITO to create a cinematic scene / a product mockup / a stunning portrait / an animated video"
• Paperclip button to attach image files as references
• @mention support for attached images
• Submitting creates a new project and generates content in one step

Prompt Modes

• Fine-tune (default): Prompt is enhanced by AI for better results
• Raw: Prompt is sent exactly as written
• Preference is saved in localStorage and persists across sessions
• Available in Image and Video tabs, not in AI Director chat

Cost and Credits

• Generate button shows credit cost before submitting
• Image cost = model cost per image × image count
• Video cost = model cost per second × duration
• Some capabilities (references, multi-shot) add incremental costs
• Insufficient credits: button turns red, links to billing page

High Demand

• Orange banner appears when a model is under heavy load
• Shows estimated wait time in minutes

Supported Scenarios

• Help users generate their first image or video
• Guide through script breakdown process
• Assist with writing effective prompts
• Explain @mentions, /tools, and prompt modes
• Explain model differences and options
• Troubleshoot generation issues

Example Conversations

Image Generation

User: How do I generate an image with Director?
Assistant: Use the prompt bar at the bottom of your canvas. Click the Image tab, type a description of what you want, choose a model, and adjust options like aspect ratio and image count. Click generate—the credit cost is shown on the button. Your images appear on the canvas once ready. You can reference existing assets by typing @ in the prompt.

Video Generation on Canvas

User: Can I generate videos on the canvas too?
Assistant: Yes! Click the Video tab in the prompt bar at the bottom of the canvas. Choose a video model and duration, add a start frame if needed, write a prompt, and click generate. Each model has different capabilities — some support multi-shot prompts, audio generation, and start/end frames. The interface adapts to show only the options supported by the selected model.

Asset Reference Guidance

User: How do I make this new character match the style of my moodboard?
Assistant: Reference your moodboard assets in your prompt using @ notation. For example:

"Portrait of [character description] in the style of @2"

This tells Director to analyze asset @2's artistic style—color palette, lighting, composition—and apply it to your new character. You can reference multiple assets: "Using the lighting from @1 and color scheme from @3"

Which moodboard assets best represent the style you want?

Using Asset References

User: How do I use my moodboard images when generating?
Assistant: Type @ in the prompt to search across all project assets, or click the thumbnails below the prompt bar, or drag assets from the canvas into the prompt area. References appear as @1, @2, @3 and help guide the style and content of your generation. You can add up to 6 references and drag them to reorder.

Using Tools

User: What are the / tools in the prompt?
Assistant: Type / in the prompt bar to open the tools menu. You'll find creative presets for camera lens, shot type, angle, lighting, film stock, and more. Selecting a tool option inserts it into your prompt automatically, helping you get professional-looking results without memorizing technical terms.

Common Issues & Solutions

Problem	Solution
Prompt bar not visible	It's at the bottom center of the canvas — make sure you're in Pre-Production (Canvas) mode
Image doesn't match prompt	Add more specific details, reference existing assets, or try a different model
Video prompt too long	Each video model has a character limit — shorten your prompt or use a model with a higher limit
@mentions not working	Some models don't support references — try a different model
Script not recognized	Make sure it's a text asset on the canvas
Chat not responding	Check internet connection, try refreshing
High demand banner showing	The model is busy — you can wait or try a different model

Related Resources

• User guide: docs/user/mito-director/overview.md
• User guide: docs/user/mito-director/guides/generating-images.md
• User guide: docs/user/mito-director/guides/generating-videos.md
• User guide: docs/user/mito-director/guides/using-the-chat.md
• User guide: docs/user/mito-director/guides/prompt-modes.md
• User guide: docs/user/mito-director/guides/referencing-assets.md
• User guide: docs/user/mito-director/guides/script-breakdown.md

Update Log

• 2024-12-02 — Initial documentation for Mito Director feature
• 2025-01-28 — Added advanced features: prompt modes, asset references, conversation context
• 2025-03-09 — Rewritten for new prompt bar UI (Image/Video tabs, @mentions, /tools, prompt modes, cost display)
• 2025-03-15 — Updated for three-tab prompt bar (Image/Video/AI Director), added video model capabilities, per-model constraints, placeholder text details, and incremental costs