BMAD-METHOD/docs/tools/nanobanana.md

241 lines
5.9 KiB
Markdown

# NanoBanana
**Category:** AI Design Tool
**Purpose:** Agent-driven asset creation and design inspiration
**Website:** <https://nanobanana.com>
---
## What It Is
NanoBanana is an AI-powered design tool (think "agent-driven Photoshop") that creates visual design assets and generates design inspiration. It's used in WDS for:
1. **Early design exploration** - Creating custom graphics, mood boards, visual concepts
2. **Sketch envisioning** - Converting sketches/specifications into visual designs (images or code)
3. **Asset creation** - Generating placeholder assets and custom graphics
**Output formats:**
- **Images** - Visual designs, graphics, illustrations
- **Code snippets** - HTML/CSS/React code for designs
---
## Why WDS Recommends It
**Asset Creation:**
- Generate custom graphics and icons
- Create design inspiration and variations
- Explore visual concepts
- Generate placeholder assets
- Brand identity exploration
**Design Exploration:**
- Multiple design variations quickly
- Explore different visual directions
- Generate creative ideas
- Inspire design decisions
---
## Setup Instructions
### 1. Account Creation
1. Go to <https://nanobanana.com>
2. Sign up for account
3. Choose plan (free tier available)
### 2. WDS Integration
NanoBanana has two integration workflows in WDS:
#### **Workflow A: Early Design Exploration**
**Key Insight:** Brand identity is INDEPENDENT of product! You can establish visual identity anytime - before, during, or after product strategy.
```
Phase 5: Visual Design Exploration (can happen ANYTIME)
NanoBanana (create brand assets/visual identity)
Save to D-Design-System/01-Visual-Design/
Refine concepts → Establish visual direction
Define design tokens (colors, typography)
Phase 4: Create scenarios with established style
Export final assets → D-Design-System/02-Assets/
```
#### **Workflow B: Sketch Envisioning (Alternative to Figma)**
```
Phase 4: Create specification
NanoBanana (sketch/spec → design as image or code)
Freya interprets output (no Object IDs - manual process)
Extract design tokens and components
Update design system manually
Create/update prototype with refined design
```
**Key Difference from Figma MCP:**
- ❌ No Object ID preservation (manual interpretation required)
- ❌ No automated bidirectional sync
- ✅ Can generate code snippets directly
- ✅ Can produce visual designs from text descriptions
- ✅ Faster for exploration (no Figma setup needed)
**Folder Structure:**
- **01-Visual-Design/** - Early exploration, mood boards, NanoBanana outputs
- **02-Assets/** - Final logos, icons, images (added later)
---
## WDS Best Practices
### DO
**1. Use for Creative Exploration**
- Generate multiple variations
- Explore different styles
- Create mood boards
- Inspire design direction
**2. Refine AI Output**
- Don't use raw AI output
- Refine in Figma
- Align with brand guidelines
- Ensure consistency
**3. Document Asset Sources**
- Track AI-generated assets
- Note generation prompts
- Maintain asset library
- Document usage rights
**4. Integrate into Design System**
- Export refined assets
- Add to design system
- Create reusable components
- Document usage guidelines
### DON'T ❌
**1. Don't Replace Human Design**
- Use as inspiration, not replacement
- Apply design thinking
- Ensure brand alignment
- Review quality
**2. Don't Skip Refinement**
- Always refine AI output
- Ensure consistency
- Match brand guidelines
- Test usability
**3. Don't Use Without Customization**
- Customize for your brand
- Adapt to design system
- Ensure uniqueness
- Avoid generic output
---
## Usage Workflow
### 1. Generate Assets
```
1. Describe desired asset
2. Generate multiple variations
3. Select best options
4. Download assets
```
### 2. Refine in Figma
```
1. Import to Figma
2. Apply brand colors
3. Adjust to design system
4. Create variants if needed
```
### 3. Integrate into WDS
```
1. Add to design system
2. Document usage
3. Use in prototypes
4. Share with team
```
### 4. Sketch Envisioning (Alternative Workflow)
**For converting sketches/specs to designs:**
```
1. Provide sketch or specification to NanoBanana
2. Receive output (image or code)
3. Freya interprets output:
- Extract design tokens (colors, spacing, typography)
- Identify components and patterns
- Map to design system (manually - no Object IDs)
4. Update design system files
5. Create/update prototype
6. Test and refine
```
**Important Notes:**
- **No Object IDs** - Manual interpretation required
- **No automation** - Freya must manually extract and map components
- **Code snippets** - Need integration into WDS structure
- **Images** - Need manual component extraction
---
## When to Use
### Use NanoBanana for:
**Early Design Exploration:**
- **ANYTIME** - Brand identity is independent of product strategy!
- When establishing brand identity (can be before, during, or after product work)
- When exploring multiple visual directions
- Before committing to design tokens
- Can happen before Phase 1, parallel to strategy work, or after Phase 4
**Sketch Envisioning (Alternative to Figma):**
- Converting sketches to visual designs quickly
- Generating code snippets from design concepts
- When Figma/MCP server not available
- Rapid prototyping without design system
- Exploring multiple design variations
### Don't use for:
- **Automated workflows** - No MCP integration (manual interpretation needed)
- **Object ID traceability** - Outputs lack Object IDs (use Figma MCP for this)
- **Final production code** - Code snippets need refinement and integration
- **Replacing design process** - Use as tool, not replacement
- **Bypassing brand guidelines** - Always align with brand standards
---
## Resources
- Website: <https://nanobanana.com>
- Documentation: Check website for latest guides
- Support: Contact via website
---
[← Back to Tools](wds-tools-guide.md)