NanoBot

AI-powered visual tools using Google's Nano Banana Pro via fal.ai.

Main Tool: afcover

afcover is an Afghan music cover art generator designed specifically for the Afghan music industry. It uses reference images and cultural style presets to create professional album covers.

Features

• Reference-Based Generation: Use previous covers as style guides
• Afghan Music Styles: Traditional, modern, fusion, romantic, folk, ghazal, etc.
• Regional Variations: Kabuli, Herati, Kandahari, Mazari, Panjshiri aesthetics
• Cultural Authenticity: Designed for Afghan music industry standards
• Dari/Pashto Support: Typography optimized for Afghan scripts

Quick Setup

1. Clone the repository:

   git clone https://github.com/yourusername/nanobot.git
   cd nanobot

2. Install dependencies:

   pip install -r requirements.txt

3. API Key Setup:

   • Get a fal.ai API key at https://fal.ai/dashboard/keys
   • Copy the example environment file:

     cp .env.example .env

   • Edit the .env file and add your API key:

     FAL_KEY=your_actual_api_key
     

4. Verify installation:

   python -m afcover.cli --list-styles

Basic Usage

Generate Cover Art (CLI)

# Generate with reference images
python -m afcover.cli \
  --ref previous_cover.jpg \
  --title "Dilbar" \
  --artist "Ahmad Zahir" \
  --style traditional

# Modern style with Kabul aesthetic
python -m afcover.cli \
  --ref ref1.jpg --ref ref2.jpg \
  --title "Kabul Nights" \
  --artist "Aryana Sayeed" \
  --style modern \
  --regional kabuli

# Generate multiple variations
python -m afcover.cli \
  --ref reference.jpg \
  --title "Watan" \
  --style folk \
  --num 4

Generate Cover Art (Python API)

from afcover.generator import generate_cover

# Basic generation
result = generate_cover(
    reference_images=["artist_photo.jpg", "style_reference.jpg"],
    title="دل تنها",
    artist="احمد ظاهر",
    style="romantic",
    regional="kabuli",
    num_variations=2,
)

# Print paths to generated images
print(result["images"])

# Cost estimate only (no generation)
result = generate_cover(
    reference_images=["reference.jpg"],
    title="Test Album",
    style="traditional",
    dry_run=True,
)
print(f"Estimated cost: {result['estimated_cost']}")

Natural Language Interface

from afcover.bot import generate_from_request

# Generate from natural language request
result = generate_from_request(
    "Make a traditional cover for 'Laili Laili' by Ahmad Zahir with gold details",
    image_paths=["reference1.jpg", "reference2.jpg"],
)

# Format response
from afcover.bot import format_response
print(format_response(result))

Available Styles

Style	Description
traditional	Classic Afghan with ornate details and gold accents
modern	Contemporary pop style, clean and polished
fusion	East-West blend, bilingual appeal
romantic	Soft, emotional for love songs
folk	Rustic, earthy, homeland imagery
ghazal	Poetic, calligraphic, literary
sufi	Mystical and spiritual aesthetic
wedding	Festive, celebratory aesthetic
patriotic	National pride and symbolism
hiphop	Urban Afghan street style
acoustic	Intimate, unplugged aesthetic

Regional Modifiers

Region	Aesthetic
kabuli	Urban, cosmopolitan Kabul
herati	Artistic, Persian classical
kandahari	Bold, traditional Pashtun
mazari	Colorful, festive Turkic
panjshiri	Mountain imagery, natural beauty
badakhshi	High mountains, gem colors
hazaragi	Bamiyan heritage, resilient
nuristani	Forest mountains, wooden

Reference Library

The generator includes a reference library for consistent styles:

# List available references
python -m afcover.cli --list-references

# Use artist references from library
python -m afcover.cli --artist-ref "Ahmad Zahir" --title "New Cover" --style traditional

# Use style references from library
python -m afcover.cli --style-ref "modern" --title "Pop Hit" --artist "New Artist"

OpenClaw Integration

The afcover tool is designed to work seamlessly with OpenClaw. See SKILL.md for the full skill definition.

Bot Interface for OpenClaw

# Parse a natural language request
python3 -m afcover.bot parse "Make a traditional cover for 'Laili' by Ahmad Zahir"

# Generate with reference images (main entry point for OpenClaw)
python3 -m afcover.bot generate \
    --prompt "Traditional cover for 'Watan' by Ahmad Zahir" \
    --images reference1.jpg reference2.jpg \
    --output ./output

# Cost estimate only
python3 -m afcover.bot generate \
    --prompt "Modern Kabuli cover" \
    --images ref.jpg \
    --dry-run

# List available styles
python3 -m afcover.bot styles

# Manage reference library
python3 -m afcover.bot library list
python3 -m afcover.bot library add --type artist --name "Ahmad Zahir" --image photo.jpg

OpenClaw Conversation Flow

1. User uploads reference images (artist photos, style examples)
2. User describes desired cover in natural language
3. OpenClaw parses the request via afcover.bot parse
4. OpenClaw generates the cover via afcover.bot generate
5. OpenClaw returns the generated image to the user

Other Tools

• scripts/generate.py - Generic text-to-image generation
• scripts/edit.py - Generic image-to-image editing

Project Structure

nanobot/
├── afcover/              # Afghan music cover art tool
│   ├── cli.py            # Command-line interface
│   ├── bot.py            # Natural language interface
│   ├── generator.py      # Core generation logic
│   ├── styles.py         # Style definitions
│   ├── library.py        # Reference image management
│   └── references/       # Store reference artwork
│       ├── artists/      # Artist-specific references
│       └── styles/       # Style-specific references
├── scripts/              # Generic utilities
│   ├── generate.py       # Text-to-image
│   └── edit.py           # Image-to-image
├── shared/               # Shared modules
│   └── api.py            # fal.ai API wrapper
├── tests/                # Test cases
│   └── test_styles.py    # Style validation tests
├── docs/                 # Documentation
│   └── TESTING.md        # Comprehensive testing guide
├── USAGE.md              # Detailed usage guide
├── ROADMAP.md            # Development roadmap
└── README.md             # This file

Cost

• Standard (1K/2K): $0.15/image
• High-res (4K): $0.30/image

Troubleshooting

See the USAGE.md file for detailed troubleshooting tips and the docs/TESTING.md for comprehensive testing guidelines.

License

MIT