Generate Image

Generate images using Gemini's imagen-3.0 models (Nano Banana Pro).

When to Use

Use this skill when the user asks to:

• Generate an image from a text prompt
• Create artwork, illustrations, or graphics
• Generate variations of an existing image (img2img)
• Create scenes with multiple reference images (character + location + other characters)

Recommended Workflow

Draft → Iterate → Final approach saves time and API costs:

1. Draft Phase (1K): Generate quickly at default resolution to test prompts
2. Iteration Phase: Refine prompts incrementally, creating new files each attempt
3. Final Phase (4K): Only produce high-res output after prompt is validated

Do not read the image back - the script outputs only the file path. This allows efficient iteration without Claude loading large image data.

Usage

bun run scripts/generate.ts "prompt" [options]

Options

• --input <path> - Reference image (can specify multiple times, up to 14 images)
• --style <id> - Apply style from the style library (see browsing-styles skill)
• --size <1K|2K|4K> - Image size (default: 1K for fast drafts)
• --aspect <ratio> - Aspect ratio: 1:1, 16:9, 9:16, 4:3, 3:4
• --negative <prompt> - Negative prompt (what to avoid)
• --count <n> - Number of images (1-4, default: 1)
• --guidance <n> - Guidance scale
• --seed <n> - Random seed for reproducibility
• --output <path> - Output path

Examples

# Simple generation
bun run scripts/generate.ts "cyberpunk cityscape at night"

# With art style (100+ available, use short names or full IDs)
bun run scripts/generate.ts "mountain landscape" --style impressionism
bun run scripts/generate.ts "portrait" --style ukiy
bun run scripts/generate.ts "city street" --style noir

# High-res with specific aspect ratio
bun run scripts/generate.ts "mountain landscape" --size 4K --aspect 16:9

# With negative prompt
bun run scripts/generate.ts "portrait of a cat" --negative "low quality, blurry"

# Combine style with other options
bun run scripts/generate.ts "cat sleeping" --style wtrc --size 4K --aspect 1:1

# Generate multiple variations
bun run scripts/generate.ts "abstract art" --count 4

# Single reference image (img2img)
bun run scripts/generate.ts "make it look like a watercolor painting" --input photo.jpg

# Multiple reference images (character consistency, scene composition)
bun run scripts/generate.ts "King on throne in dramatic lighting" \
  --input character.png \
  --input throne-room.png \
  --aspect 16:9 --size 2K

# Multiple characters in a scene
bun run scripts/generate.ts "Two warriors facing each other in combat" \
  --input warrior1.png \
  --input warrior2.png \
  --input battlefield.png \
  --aspect 16:9

Multiple Reference Images

Gemini supports up to 14 reference images per request:

• 6 objects - Locations, items, environments
• 5 humans - Character consistency across generations
• Use detailed prompts describing how reference images should be combined
• Reference images help maintain consistency across scene generations

Available Styles

Use the browsing-styles skill to browse all 100+ styles:

• bun run ../browsing-styles/scripts/preview_server.ts - Interactive browser
• bun run ../browsing-styles/scripts/list_styles.ts --table - CLI list

Popular styles: impr (Impressionism), ukiy (Ukiyo-e), cybr (Cyberpunk), pixl (Pixel Art), noir (Film Noir), anim (Anime), wtrc (Watercolor)

Model

Uses gemini-3-pro-image-preview for image generation.