Art
Art Skill — Image Generation
A CLI tool for generating images from text prompts using multiple AI models. Used by the brand-skill in Phase 2 (Visual Direction) and Phase 3 (Mark Development) for reference image generation.
The Tool
Tools/Generate.ts — A multi-model image generation CLI that supports:
| Model | Flag | API Key | Best for |
|---|---|---|---|
| Gemini (nano-banana-pro) | --model nano-banana-pro |
GOOGLE_API_KEY |
Best quality, text rendering |
| Replicate (nano-banana) | --model nano-banana |
REPLICATE_API_TOKEN |
Faster iteration |
| Flux | --model flux |
REPLICATE_API_TOKEN |
High quality, stylistic variety |
| GPT Image | --model gpt-image-1 |
OPENAI_API_KEY |
Alternative generation |
Setup
- Install bun runtime:
curl -fsSL https://bun.sh/install | bash - Set at least one API key as an environment variable
- Run with:
bun run Tools/Generate.ts --model [MODEL] --prompt "[PROMPT]" --output ~/Downloads/output.png
Usage
# Basic generation
bun run Tools/Generate.ts \
--model nano-banana-pro \
--prompt "Abstract minimalist logo concept: [description]. Clean vector style, dark background. No text." \
--size 2K \
--aspect-ratio 1:1 \
--output ~/Downloads/brand-ref-1.png
# With background removal (useful for marks)
bun run Tools/Generate.ts \
--model nano-banana-pro \
--prompt "[PROMPT]" \
--size 2K \
--remove-bg \
--output ~/Downloads/mark-reference.png
# With reference image for style consistency
bun run Tools/Generate.ts \
--model nano-banana-pro \
--prompt "[PROMPT]" \
--reference-image existing-mark.png \
--size 2K \
--output ~/Downloads/variation.png
Flags
| Flag | Options | Default | Purpose |
|---|---|---|---|
--model |
nano-banana-pro, nano-banana, flux, gpt-image-1 | nano-banana-pro | Image generation model |
--prompt |
text | (required) | The generation prompt |
--size |
1K, 2K, 4K (or aspect like 16:9) | 16:9 | Output resolution or aspect ratio |
--aspect-ratio |
1:1, 16:9, 9:16, 3:2, 21:9 | 16:9 | Aspect ratio (Gemini nano-banana-pro) |
--output |
path | ~/Downloads/generated-image.png | Output file path |
--remove-bg |
(flag) | off | Remove background (requires REMOVEBG_API_KEY) |
--thumbnail |
(flag) | off | Generate both transparent + background versions (enables --remove-bg) |
--reference-image |
path (repeatable) | none | Style/content reference (nano-banana-pro only) |
--creative-variations |
number | 1 | Generate multiple variations (-v1, -v2, etc.) |
In the Brand Process
Phase 2 — Visual Direction (Mode A):
Generate 3-4 reference images exploring different visual interpretations of the brand concept. Use --aspect-ratio 1:1 for logo references (default is 16:9). High resolution, abstract/minimalist prompts.
Phase 3 — Mark Development (Path A):
Generate high-contrast reference images for bitmap tracing with vtracer. Use --remove-bg for clean marks. Trace the result to SVG.
Tips
- Always output to
~/Downloads/first for preview - Square (1:1) aspect ratio works best for logo references
- Include "no text" in prompts for logo/mark references
- Use
--remove-bgwhen generating marks for tracing - Keep prompts abstract: "minimalist logo concept" not "a logo for my company"
nano-banana-prohandles text rendering best if you need labels
More from b1rdmania/claude-brand-skills
brand
Complete brand development system with emotive foundation. Triggers on requests for brand identity, logos, visual systems, or design guidelines. Creates distinctive, anti-AI-slop design from strategy through delivery.
20canvas-design
Create beautiful visual art in .png and .pdf documents using design philosophy. You should use this skill when the user asks to create a poster, piece of art, design, or other static piece. Create original visual designs, never copying existing artists' work to avoid copyright violations.
12frontend-design
Create distinctive, production-grade frontend interfaces with high design quality. Use this skill when the user asks to build web components, pages, or applications. Generates creative, polished code that avoids generic AI aesthetics.
12