fal.ai Image Generator

Use the fal.ai API to generate images from text prompts.

Official docs: https://fal.ai/docs

When to Use

Use this skill when you need to:

• Generate images from text descriptions
• Create illustrations or visual content
• Generate blog headers, thumbnails, or social media images

Prerequisites

1. Sign up at fal.ai
2. Get your API key from the dashboard

export FAL_TOKEN="your-api-key"

How to Use

1. Generate Image (nano-banana-pro - fast)

Write to /tmp/fal_request.json:

{
  "prompt": "A futuristic city at sunset, cyberpunk style"
}

Then run:

curl -s -X POST "https://fal.run/fal-ai/nano-banana-pro" --header "Authorization: Key $(printenv FAL_TOKEN)" --header "Content-Type: application/json" -d @/tmp/fal_request.json | jq -r '.images[0].url'

2. Generate Image (flux/schnell - fast)

Write to /tmp/fal_request.json:

{
  "prompt": "A cute cat eating a cookie"
}

Then run:

curl -s -X POST "https://fal.run/fal-ai/flux/schnell" --header "Authorization: Key $(printenv FAL_TOKEN)" --header "Content-Type: application/json" -d @/tmp/fal_request.json | jq -r '.images[0].url'

3. Generate Image (recraft-v3 - high quality)

Write to /tmp/fal_request.json:

{
  "prompt": "Abstract art, vibrant colors"
}

Then run:

curl -s -X POST "https://fal.run/fal-ai/recraft-v3" --header "Authorization: Key $(printenv FAL_TOKEN)" --header "Content-Type: application/json" -d @/tmp/fal_request.json | jq -r '.images[0].url'

4. Generate with Custom Size

Write to /tmp/fal_request.json:

{
  "prompt": "Mountain landscape",
  "image_size": "landscape_16_9"
}

Then run:

curl -s -X POST "https://fal.run/fal-ai/nano-banana-pro" --header "Authorization: Key $(printenv FAL_TOKEN)" --header "Content-Type: application/json" -d @/tmp/fal_request.json | jq -r '.images[0].url'

5. Download Generated Image

Write to /tmp/fal_request.json:

{
  "prompt": "A minimalist workspace"
}

Then run:

curl -s -X POST "https://fal.run/fal-ai/nano-banana-pro" --header "Authorization: Key $(printenv FAL_TOKEN)" --header "Content-Type: application/json" -d @/tmp/fal_request.json | jq -r '.images[0].url' | xargs curl -sL -o /tmp/image.png

6. Pipe Prompt from Echo (JSON escaped)

echo "A dragon breathing fire, epic fantasy art" | jq -Rs '{prompt: .}' > /tmp/fal_request.json
curl -s -X POST "https://fal.run/fal-ai/nano-banana-pro" --header "Authorization: Key $(printenv FAL_TOKEN)" --header "Content-Type: application/json" -d @/tmp/fal_request.json | jq -r '.images[0].url'

7. Pipe Prompt from File (JSON escaped)

cat /tmp/prompt.txt | jq -Rs '{prompt: .}' > /tmp/fal_request.json
curl -s -X POST "https://fal.run/fal-ai/nano-banana-pro" --header "Authorization: Key $(printenv FAL_TOKEN)" --header "Content-Type: application/json" -d @/tmp/fal_request.json | jq -r '.images[0].url'

8. Pipe with Additional Parameters

echo "Neon city at night" | jq -Rs '{prompt: ., image_size: "landscape_16_9"}' > /tmp/fal_request.json
curl -s -X POST "https://fal.run/fal-ai/nano-banana-pro" --header "Authorization: Key $(printenv FAL_TOKEN)" --header "Content-Type: application/json" -d @/tmp/fal_request.json | jq -r '.images[0].url'

Available Models

See more at: https://fal.ai/models

Image Sizes

Prompt Guidelines

For best results:

1. Be specific - Describe the subject clearly
2. Add style hints - "modern", "minimalist", "photorealistic", "digital art", "cinematic"
3. Specify colors/mood - "blue and purple gradient", "warm tones", "dark and moody"
4. Keep it concise - Clear and focused descriptions work better