Image to Image Generation Agent

Transform an existing image into a new one with `FalTools(enable_image_to_image=True)`.

Create an agent that generates new images from existing images using the Fal AI API.

Code

image_to_image_agent.py
from agno.agent import Agent
from agno.models.openai import OpenAIResponses
from agno.tools.fal import FalTools

agent = Agent(
    model=OpenAIResponses(id="gpt-5.2"),
    id="image-to-image",
    name="Image to Image Agent",
    tools=[FalTools(enable_image_to_image=True)],
    markdown=True,
    instructions=[
        "You have to use the `image_to_image` tool to generate the image.",
        "You are an AI agent that can generate images using the Fal AI API.",
        "You will be given a prompt and an image URL.",
        "You have to return the image URL as provided, don't convert it to markdown or anything else.",
    ],
)

agent.print_response(
    "a cat dressed as a wizard with a background of a mystic forest. Make it look like 'https://fal.media/files/koala/Chls9L2ZnvuipUTEwlnJC.png'",
    stream=True,
)

Usage

Set up your virtual environment

uv venv --python 3.12
source .venv/bin/activate

Install dependencies

uv pip install -U agno openai fal-client

Export your API keys

  export OPENAI_API_KEY="your_openai_api_key_here"
  export FAL_KEY="your_fal_api_key_here"

Run Agent

python image_to_image_agent.py