Image generation with Azure OpenAI DALL-E 3

DALL-E 3 is a Generative AI model from OpenAI, which can produce images based on your textual input (prompt). You will find a Jupyter notebook in this repo, that utilises Azure OpenAI deployment of DALL-E 3 to generate caricature image of a famous Disney character.

To build this demo, I used the latest version of OpenAI Python SDK - v1.x. To upgrade your openai Python package, please use the following pip command:

pip install --upgrade openai

Table of contents:

• Part 1: Configuring solution environment
• Part 2: Generating and visualising required image

Part 1: Configuring solution environment

1. To use API key authentication, assign the API endpoint name, version and key, along with the Azure OpenAI deployment name of DALL-E 3 model to OPENAI_API_BASE, OPENAI_API_VERSION, OPENAI_API_KEY and OPENAI_API_DEPLOY_DALLE environment variables respectively.

Note: If you want to use the Entra ID (former Azure Active Directory) authentication instead, you may find some implementation options here.

2. Install the required Python packages, by using the pip command and the provided requirements.txt file.

pip install -r requirements.txt

Part 2: Generating and visualising required image

1. Connection with the backend Azure OpenAI service is established through the openai Python SDK v1. Current implementation passes the Azure OpenAI endpoint's API key as a parameter value of the AzureOpenAI class. If necessary, you can switch to the Entra ID authentication instead.

client = AzureOpenAI(
    api_version = AOAI_API_VERSION,
    api_key = AOAI_API_KEY,
    azure_endpoint = AOAI_API_BASE,
)

2. We define then a helper function, that would call our DALL-E 3 model in Azure OpenAI and return URL of generated image.

def image_generator(prompt):
    # Generate image with DALL-E 3
    response = client.images.generate(
        model = AOAI_DEPLOYMENT,
        prompt = prompt,
    )
    
    # Extract image URL
    json_response = json.loads(response.model_dump_json())
    image_url = json_response["data"][0]["url"]
    return image_url

3. We can then execute our helper function with an image prompt, e.g. "Caricature picture of Mickey Mouse with a hat"
4. If successful, we should be able to download generated image with Pillow and visualise it in default graphic editor.

image = Image.open(requests.get(image_url, stream=True).raw)
image.show()

5. Below is an example from my run. Your output may look different.