Create An API That Generates Images Based On A Prompt Using Python And Django

In this blog, you will learn how to create an API that generates images based on a prompt using Python and Django.

Create a new Django project.

django-admin startproject image_generator

Create a new Django app.

cd image_generator

python manage.py startapp image_api

Install the required libraries.

pip install tensorflow numpy pillow django-rest-framework

Define the model architecture that generates images from text prompts. For this example, we will use DALL-E 2, a neural network developed by OpenAI that can generate images from textual descriptions. You can download the model from the OpenAI API or use a pre-trained model available on GitHub.

Define the Django view that will receive the text prompt, generate an image using the DALL-E 2 model, and return the image as a response. You can create a new file views.py in your image_api app, and define a function like this:

from rest_framework.decorators import api_view
from rest_framework.response import Response
import numpy as np
from PIL import Image
import requests
import json

@api_view(['POST'])
def generate_image(request):
    prompt = request.data.get('prompt', '')
    model_url = 'https://api.openai.com/v1/images/generations/dall-e-2'
    headers = {
        'Content-Type': 'application/json',
        'Authorization': f'Bearer {YOUR_OPENAI_API_KEY}'
    }
    data = {
        'model': 'image-alpha-001',
        'prompt': prompt,
        'num_images': 1,
        'size': '256x256',
        'response_format': 'url'
    }
    response = requests.post(model_url, headers=headers, data=json.dumps(data))
    image_url = response.json()['data'][0]['url']
    image = Image.open(requests.get(image_url, stream=True).raw).convert('RGB')
    image_bytes = image.tobytes()
    return Response(image_bytes, content_type='image/jpeg')

Define a URL pattern that maps to the generate_image view. You can create a new file urls.py in your image_api app, and define a URL pattern like this

from django.urls import path
from .views import generate_image

urlpatterns = [
    path('generate_image/', generate_image),
]

Run the Django server and test the API. You can run the server using the following command:

python manage.py runserver

Test the API using a tool like cURL or Postman. For example, you can send a POST request to the http://localhost:8000/generate_image/ URL with a JSON payload like this

{
    "prompt": "a red apple on a table"
}

This should generate an image of a red apple on a table and return the image as a response.

That's it! You now have a Django API that generates images based on prompts using the DALL-E 2 model.