Multi-Image Composite Generator

Project Overview

This tool accepts 3–5 reference images along with a user-provided instruction prompt. It generates a new composite image by blending the described visual elements using OpenAI's DALL-E 3 API. The final output is saved in PNG format and also returned as a Base64-encoded string.

Features

• Supports 3–5 input images (URLs or local paths).
• Preprocesses images (resize to 1024x1024 and convert to RGB).
• Constructs detailed prompts for image generation.
• Saves the generated image locally and encodes it in Base64.
• Error handling for API issues and missing files.

Setup Instructions

1. Clone the repository or download the project files.
2. Install dependencies:

   pip install -r requirements.txt

3. Create a .env file in the project root with your OpenAI API key:

   OPENAI_API_KEY=your_api_key_here
   

4. Run the script:

   python scraper.py

Output

• Generated images are saved in the output/ directory.
• The script prints both the output path and a snippet of the Base64-encoded image string.

Requirements

• Python 3.9+
• OpenAI Python SDK >= 1.11.1
• Pillow >= 10.0.0
• Requests >= 2.31.0
• Python-dotenv >= 1.0.1

Notes

• This solution uses OpenAI’s DALL-E 3 API, which only supports text-based prompts and does not accept raw images as inputs.
• API keys should be kept secure and are not committed into the codebase.