README.md

Multimodality

A Go CLI tool for generating, editing, and analyzing images using OpenAI's API. Supports image generation from text prompts, image inpainting (completion), and image vision (analysis) tasks.

Features

• Image Generation: Create images from text prompts using OpenAI's image generation API.
• Image Completion (Inpainting): Edit or complete images by providing an original and a masked image.
• Image Vision: Analyze and answer questions about images using OpenAI's vision models.

Directory Structure

multimodality/
├── .env                # Environment variables (API keys, etc.)
├── go.mod, go.sum      # Go module files and dependencies
├── main.go             # Main CLI entry point
├── openai/             # OpenAI API client and image operation logic
└── test-files/         # Example images and test assets

Setup

1. Clone the repository and navigate to the multimodality directory.

2. Install dependencies (requires Go 1.24+):

   go mod tidy

3. Set up environment variables:

   • Copy .env and set your OpenAI API key:

     export OPEN_API_KEY=your-openai-api-key
     

Usage

Run the CLI with the desired action and flags:

Image Generation

Generate an image from a text description:

go run main.go -action=image-gen -image-desc="An astronaut riding a bicycle on the moon"

Image Completion (Inpainting)

Edit or complete an image using a mask:

go run main.go -action=image-complete -image-desc="Ancient Konark temple dedicated for Lord Surya (Sun) before it was destroyed." -image-path=./test-files/image.png -masked-image-path=./test-files/masked.png

Image Vision (Analysis)

Ask a question about an image (local file or URL):

go run main.go -action=image-vision -query="What did the Ancient Konark temple dedicated for Lord Surya (Sun) look like before it was destroyed." -image-path=./test-files/image.png

or

go run main.go -action=image-vision -query="What did the Ancient Konark temple dedicated for Lord Surya (Sun) look like before it was destroyed." -image-url=https://ik.imagekit.io/1hhs6vx06v/konark.png

Actions and Flags

Flag	Description
-action	One of: image-gen, image-complete, image-vision
-image-desc	Description for image generation or completion
-image-path	Path to the input image file
-masked-image-path	Path to the masked image file (for inpainting)
-query	Question to ask about the image (for vision)

Dependencies

• Go 1.24+
• OpenAI Go SDK
• caarlos0/env
• pkg/errors

License

MIT License