Inspiration
We wanted a focused image workspace where people can move from a prompt or reference photo to an image they can download, without installing a desktop application.
What it does
ImgKick supports text-to-image generation, reference-based photo editing, saved outputs and a library of 87 tested prompt recipes. Visitors can try one shared generation or editing trial before signing in. Additional credits are available as one-time purchases.
How we built it
The browser interface uses React. A Go server serves the website and APIs, with SQLite for sessions, generation jobs and assets. Image-model requests run on the server so provider credentials stay out of the browser. The project is derived from the MIT-licensed Open Generative AI codebase, adapted for browser-only use.
Challenges and lessons
Image requests can outlive a browser session or encounter provider timeouts. We use durable job records and retain provider request IDs so ambiguous failures do not automatically trigger duplicate paid requests. Another practical task was testing prompt recipes and replacing source previews with outputs generated through ImgKick.
Try it
Use the image workspace or browse the tested prompt library. The prompt collection is also available on GitHub.
This project page is maintained by the ImgKick team.
Log in or sign up for Devpost to join the conversation.