Amazing.photos

The platform is built on the Flux generative AI architecture and uses Replicate's cloud infrastructure for model training and inference, making professional-quality personalized AI photography accessible to anyone without requiring local GPU hardware or machine learning expertise.

The generation workflow begins with the user uploading a minimum of five photos of themselves, with better results achieved from 20 or more diverse reference images showing different angles, lighting conditions, and facial expressions.

These photos are sent to Replicate's cloud to train a custom Flux fast-trainer model, a process that completes in approximately two minutes on H100 GPU hardware. Once training is finished, Replicate sends a webhook to the backend and the personalized model is immediately available for inference requests.

Because Amazing.photos is fully open-source under an MIT license, it occupies a distinct position in the AI photo market: developers and technically inclined users can self-host the entire platform on their own infrastructure, gaining complete control over training data, generated images, and model storage.

The codebase is built with Next.js, providing a clean, modern web interface that can be customized and extended. The cloud-hosted version allows non-technical users to experience the same capabilities without any setup.

A notable characteristic of Amazing.photos is its unrestricted content generation policy.

Unlike many consumer AI image platforms that apply safety filters to restrict certain image types, the self-hosted deployment has no built-in content restrictions, giving users complete creative freedom over the images they generate.