UNO

UNO is a local AI image generation framework for researchers and creators who want to preserve subjects from reference images in generated scenes. One model handles both single-subject images and compositions with several subjects.

• Reference-based generation: Multiple input images guide the appearance of subjects in the generated output.
• Multiple subjects: Combine distinct subjects in one image, with a method designed to reduce the mixing of attributes between them.
• Generation and training: The project includes inference and training code, plus a Gradio interface for running generation on your own hardware.

Checkpoints download from Hugging Face. GPU memory is a practical constraint: the flux-dev-fp8 mode with offloading has a stated peak memory use of 16 GB, with an NVIDIA RTX 3090 used as the inference example.

The code is open source under Apache 2.0. Model weights use CC BY-NC 4.0, which restricts commercial use. Models related to FLUX.1-dev must also follow that base model's licensing terms.