Summary

UNO is a universal image generation framework from ByteDance's Intelligent Creation team, aimed at providing content creators and designers with unprecedented control and flexibility in generating high-quality images. Built on the principle of Less-to-More Generalization, UNO enables users to create both single-subject and complex multi-subject scenes with remarkable consistency and fidelity, leveraging diffusion transformers and a model-data co-evolution paradigm in which improved models generate better training data, which in turn improves the model.

Key Features

Supports both single and multi-subject image generation with high consistency
Progressive cross-modal alignment for precise subject control
Universal rotary position embedding (UnoPE) to prevent attribute confusion
High-consistency data synthesis pipeline using diffusion transformer in-context generation
Enables multi-image conditional input for complex scene creation
Built on FLUX.1-dev with minimal architectural modification
State-of-the-art results on DreamBench for single and multi-subject generation
Open-source code and models under Apache 2.0 and CC BY-NC 4.0

About UNO


At the core of UNO's technology are two major innovations: progressive cross-modal alignment and universal rotary position embedding (UnoPE). The progressive cross-modal alignment uses a two-stage training strategy, first fine-tuning a base text-to-image model with single-subject data, then further training with generated multi-subject data pairs. The UnoPE technique solves the problem of attribute confusion, ensuring the model can distinguish and preserve the features of each subject even in highly detailed or crowded scenes.


UNO's high-consistency data synthesis pipeline harnesses the inherent in-context generation capabilities of diffusion transformers to generate paired data with high consistency, supporting tasks such as virtual try-ons, product displays and brand-customized content creation. Presented at ICCV 2025, UNO achieves state-of-the-art results on DreamBench for both single-subject and multi-subject driven generation, and it is open-sourced under the Apache 2.0 license for code and CC BY-NC 4.0 for model weights.

Get more likes & reach the top of search results by adding this button on your site!

Embed button preview - Light theme
Embed button preview - Dark theme
TurboType Banner

Subscribe to the AI Search Newsletter

Get top updates in AI to your inbox every weekend. It's free!