Innovate Futures @ Benji
patreon
ByteDance Uno In ComfyUI - The In-Context Image Generation Better Than Alibaba?
🕑 Added 2025-04-16 13:00:29 +0000 UTC
Tutorial Video : https://youtu.be/Qexw5nWokfI
Discover Bytedance's Uno Framework for AI-Powered In-Context Image Generation!
In this video, we dive into Uno, a cutting-edge framework built on Flux and LoRA models, designed to merge multiple image references into creative outputs. Learn how Uno handles style transfers, object composition, and virtual try-ons—while exploring its pros, cons, and comparisons to Alibaba's in-context LoRA. We also provide a step-by-step guide to installing Uno in ComfyUI, including VRAM requirements, model setup, and practical tests for e-commerce, design, and AI-driven content creation.
Who Is This For?
AI developers and researchers exploring image diffusion models.
E-commerce professionals interested in virtual try-on solutions.
Content creators seeking AI tools for creative image generation.
Tech enthusiasts curious about Bytedance's latest AI frameworks.
Why It Matters:
Uno represents a leap in contextual AI image synthesis, enabling dynamic multi-reference generation. While still experimental, its potential for rapid prototyping, creative projects, and niche applications makes it a tool worth exploring—even as it evolves to address accuracy and scalability challenges.
Resources:
Project Page:
https://github.com/bytedance/UNO
https://bytedance.github.io/UNO/
ComfyUI Custom Node:
I got error with this: https://github.com/jax-explorer/ComfyUI-UNO/
I can run with this : https://github.com/HM-RunningHub/ComfyUI_RH_UNO
Download Models for using RunningHub's custom node:
Text Encoder from : https://huggingface.co/black-forest-labs/FLUX.1-schnell/tree/main
For text encoder download the following 4 folders:
text_encoder , text_encoder_2, tokenizer , tokenizer_2
For UNO / dit_lora.safetensors, download in here : https://huggingface.co/bytedance-research/UNO
