ByteDance USO Image Model - This New AI Can Copy ANY Art Style With Just One Photo!
Added 2025-09-08 13:00:25 +0000 UTC
Tutorial Video : https://youtu.be/zCrwq-bp5lw
In this video, we dive into the newly released ByteDance Unified Style-Subject Optimized (USO) custom model, an exciting new add-on model built on top of Flux that brings powerful style transfer capabilities to local AI image generation. Using just one reference image, USO lets you transfer a character’s identity across different environments, artistic styles, and visual aesthetics—like turning a 3D character into ultra-realism, anime, or even Pixar-style renders—while maintaining strong subject consistency. We’ll show you exactly how it works inside ComfyUI, from setup to generating stunning results.
This content is perfect for AI art creators, generative model enthusiasts, and ComfyUI users who want to push the boundaries of what’s possible with local image generation. Whether you're into character design, animation, or creating viral transformation videos (like anime to realism), this tool opens up new creative possibilities. With USO, you no longer need custom LoRAs or complex training—just a single image and a few clicks.
Why does this matter? Because Flux1-dev has been around for over a year, and many thought its development had slowed down. But USO proves there’s still life in the model, with new, high-quality add-ons emerging that enhance style transfer and subject consistency like never before. This is a game-changer for creators looking to generate consistent characters across styles—especially for video projects using first-and-last-frame techniques in tools like WAN 2.2. Watch to see how you can use it today.
ByteDance Unified Style-Subject Optimized (USO)
https://bytedance.github.io/USO/
https://github.com/bytedance/USO
https://huggingface.co/bytedance-research/USO
How to Update ComfyUI
https://docs.comfy.org/installation/update_comfyui
USO In ComfyUI
https://docs.comfy.org/tutorials/flux/flux-1-uso
Comfy USO_1.0_Repackaged
https://huggingface.co/Comfy-Org/USO_1.0_Repackaged
Model links
Checkpoints
flux1-dev-fp8.safetensors
https://huggingface.co/Comfy-Org/flux1-dev/resolve/main/flux1-dev-fp8.safetensors
loras
uso-flux1-dit-lora-v1.safetensors
model_patches
uso-flux1-projector-v1.safetensors
clip_visions
sigclip_vision_patch14_384.safetensors
Attached Flux USO with Wan 2.2 FirstLastFrame2Video workflow mentioned in the tutorial video.