Free ComfyUI Workflow: ByteDance USO for Style & Character Transfer

Tired of choosing between maintaining a character's identity and perfectly capturing an artistic style? This groundbreaking ComfyUI workflow featuring ByteDance's USO model finally unifies these two powerful concepts into a single, easy-to-use framework.

USO (Unified Style-Subject Optimized) is a state-of-the-art model built on the powerful FLUX.1 architecture. It's designed to solve a classic problem in AI art: how to apply the style of one image to another or place a specific character into a new scene, all without losing the crucial details. This workflow lets you do both—or even combine them.

What is ByteDance USO?

Developed by Bytedance's UXO team, USO is a model that treats style-driven and subject-driven generation as two sides of the same coin. Instead of using separate tools like IPAdapters for style and LoRAs for characters, USO's advanced training allows it to intelligently "disentangle" the subject (the "what") from the style (the "how") in any reference image.

This means you can achieve incredible consistency and control, whether you're transferring a painting's aesthetic or keeping a character's face the same across multiple images.

Three Powerful Generation Modes in One Workflow

This ComfyUI workflow unlocks the three core capabilities of the USO model:

• 👤 Subject-Driven Mode: Provide an image of a person, object, or character. The workflow will place that exact subject into a completely new scene based on your text prompt, ensuring identity consistency.
• 🎨 Style-Driven Mode: Provide an image with a distinct artistic style (e.g., a watercolor painting, a retro photograph). The workflow will apply that exact style to a new image generated from your text prompt.
• ✨ Combined Mode: The ultimate creative tool. Use one image as a subject reference and a second image as a style reference. The workflow will generate an image featuring your character rendered in your chosen artistic style.

How to Use This ComfyUI Workflow

Getting started is incredibly simple:

1. Load Your Image(s): Drag and drop your reference image(s) into the "Load Image" node(s). Use one for subject, one for style, or the same one for both.
2. Input Your Prompt: Write a text prompt describing the new scene or composition you want to create.
3. Generate Your Image: Click "Queue Prompt" and watch as USO perfectly blends your references and prompt into a new, coherent masterpiece.

Why is This a Game-Changer for AI Artists?

• Unified Simplicity: Replaces the need for complex, multi-node setups. One model handles both character and style locking.
• Superior Consistency: The model's unique training on disentangling content and style leads to highly reliable and accurate results.
• Built on FLUX.1: Leverages the power and efficiency of the next-generation FLUX.1 architecture for high-quality outputs.
• Unlocks New Creative Possibilities: The combined mode opens up a world of artistic experimentation that was previously difficult to achieve.

Frequently Asked Questions (FAQ)

How is this different from using an IPAdapter? While IPAdapters are excellent for style and composition, USO is specifically trained to unify both style and subject identity in a single, cohesive model. Its disentangled learning approach is designed to provide a higher degree of accuracy for both tasks simultaneously.

What kind of machine do I need? Like the base FLUX.1 model, USO is resource-intensive. A high-VRAM GPU (such as a "Large-Pro" tier machine) is recommended for the best performance.

Can I use it for more than one character or style? This specific workflow is optimized for a single subject and/or style reference to ensure the highest quality and consistency.

Download the Ultimate Style & Subject Workflow

Stop compromising and start creating. Download this free ComfyUI workflow to harness the power of ByteDance USO and take your image-to-image generations to the next level.

[Download the ByteDance USO Img2Img Workflow Now]