Scheduled Style Injection: Expanding the Style-Content Pareto Frontier in Training-Free Diffusion-based Style Transfer
Researchers have developed a new method for training-free diffusion-based style transfer that improves the balance between style fidelity and content preservation. By systematically exploring the optimal injection points for style across different decoder layers and denoising timesteps, they found that decreasing schedules, with stronger structural signal injection in earlier layers and timesteps, yield superior results. This approach, which also incorporates ControlNet geometric conditioning, expands the Pareto frontier, offering better tradeoffs than existing methods like StyleID. The new configuration achieved a 6.1% relative improvement in ArtFID score and has been validated across numerous configurations and metrics. AI
IMPACT This research offers improved control over style transfer in diffusion models, potentially leading to more nuanced and higher-quality image stylization for creative applications.
