arXiv:2605.26538v1 Announce Type: new 
Abstract: Style transfer with pre-trained diffusion models has advanced rapidly, but a core question remains underexplored: where in the model should style injection be strongest? StyleID, the leading training-free method, uses a single global parameter (gamma) uniformly across all layers and timesteps, which forces a fixed tradeoff between style quality and content preservation. We show this tradeoff is unnecessarily rigid. We systematically explore four dimensions of control: varying style injection strength across decoder layers, across denoising timesteps, and scheduling ControlNet geometric conditioning along both axes. The pattern is consistent everywhere: decreasing schedules, with stronger structural signal injection in shallower layers and earlier timesteps, reliably outperform the reverse. Beyond direction, schedule shape matters: cosine and square-root timestep schedules outperform linear. Most importantly, we find that gamma scheduling and ControlNet conditioning are nearly independent. The resulting combined configurations expand the Pareto frontier, offering superior tradeoffs between style fidelity and content preservation compared to any single baseline setting. Our best balanced configuration achieves ArtFID of 27.036 versus StyleID's 28.801 - a 6.1% relative improvement, with consistent gains across the full style-content tradeoff frontier. Results are validated across 35 configurations totaling over 28,000 stylized images using four complementary metrics. These findings generalize across SD backbones with identical rank ordering. All modifications are training-free, parameter-free, and require only a few lines of scheduling code; code is available at https://github.com/ameyskulkarni/scheduled_style_injection.

قدمت دراسة حديثة نهجًا مبتكرًا لنقل الأسلوب باستخدام نماذج الانتشار المدربة مسبقًا، مع التركيز على تحسين حقن الأسلوب عبر طبقات مختلفة وخطوات إزالة الضوضاء. تكشف الأبحاث أن الجدولة المرنة لحقن الأسلوب يمكن أن تعزز بشكل كبير كل من جودة الأسلوب والحفاظ على المحتوى، مما يتحدى التوازن الصارم الذي وضعته الطريقة الرائدة، StyleID.

Un estudio reciente ha presentado un enfoque novedoso para la transferencia de estilo utilizando modelos de difusión preentrenados, centrándose en la optimización de la inyección de estilo a través de diferentes capas y pasos de eliminación de ruido. La investigación revela que una programación flexible de la inyección de estilo puede mejorar significativamente tanto la calidad del estilo como la preservación del contenido, desafiando el compromiso rígido establecido por el método líder, StyleID.

Une étude récente a introduit une approche novatrice du transfert de style utilisant des modèles de diffusion pré-entraînés, se concentrant sur l'optimisation de l'injection de style à travers différentes couches et étapes de débruitage. La recherche révèle qu'un calendrier flexible d'injection de style peut améliorer de manière significative à la fois la qualité du style et la préservation du contenu, remettant en question le compromis rigide précédemment établi par la méthode leader, StyleID.

A recent study has introduced a novel approach to style transfer using pre-trained diffusion models, focusing on optimizing style injection across different layers and timesteps. The research reveals that a flexible scheduling of style injection can significantly enhance both style quality and content preservation, challenging the previously rigid tradeoff established by the leading method, StyleID.