TP-Blend: Textual-Prompt Attention Pairing for Precise Object-Style Blending in Diffusion Models

Researchers introduced TP-Blend, a training-free framework for diffusion models that enables simultaneous object and style blending using two separate text prompts. The system uses Cross-Attention Object Fusion and Self-Attention Style Fusion to produce high-resolution, photo-realistic edits with precise control over both content and appearance.

• →TP-Blend addresses a key limitation in current text-conditioned diffusion editors that struggle with simultaneous object and style introduction.
• →The framework uses two complementary attention processors: CAOF for object fusion and SASF for style injection.
• →The system is training-free and lightweight, operating through attention mechanism modifications during the denoising process.
• →SASF uses Detail-Sensitive Instance Normalization with Gaussian filtering to preserve texture while maintaining global geometry.
• →Experimental results show superior performance in quantitative fidelity, perceptual quality, and inference speed compared to existing baselines.