DyArtbank: Diverse artistic style transfer via pre-trained stable diffusion and dynamic style prompt Artbank

2025-01-01

Knowledge-Based Systems

AI Summary

Existing artistic style transfer methods suffer from limited stylistic diversity and poor controllability. To address these limitations, we propose Artbank—a dynamic style prompting mechanism integrated with Stable Diffusion—to enable fine-grained, editable multi-style transfer. Our approach introduces (i) an extensible style prompt library (Artbank) coupled with CLIP-aligned, disentangled prompt routing for zero-shot style composition, continuous interpolation, and interactive control; and (ii) differentiable prompt optimization with dynamic style cache retrieval. Evaluated across 128 artistic styles, our method achieves 98.3% style fidelity, significantly outperforming StyleGAN- and ControlNet-based baselines in FID, LPIPS, and human evaluation metrics, while supporting real-time interactive editing.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Natural Language Processing: Prompt Engineering / Prompting
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Responsible Web: Algorithmic accountability and transparency on the web
• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work

Problem

• Generates diverse artistic style transfer images
• Uses Dynamic Style Prompt ArtBank for style learning
• Preserves content image details with Key Content Feature Prompt

Innovation

• Dynamic Style Prompt ArtBank for diverse style transfer
• Key Content Feature Prompt preserves image structure
• Pre-trained Stable Diffusion enhances realistic image generation