Research map / Image generation / Editing and style transfer
Learning Dynamic Style Kernels for Artistic Style Transfer
Wenju Xu, Chengjiang Long, Yongwei Nie
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use spatially adaptive style kernels generated from content-style aligned features instead of global style modulation | StyleKernel · Style transfer evaluation | Style Loss: 1.04 -> 0.98 | Improved |
| Add adversarial loss to the attention-based Base model (ablation) | StyleKernel · Style transfer evaluation | Style Loss: 2.92 -> 2.14 | Improved |
| Add REMD loss to the adversarial-loss variant (ablation) | StyleKernel · Style transfer evaluation | Style Loss: 2.14 -> 1.48 | Improved |
| Add style kernels to the adversarial- and REMD-loss variant (ablation) | StyleKernel · Style transfer evaluation | Style Loss: 1.48 -> 1.11 | Improved |
| Add CGM to the style-kernel variant (ablation) | StyleKernel · Style transfer evaluation | Style Loss: 1.11 -> 1.08 | Improved |
| Add grouped shuffling to the CGM and style-kernel variant (ablation) | StyleKernel · Style transfer evaluation | Style Loss: 1.08 -> 0.98 | Improved |
| Increase style kernel size from k=3 to k=5 (ablation) | StyleKernel · Style transfer evaluation | LPIPS: 0.30 -> 0.31 | Hurt |
| Increase style kernel size from k=3 to k=5 (ablation) | StyleKernel · Style transfer evaluation | T512: 0.035 -> 0.044 | Hurt |
| Increase style kernel size from k=3 to k=5 (ablation) | StyleKernel · Style transfer evaluation | Style Loss: 0.98 -> 0.97 | Improved |
| Replace globally shared style modulation with per-position kernels informed by content-style alignment | StyleKernel · Artistic style transfer | content structure preservation | Improved |
| Use CGM to suppress weakly correlated style features during attention aggregation (ablation) | StyleKernel · Artistic style transfer | semantic region completeness | Improved |
| Use grouped shuffling in the content feature (ablation) | StyleKernel · Artistic style transfer | style consistency | Improved |
Closest papers by meaning
- Adjustable Real-time Style Transfer (ICLR 2020 · similarity 0.89)
- Learning Linear Transformations for Fast Image and Video Style Transfer (CVPR 2019 · similarity 0.88)
- HSI: A Holistic Style Injector for Arbitrary Style Transfer (CVPR 2025 · similarity 0.88)
- SigStyle: Signature Style Transfer via Personalized Text-to-Image Models (AAAI 2025 · similarity 0.87)
- Style Injection in Diffusion: A Training-free Approach for Adapting Large-scale Diffusion Models for Style Transfer (CVPR 2024 · similarity 0.87)
- CSGO: Content-Style Composition in Text-to-Image Generation (NeurIPS 2025 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
