Research map / Learning theory and algorithms / Optimal transport
User-Controllable Arbitrary Style Transfer via Entropy Regularization
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 ε-controlled entropy regularization for assignment | ε-AM · MS-COCO/PBN | pixel distance (PD): 0.088 -> 0.117 | Helped |
| Use ε-controlled entropy regularization for assignment | ε-AM · MS-COCO/PBN | LPIPS: 0.248 -> 0.375 | Helped |
| Set ε to 1e-2 | ε-AM · PBN artist deception evaluation | deception rate (%): 40.65 -> 47.35 | Helped |
| Set ε to 1e-0 instead of 1e-2 | ε-AM · PBN artist deception evaluation | deception rate (%): 47.35 -> 34.70 | Hurt |
| Increase ε from 1e-0 to 1e-2 | ε-AM · PBN artist deception evaluation | inference time (ms): 23 -> 48 | Hurt |
| Allow assignment-matrix gradients to update the encoder (ablation) | ε-AM · MS-COCO/PBN style transfer evaluation | content loss: 2.26 -> 1.97 | Helped |
| Allow assignment-matrix gradients to update the encoder (ablation) | ε-AM · MS-COCO/PBN style transfer evaluation | style loss: 3.02 -> 2.28 | Helped |
| Add quality-aware content mixing (ablation) | ε-AM · MS-COCO/PBN style transfer evaluation | content loss: 2.09 -> 1.97 | Helped |
| Add quality-aware content mixing (ablation) | ε-AM · MS-COCO/PBN style transfer evaluation | style loss: 2.82 -> 2.28 | Helped |
| Replace quality-aware mixing with constant content mixing (ablation) | ε-AM · MS-COCO/PBN style transfer evaluation | content loss: 2.01 -> 1.97 | Helped |
| Replace quality-aware mixing with constant content mixing (ablation) | ε-AM · MS-COCO/PBN style transfer evaluation | style loss: 2.59 -> 2.28 | Helped |
Closest papers by meaning
- OmniStyle: Filtering High Quality Style Transfer Data at Scale (CVPR 2025 · similarity 0.86)
- StyleSSP: Sampling StartPoint Enhancement for Training-free Diffusion-based Method for Style Transfer (CVPR 2025 · similarity 0.85)
- SigStyle: Signature Style Transfer via Personalized Text-to-Image Models (AAAI 2025 · similarity 0.84)
- StyleStudio: Text-Driven Style Transfer with Selective Control of Style Elements (CVPR 2025 · similarity 0.84)
- HSI: A Holistic Style Injector for Arbitrary Style Transfer (CVPR 2025 · similarity 0.84)
- Style Injection in Diffusion: A Training-free Approach for Adapting Large-scale Diffusion Models for Style Transfer (CVPR 2024 · similarity 0.84)
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.