Research map / Image generation / Editing and style transfer
Style Transfer by Relaxed Optimal Transport and Self-Similarity
Nicholas Kolkin, Jason Salavon, Gregory Shakhnarovich
Check 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 relaxed Earth Mover's Distance with both R_A and R_B instead of either term alone (ablation) | STROTSS style loss · Style transfer | stylization quality | Improved |
| Add moment matching to relaxed Earth Mover's Distance (ablation) | STROTSS style loss · Style transfer | feature magnitude and saturation artifacts | Improved |
| Add palette matching loss to relaxed Earth Mover's Distance and moment matching (ablation) | STROTSS style loss · Style transfer | palette matching | Improved |
| Use moment matching alone instead of the full style loss (ablation) | STROTSS · Style transfer | capture of larger style structures | Hurt |
| Use only R_A instead of the full relaxed Earth Mover's Distance (ablation) | STROTSS · Style transfer | content reconstruction | Hurt |
| Use only R_B instead of the full relaxed Earth Mover's Distance (ablation) | STROTSS · Style transfer | style-feature matching | Hurt |
| Replace cosine distance with Euclidean distance as the relaxed EMD ground metric (ablation) | STROTSS · Style transfer | stylization quality | Hurt |
| Optimize output pixels directly instead of Laplacian-pyramid entries (ablation) | STROTSS · Style transfer | convergence speed | Hurt |
| Use a single resolution with 800 updates instead of four increasing resolutions with 200 updates each (ablation) | STROTSS · Style transfer | stylization quality | Mixed |
| For a given level of content preservation, use STROTSS rather than prior methods | Style transfer methods in the human evaluation · Paired, unpaired, and texture regimes | style score | Improved |
Closest papers by meaning
- StyleGallery: Training-free and Semantic-aware Personalized Style Transfer from Arbitrary Image References (CVPR 2026 · similarity 0.89)
- SigStyle: Signature Style Transfer via Personalized Text-to-Image Models (AAAI 2025 · similarity 0.89)
- Domain-Aware Universal Style Transfer (ICCV 2021 · similarity 0.89)
- OmniStyle: Filtering High Quality Style Transfer Data at Scale (CVPR 2025 · similarity 0.89)
- StyleSSP: Sampling StartPoint Enhancement for Training-free Diffusion-based Method for Style Transfer (CVPR 2025 · similarity 0.88)
- Style Injection in Diffusion: A Training-free Approach for Adapting Large-scale Diffusion Models for Style Transfer (CVPR 2024 · similarity 0.88)
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.
