Research map / Image generation / Text-to-image diffusion
StyleDistillation: A New Insight of Image Style Enables Personalized Aesthetic Manipulation
Yuxin Wang, Xiaoyu Geng, Yuke Li, Zheng Wang
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 |
|---|---|---|---|
| add Prompt Alignment Enhancement at inference (ablation) | StyleDistillation · Style benchmark | CLIP-Text x100: 27.32 -> 28.95 | Improved |
| add Prompt Alignment Enhancement at inference (ablation) | StyleDistillation · Style benchmark | CSD score x100: 59.24 -> 53.26 | Hurt |
| add reconstruction loss (ablation) | StyleDistiller · Style benchmark (without PAE) | CSD score x100: 58.49 -> 59.24 | Improved |
| add orthogonality loss (ablation) | StyleDistiller · Style benchmark | CSD score x100: 61.22 -> 58.49 | Hurt |
| add orthogonality loss (ablation) | StyleDistiller · Qualitative content leakage | content leakage | Improved |
| add anchor loss (ablation) | StyleDistiller · Style benchmark | CSD score x100: 33.21 -> 61.22 | Improved |
| increase gamma in reconstruction loss (ablation) | StyleDistiller · Style benchmark | CSD score: CSD drops by an average of 13.5% | Hurt |
| increase lambda in PAE (ablation) | StyleDistillation · Prompt alignment vs style | CLIP-Text and CSD | Mixed |
| subtract content text embedding from image embedding (InstantStyle) instead of StyleDistiller | style extraction · Qualitative | style fidelity and content removal | Hurt |
| use StyleDistillation instead of IP-Adapter | SDXL · Style benchmark | Composite Quality score: 8.61 -> 10.73 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- CSGO — CSGO: Content-Style Composition in Text-to-Image Generation (NeurIPS 2025) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- StyleCLIP: Text-Driven Manipulation of StyleGAN Imagery (ICCV 2021 · similarity 0.91)
- 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)
- Style Aligned Image Generation via Shared Attention (CVPR 2024 · similarity 0.89)
- StyleStudio: Text-Driven Style Transfer with Selective Control of Style Elements (CVPR 2025 · similarity 0.88)
- FineStyle: Fine-grained Controllable Style Personalization for Text-to-image Models (NeurIPS 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.
