Research map / Image generation / Editing and style transfer
PHOTOSWAP: Personalized Subject Swapping in Images
Jing Gu, Yilin Wang, Nanxuan Zhao, Tsu-Jui Fu, Wei Xiong, Qing Liu, Zhifei Zhang, HE Zhang, Jianming Zhang, HyunJoon Jung, Xin Eric 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 |
|---|---|---|---|
| use Photoswap instead of P2P+DreamBooth | DreamBooth subject swapping · Human eval | overall quality preference: 37.3% vs. 27.1% | Improved |
| use Photoswap instead of P2P+DreamBooth | DreamBooth subject swapping · Human eval | background preservation: 39.3% vs. 30.2% | Improved |
| use Photoswap instead of P2P+DreamBooth | DreamBooth subject swapping · Human eval | subject swapping: 42.4% vs 30.0% | Improved |
| use Photoswap instead of PnP+DreamBooth | DreamBooth subject swapping · Human eval | overall quality preference: 55.1% vs 22.4% | Improved |
| use Photoswap instead of MasaCtrl+DreamBooth | DreamBooth subject swapping · Human eval | overall quality preference: 83.3% vs 10.3% | Improved |
| start attention swap from the earliest diffusion steps (ablation) | Photoswap · qualitative | layout preservation | Improved |
| swap self-attention output and maps in all decoder layers of the U-Net (ablation) | Photoswap · qualitative | subject swapping quality | Improved |
| vary latent size of the swapped U-Net layer (ablation) | Photoswap · qualitative | subject swapping quality | No clear effect |
| increase swapping step lambda_M for self-attention map M (ablation) | Photoswap · qualitative | identity similarity to source | Mixed |
| copy cross-attention map from source to target (ablation) | Photoswap · qualitative | layout alignment | Improved |
Later papers that compared with this method
1 later accepted paper reports Photoswap as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- SwapAnything: Enabling Arbitrary Object Swapping in Personalized Image Editing (ECCV 2024 · better on 4 of 4 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
- PS-Diffusion: Photorealistic Subject-Driven Image Editing with Disentangled Control and Attention (CVPR 2025 · similarity 0.87)
- MS-Diffusion: Multi-subject Zero-shot Image Personalization with Layout Guidance (ICLR 2025 · similarity 0.86)
- VideoSwap: Customized Video Subject Swapping with Interactive Semantic Point Correspondence (CVPR 2024 · similarity 0.85)
- Multi-party Collaborative Attention Control for Image Customization (CVPR 2025 · similarity 0.85)
- DynASyn: Multi-Subject Personalization Enabling Dynamic Action Synthesis (AAAI 2025 · similarity 0.84)
- InstantSwap: Fast Customized Concept Swapping across Sharp Shape Differences (ICLR 2025 · 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.
