Research map / Image generation / Editing and style transfer
Smooth-Swap: A Simple Enhancement for Face-Swapping With Smoothness
Jiseob Kim, Jihoon Lee, Byoung-Tak Zhang
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 a supervised-contrastive smooth identity embedder instead of ArcFace (ablation) | Smooth-Swap · FaceForensics++ | Overall: -0.617 vs -0.329 | Improved |
| train for 50% of the steps (ablation) | Smooth-Swap · FaceForensics++ | Overall: -0.617 -> -0.398 | Hurt |
| reduce identity-loss weight to lambda_id=1 (ablation) | Smooth-Swap with the proposed embedder · FaceForensics++ | Overall: -0.617 -> -0.419 | Hurt |
| use ArcFace instead of the proposed embedder with lambda_id=1 (ablation) | Smooth-Swap generator · FaceForensics++ | Overall: -0.617 -> 0.266 | Hurt |
| use ArcFace instead of the proposed embedder with lambda_id=4 (ablation) | Smooth-Swap generator · FaceForensics++ | Overall: -0.617 -> 0.699 | Hurt |
| use the proposed contrastive embedder rather than ArcFace (ablation) | Smooth-Swap generator · FaceForensics++ training | identity-loss convergence speed | Improved |
| use the proposed contrastive embedder rather than ArcFace (ablation) | Smooth-Swap generator · FaceForensics++ training | change-loss stability | Improved |
| use the proposed contrastive embedder rather than ArcFace | identity embedding · FFHQ test split | d_smooth (r=0.25, samples): 0.360 -> 0.116 | Improved |
| use the proposed contrastive embedder rather than ArcFace | identity embedding · FFHQ test split | d_smooth (r=0.5, samples): 0.380 -> 0.135 | Improved |
| use the proposed contrastive embedder rather than ArcFace | identity embedding · StyleGAN2-generated faces | d_smooth (r=0.5, GAN): 0.802 -> 0.671 | Improved |
| use the proposed contrastive embedder rather than ArcFace | identity embedding · LFW | Verification AUC: 0.995 -> 0.999 | No clear effect |
Later papers that compared with this method
1 later accepted paper reports SmoothSwap as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- MFIM: Megapixel Facial Identity Manipulation (ECCV 2022 · 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
- High-Fidelity Diffusion Face Swapping with ID-Constrained Facial Conditioning (CVPR 2026 · similarity 0.92)
- DynamicFace: High-Quality and Consistent Face Swapping for Image and Video using Composable 3D Facial Priors (ICCV 2025 · similarity 0.91)
- BlendFace: Re-designing Identity Encoders for Face-Swapping (ICCV 2023 · similarity 0.90)
- StyleSwap: Style-Based Generator Empowers Robust Face Swapping (ECCV 2022 · similarity 0.89)
- Fast Face-Swap Using Convolutional Neural Networks (ICCV 2017 · similarity 0.89)
- Learning Disentangled Representation for One-Shot Progressive Face Swapping (TPAMI 2024 · similarity 0.89)
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.
