Research map / Image generation / Editing and style transfer
Warping the Residuals for Image Editing with StyleGAN
Ahmet Burak Yildirim, Hamza Pehlivan, Ayşegül Dündar
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 trained feature-space flow with flow loss (vs StyleRes without warping) | StyleRes · CelebA-HQ reconstruction | FID: 7.04 -> 6.59 | Improved |
| add trained feature-space flow with flow loss | StyleRes · CelebA-HQ pose editing | Id(+): 0.58 -> 0.74 | Improved |
| add trained feature-space flow with flow loss | StyleRes · CelebA-HQ pose editing | FID(+): 11.31 -> 10.40 | Improved |
| add flow estimation to StyleRes | StyleRes · CelebA-HQ runtime | runtime (sec): 0.0429 -> 0.1254 | Hurt |
| add flow guidance loss from pretrained flow network (ablation) | trained flow WarpRes · CelebA-HQ pose editing | FID(+): 16.38 -> 10.40 | Improved |
| add flow guidance loss from pretrained flow network (ablation) | trained flow WarpRes · CelebA-HQ smile removal | FID(-): 24.54 -> 21.52 | Improved |
| add flow guidance loss from pretrained flow network (ablation) | trained flow WarpRes · CelebA-HQ reconstruction | FID: 7.24 -> 6.59 | Improved |
| replace pretrained image-space flow with trained feature-space flow (ablation) | WarpRes · CelebA-HQ reconstruction | FID: 7.55 -> 6.59 | Improved |
| replace pretrained image-space flow with trained feature-space flow (ablation) | WarpRes · CelebA-HQ pose editing | Id(+): 0.67 -> 0.74 | Improved |
| add WarpRes on top of pSp encoder | pSp · CelebA-HQ reconstruction | FID: 23.86 -> 6.60 | Improved |
| add WarpRes instead of StyleRes on pSp encoder | pSp + StyleRes · CelebA-HQ pose editing | Id(+): 0.58 -> 0.72 | Improved |
| add WarpRes on top of StyleTransformer | StyleTransformer · CelebA-HQ reconstruction | FID: 21.82 -> 6.85 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- StyleRes — StyleRes: Transforming the Residuals for Real Image Editing With StyleGAN (CVPR 2023) · better on 10 of 11 shared results
- HyperStyle — HyperStyle: StyleGAN Inversion With HyperNetworks for Real Image Editing (CVPR 2022) · 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
- StyleRes: Transforming the Residuals for Real Image Editing With StyleGAN (CVPR 2023 · similarity 0.96)
- Delving StyleGAN Inversion for Image Editing: A Foundation Latent Space Viewpoint (CVPR 2023 · similarity 0.95)
- ReGANIE: Rectifying GAN Inversion Errors for Accurate Real Image Editing (AAAI 2023 · similarity 0.94)
- The Devil is in the Details: StyleFeatureEditor for Detail-Rich StyleGAN Inversion and High Quality Image Editing (CVPR 2024 · similarity 0.93)
- Gradual Residuals Alignment: A Dual-Stream Framework for GAN Inversion and Image Attribute Editing (AAAI 2024 · similarity 0.90)
- WarpGAN: Warping-Guided 3D GAN Inversion with Style-Based Novel View Inpainting (NeurIPS 2025 · 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.
