Research map / Image generation / Editing and style transfer
Diverse Inpainting and Editing with GAN Inversion
Ahmet Burak Yildirim, Hamza Pehlivan, Bahri Batuhan Bilecen, Aysegul Dundar
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 full-reconstruction loss on generated images with sampled latent (ablation) | inpainting encoder + mixing network · FFHQ val | LPIPS (diversity): 0.0073 -> 0.1440 | Improved |
| add gated mixing network instead of fully connected layer (ablation) | mixing network · FFHQ val | FID: 20.91 -> 16.65 | Improved |
| add second-stage skip-connection training (ablation) | diverse model · FFHQ val | FID: 16.65 -> 5.92 | Improved |
| add second-stage training to non-diverse model (ablation) | Model 1 · FFHQ val | FID: 10.92 -> 7.87 | Improved |
| add diversity (full reconstruction training) before second stage (ablation) | inpainting model · FFHQ val | FID: 10.92 -> 16.65 | Hurt |
| add diversity with second stage (ablation) | inpainting model · FFHQ val | FID: 7.87 -> 5.92 | Improved |
| use our method instead of InvertFill on easy masks (0-0.4) | inpainting · FFHQ val easy masks | FID | Hurt |
| use our method instead of CoModGAN for diversity | inpainting · FFHQ val | LPIPS (diversity): 0.0790 -> 0.2026 | 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.
- 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
- High-Fidelity Image Inpainting with GAN Inversion (ECCV 2022 · similarity 0.94)
- In-Domain GAN Inversion for Faithful Reconstruction and Editability (TPAMI 2024 · similarity 0.93)
- High-Fidelity Image Inpainting with Multimodal Guided GAN Inversion (IJCV 2025 · similarity 0.93)
- WarpGAN: Warping-Guided 3D GAN Inversion with Style-Based Novel View Inpainting (NeurIPS 2025 · similarity 0.91)
- Gradual Residuals Alignment: A Dual-Stream Framework for GAN Inversion and Image Attribute Editing (AAAI 2024 · similarity 0.90)
- The Devil is in the Details: StyleFeatureEditor for Detail-Rich StyleGAN Inversion and High Quality Image Editing (CVPR 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.
