Research map / Image generation / Text-to-image diffusion
From Missing Pieces to Masterpieces: Image Completion With Context-Adaptive Diffusion
Pourya Shamsolmoali, Masoumeh Zareapoor, Huiyu Zhou, Michael Felsberg, Dacheng Tao, Xuelong Li
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 CAD instead of Wasserstein discrepancy (ablation) | ConFill · Places2 (Wide2 mask) | FID: 11.97 -> 10.39 | Improved |
| use CAD instead of L2 discrepancy (ablation) | ConFill · Places2 (Wide2 mask) | FID: 12.32 -> 10.39 | Improved |
| increase gradient steps from 2 to 5 (ablation) | ConFill with CAD · Places2 (Wide2 mask) | FID: 10.39 -> 10.56 | Hurt |
| use 2 gradient steps instead of 1 (ablation) | ConFill with CAD · Places2 (Wide2 mask) | LPIPS: 0.129 -> 0.120 | Improved |
| combine texture and edge criteria in dynamic sampling instead of texture only (ablation) | ConFill · CelebA (Expand mask) | FID: 15.86 -> 14.27 | Improved |
| combine texture and edge criteria in dynamic sampling instead of edge only (ablation) | ConFill · CelebA (Expand mask) | FID: 16.21 -> 14.27 | Improved |
| use combined handcrafted and learned context features instead of handcrafted only (ablation) | ConFill · ImageNet-1k (Expand mask) | FID: 55.34 -> 50.35 | Improved |
| use combined handcrafted and learned context features instead of learned only (ablation) | ConFill · ImageNet-1k (Expand mask) | FID: 52.95 -> 50.35 | Improved |
| raise upsilon from 0.05 to 0.10 (ablation) | ConFill · CelebA-HQ (Wide1 mask) | LPIPS: 0.130 -> 0.121 | Improved |
| vary tau from 0.01 to 0.03 (ablation) | ConFill · CelebA-HQ (Wide1 mask) | FID: 13.26 -> 13.27 | No clear effect |
| increase time travel interval from 2 to 10 (ablation) | ConFill · Places2 (Wide2 mask) | FID: 10.48 -> 10.39 | Improved |
| increase time travel frequency J from 1 to 5 (ablation) | ConFill · Places2 (Wide2 mask) | FID: 10.39 -> 10.42 | No clear effect |
Closest papers by meaning
- GeoComplete: Geometry-Aware Diffusion for Reference-Driven Image Completion (NeurIPS 2025 · similarity 0.88)
- RecDiffusion: Rectangling for Image Stitching with Diffusion Models (CVPR 2024 · similarity 0.87)
- Adapting Diffusion Models for Improved Prompt Compliance and Controllable Image Synthesis (NeurIPS 2024 · similarity 0.86)
- RAD: Region-Aware Diffusion Models for Image Inpainting (CVPR 2025 · similarity 0.86)
- High-Fidelity Diffusion-Based Image Editing (AAAI 2024 · similarity 0.86)
- Diffusing Differentiable Representations (NeurIPS 2024 · similarity 0.86)
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.
