Research map / Image generation / Editing and style transfer
Constructing Non-isotropic Gaussian Diffusion Model Using Isotropic Gaussian Diffusion Model for Image Editing
Xi Yu, Xiang Gu, Haozhi Liu, Jian Sun
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 |
|---|---|---|---|
| increase a from 6.0 to 14.0 (b=5.0) (ablation) | NGDM · AFHQ | FID: 87.01 -> 46.75 | Improved |
| increase a from 6.0 to 14.0 (b=5.0) (ablation) | NGDM · AFHQ | SSIM: 0.556 -> 0.425 | Hurt |
| increase b from 3.0 to 7.0 (a=10.0) (ablation) | NGDM · AFHQ | FID: 42.34 -> 88.49 | Hurt |
| increase b from 3.0 to 7.0 (a=10.0) (ablation) | NGDM · AFHQ | SSIM: 0.373 -> 0.601 | Improved |
| use NGDM instead of EGSDE | NGDM · AFHQ | FID: 65.82 -> 61.39 | Improved |
| use NGDM instead of SDDM | NGDM · AFHQ | FID: 62.29 -> 61.39 | Improved |
| use NGDM instead of SDEdit | NGDM · AFHQ | SSIM: 0.423 -> 0.478 | Improved |
| use NGDM instead of StarGAN v2 | NGDM · AFHQ | FID: 54.88 -> 61.39 | Hurt |
| use NGDM instead of CUT | NGDM · AFHQ | SSIM: 0.601 -> 0.478 | Hurt |
| use NGDM instead of EGSDE in user study | NGDM · AFHQ user study | vote share: 12.5% -> 65.5% | Improved |
| use NGDM instead of DiffEdit in user study | NGDM · ImageNet/Imagen/COCO-S/DreamBooth | vote share: 10.0% -> 72.0% | Improved |
| use soft weighting matrix instead of hard threshold mask (ablation) | NGDM · image editing | artifacts | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 4 on most shared results.
- SDEdit — SDEdit: Guided Image Synthesis and Editing with Stochastic Differential Equations (ICLR 2022) · better on 3 of 3 shared results
- EGSDE — EGSDE: Unpaired Image-to-Image Translation via Energy-Guided Stochastic Differential Equations (NeurIPS 2022) · better on 3 of 3 shared results
- SDDM — SDDM: Score-Decomposed Diffusion Models on Manifolds for Unpaired Image-to-Image Translation (ICML 2023) · better on 2 of 2 shared results
- StarGAN v2 — StarGAN v2: Diverse Image Synthesis for Multiple Domains (CVPR 2020) · better on 1 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-Based Image Editing (AAAI 2024 · similarity 0.86)
- Contrastive Denoising Score for Text-guided Latent Diffusion Image Editing (CVPR 2024 · similarity 0.85)
- Residual Denoising Diffusion Models (CVPR 2024 · similarity 0.85)
- Denoising Diffusion Bridge Models (ICLR 2024 · similarity 0.85)
- Directly Denoising Diffusion Models (ICML 2024 · similarity 0.85)
- GaussianEditor: Editing 3D Gaussians Delicately with Text Instructions (CVPR 2024 · similarity 0.85)
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.
