Research map / Image generation / Text-to-image diffusion
SONIC: Spectral Optimization of Noise for Inpainting with Consistency
Seungyeon Baek, Erqun Dong, Shadan Namazifard, Mark J Matthews, Kwang Moo Yi
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 spectral preconditioning (optimize noise in spectral domain) (ablation) | SONIC · FFHQ | FID: 14.579 -> 12.883 | Improved |
| add spatial constraining of gradient updates (ablation) | SONIC · FFHQ | FID: 33.269 -> 12.883 | Improved |
| add spatial constraining of gradient updates (ablation) | SONIC · FFHQ | LPIPS: 0.165 -> 0.121 | Improved |
| add spectral preconditioning (ablation) | SONIC · FFHQ | IR: 0.216 -> 0.280 | Improved |
| use ground-truth image instead of nearest-neighbor fill-in for the encoder (ablation) | SONIC · FFHQ | FID: 12.883 -> 12.773 | No clear effect |
| increase optimization iterations from 10 to 20 (ablation) | SONIC · FFHQ | PSNR: 22.088 -> 22.562 | Improved |
| increase optimization iterations from 20 to 30 (ablation) | SONIC · FFHQ | FID: 12.883 -> 12.840 | No clear effect |
| optimize initial noise with SONIC on BLD (SD3.5) | BLD SD3.5 / FlowChef-type baselines · FFHQ | LPIPS: 0.180 -> 0.121 | Improved |
| add SONIC noise optimization to SD1.5-BLD versus InitNo | SD1.5-BLD · FFHQ | FID: 35.945 -> 20.925 | Improved |
| add GoldenNoise to SDXL-BLD versus SONIC | SDXL-BLD · FFHQ | FID: 18.817 -> 18.006 | Improved |
| add FreeInpaint noise optimization to SD3.5-BLD versus SONIC | SD3.5-BLD · FFHQ | FID: 29.754 -> 12.883 | Improved |
| use 400 NFEs instead of default for FLAIR | FLAIR · FFHQ | FID: 18.982 -> 19.873 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- FlowChef — FlowChef: Steering of Rectified Flow Models for Controlled Generations (ICCV 2025) · better on 4 of 4 shared results
- FlowDPS — FlowDPS : Flow-Driven Posterior Sampling for Inverse Problems (ICCV 2025) · better on 3 of 3 shared results
- BrushNet — BrushNet: A Plug-and-Play Image Inpainting Model with Decomposed Dual-Branch Diffusion (ECCV 2024) · better on 3 of 3 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
- TurboFill: Adapting Few-step Text-to-image Model for Fast Image Inpainting (CVPR 2025 · similarity 0.86)
- Paint by Inpaint: Learning to Add Image Objects by Removing Them First (CVPR 2025 · similarity 0.86)
- Towards Enhanced Image Inpainting: Mitigating Unwanted Object Insertion and Preserving Color Consistency (CVPR 2025 · similarity 0.86)
- HD-Painter: High-Resolution and Prompt-Faithful Text-Guided Image Inpainting with Diffusion Models (ICLR 2025 · similarity 0.86)
- Image Inpainting via Tractable Steering of Diffusion Models (ICLR 2024 · similarity 0.85)
- RAD: Region-Aware Diffusion Models for Image Inpainting (CVPR 2025 · 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.
