Research map / Image generation / Fast diffusion sampling
Improving Image-to-Image Translation via a Rectified Flow Reformulation
Satoshi Iizuka, Shun Okamoto, Kazuhiro Fukui
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 |
|---|---|---|---|
| apply I2I-RFR (noisy-target concat, t-reweighted L1, 3 Euler steps) | SwinIR · Set5 SR x4 | LPIPS: 0.1687 -> 0.1665 | Improved |
| apply I2I-RFR | NLSN / SwinIR / U-Net · SR x4 benchmarks | PSNR: NLSN 32.84 -> 32.58; SwinIR 32.80 -> 32.84 | Mixed |
| apply I2I-RFR | Restormer · RealBlur-J | LPIPS: 0.1522 -> 0.1408 | Improved |
| apply I2I-RFR | Palette (diffusion U-Net) · Set5 SR x4 | PSNR: 22.28 -> 32.09 | Improved |
| apply I2I-RFR | LEDNet · LOLBlur | PSNR: 26.04 -> 27.42 | Improved |
| apply I2I-RFR (plain) | DarkIR · LOLBlur | PSNR: 27.31 -> 27.19 | No clear effect |
| keep composite loss, reweight only pixel term (I2I-RFR-CL) | DarkIR · LOLBlur | PSNR: 27.31 -> 27.72 | Improved |
| apply I2I-RFR | SFGNet · LSUI / UIEB | PSNR: LSUI 27.54 -> 26.09; UIEB 18.82 -> 19.69 | Mixed |
| apply I2I-RFR | RTN · DAVIS | LPIPS: 0.1562 -> 0.2100 | Hurt |
| apply I2I-RFR | RRTN · DAVIS | PSNR: 22.66 -> 25.87 | Improved |
| predict velocity instead of clean target (ablation) | I2I-RFR on DarkIR · LOLBlur | PSNR: 27.19 -> 16.73 | Hurt |
| add sinusoidal time embedding (ablation) | I2I-RFR on RRTN · DAVIS | PSNR: 25.87 -> 23.18 | Hurt |
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.
- Restormer — Restormer: Efficient Transformer for High-Resolution Image Restoration (CVPR 2022) · better on 2 of 2 shared results
- LEDNet — LEDNet: Joint Low-Light Enhancement and Deblurring in the Dark (ECCV 2022) · better on 1 of 1 shared result
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
- Taming Rectified Flow for Inversion and Editing (ICML 2025 · similarity 0.86)
- FlowVid: Taming Imperfect Optical Flows for Consistent Video-to-Video Synthesis (CVPR 2024 · similarity 0.86)
- Semantic Image Inversion and Editing using Rectified Stochastic Differential Equations (ICLR 2025 · similarity 0.86)
- Improving the Training of Rectified Flows (NeurIPS 2024 · similarity 0.86)
- Scaling Rectified Flow Transformers for High-Resolution Image Synthesis (ICML 2024 · similarity 0.85)
- Text-to-Image Rectified Flow as Plug-and-Play Priors (ICLR 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.
