Research map / Image generation / Fast diffusion sampling
FlowChef: Steering of Rectified Flow Models for Controlled Generations
Maitreya Patel, Song Wen, Dimitris N. Metaxas, Yezhou Yang
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 FlowChef instead of DPS | DPS · Pixel-space box inpainting (easy) | PSNR: 23.61 -> 26.32 | Improved |
| use FlowChef instead of PnP-Flow | PnP-Flow · Pixel-space deblurring (hard) | PSNR: 19.50 -> 20.31 | Improved |
| use FlowChef instead of DPS | DPS · Pixel-space super-resolution (hard) | PSNR: 21.52 -> 21.62 | No clear effect |
| use FlowChef instead of D-Flow (time) | D-Flow · Pixel-space inverse problems | time (sec): 80.42 -> 4.31 | Improved |
| use FlowChef (Rectified Diffusion) instead of PSLD (500 NFEs) | PSLD · Latent-space box inpainting | PSNR: 28.30 -> 25.96 | Hurt |
| use FlowChef (Rectified Diffusion) instead of D-Flow | D-Flow · Latent-space box inpainting | PSNR: 19.68 -> 25.96 | Improved |
| scale FlowChef from InstaFlow to Flux | FlowChef · Latent-space super-resolution | PSNR: 25.83 -> 20.25 | Hurt |
| use human-annotated masks instead of ConceptAttention masks (ablation) | FlowChef-Edit (Flux) · PIE-Bench background preservation | PSNR: 24.31 -> 27.42 | Improved |
| use FlowChef-Edit (HA) instead of RF-Inversion | RF-Inversion · PIE-Bench background preservation | PSNR: 20.59 -> 27.42 | Improved |
| use FlowChef-Edit (HA) instead of InfEdit (HA) | InfEdit · PIE-Bench background preservation (SD v1.5) | PSNR: 28.45 -> 29.67 | Improved |
| use FlowChef-Edit (HA) instead of InfEdit (HA) | InfEdit · PIE-Bench structure distance | structure (lower better): 0.0091 -> 0.0347 | Hurt |
| use FlowChef-Edit (HA) instead of RF-Solver | RF-Solver · PIE-Bench CLIP similarity (whole) | CLIP similarity: 26.61 -> 25.69 | Hurt |
Later papers that compared with this method
8 later accepted papers report FlowChef as a baseline in their results tables; 7 of them report a better number on most of the results they share with it.
- TFTF: Training-Free Targeted Flow for Conditional Sampling (ICML 2026 · better on 6 of 7 shared results)
- Efficient Zero-shot Inpainting with Decoupled Diffusion Guidance (ICLR 2026 · better on 5 of 5 shared results)
- Variational Flow Maps: Make Some Noise for One-Step Conditional Generation (ICML 2026 · better on 2 of 4 shared results)
- SONIC: Spectral Optimization of Noise for Inpainting with Consistency (ECCV 2026 · better on 4 of 4 shared results)
- Solving Inverse Problems with Flow-based Models via Model Predictive Control (ICML 2026 · better on 2 of 3 shared results)
- Triadic Dynamics Aware Diffusion Posterior Sampling for Inverse Problems: Optimizing Guidance and Stochasticity Schedules (ICML 2026 · better on 2 of 2 shared results)
- FlowDPS : Flow-Driven Posterior Sampling for Inverse Problems (ICCV 2025 · better on 9 of 9 shared results)
- Solving Inverse Problems with FLAIR (NeurIPS 2025 · better on 4 of 4 shared results)
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.
- PnP-Flow — PnP-Flow: Plug-and-Play Image Restoration with Flow Matching (ICLR 2025) · better on 7 of 7 shared results
- D-Flow — DFlow: A Generative Model Combining Denoising AutoEncoder and Normalizing Flow for High Fidelity Waveform Generation (ICML 2024) · better on 2 of 2 shared results
- FlowEdit — FlowEdit: Inversion-Free Text-Based Editing Using Pre-Trained Flow Models (ICCV 2025) · 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
- D-Flow: Differentiating through Flows for Controlled Generation (ICML 2024 · similarity 0.90)
- Semantic Image Inversion and Editing using Rectified Stochastic Differential Equations (ICLR 2025 · similarity 0.90)
- Rectified Diffusion: Straightness Is Not Your Need in Rectified Flow (ICLR 2025 · similarity 0.89)
- Improving the Training of Rectified Flows (NeurIPS 2024 · similarity 0.89)
- Inverse Flow and Consistency Models (ICML 2025 · similarity 0.89)
- Simple ReFlow: Improved Techniques for Fast Flow Models (ICLR 2025 · 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.
