AI Paper Atlas

Research map / Image generation / Fast diffusion sampling

FlowSteer: Guiding Few-Step Image Synthesis with Authentic Trajectories

Lei Ke, Hubery Yin, Gongye Liu, Zhengyao Lv, Jingcai Guo, Chen Li, Wenhan Luo, Yujiu Yang, Jing Lyu

CVPR 2026 · Image generation › Fast diffusion sampling · approach: Rectified flow straightening · 2 citations (OpenAlex)

Read the paper

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 changedWhereMeasuredEffect
Use authentic teacher trajectory start point instead of linear interpolationSD3 teacher · COCO 1k promptsHPSv2: 28.63 -> 29.43Improved
Add Online Trajectory Alignment (ablation)PeRFlow baseline, SD3-Medium · SD3M 4-step ablationHPSv2: 26.36 -> 26.79Improved
Add adversarial distillation on ODE trajectory (ablation)PeRFlow baseline, SD3-Medium · SD3M 4-step ablationHPSv2: 26.36 -> 27.73Improved
Add adversarial distillation on ODE trajectory (ablation)PeRFlow baseline, SD3-Medium · SD3M 4-step ablationPickScore: 22.19 -> 22.18No clear effect
Add improved scheduler (ablation)PeRFlow baseline, SD3-Medium · SD3M 4-step ablationPickScore: 22.19 -> 22.50Improved
Combine OTA, adversarial distillation and scheduler (full method) (ablation)PeRFlow baseline, SD3-Medium · SD3M 4-step ablationHPSv2: 26.36 -> 28.60Improved
Use improved scheduler with linear spacing including terminal sigmaSD3M original scheduler, shift 3, 4 steps · SD3M few-step inferenceHPSv2: 15.91 -> 19.34Improved
Use improved scheduler at 32 stepsSD3M, shift 3 · SD3M inferenceHPSv2: 29.50 -> 29.51No clear effect
Use 12-block discriminator backbone instead of 6, 18 or 24 blocks (ablation)FlowSteer discriminator · SD3M 4-stepHPSv2: 28.25 (24 blocks) -> 28.60 (12 blocks)Improved
Use one discriminator head on last block instead of per-block heads (ablation)FlowSteer discriminator · SD3M 4-stepHPSv2: 27.44 -> 28.60Improved
Fully fine-tune discriminator backbone instead of frozen or LoRA (ablation)FlowSteer discriminator · SD3M 4-stepHPSv2: 27.45 (frozen) / 28.07 (LoRA) -> 28.60Improved
Use hinge loss instead of LSGAN or WGAN (ablation)FlowSteer adversarial loss · SD3M 4-stepHPSv2: 27.64 (LSGAN), 27.28 (WGAN) -> 28.60Improved

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.

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

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy