FlowSteer: Guiding Few-Step Image Synthesis with Authentic Trajectories

TL;DR

FlowSteer enhances ReFlow-based image synthesis by guiding models along authentic trajectories, addressing distribution mismatches and scheduler flaws to improve efficiency and quality in few-step generation.

Contribution

This work introduces FlowSteer, a novel method that improves ReFlow-based distillation through trajectory guidance and fixes a scheduler flaw, advancing flow matching techniques.

Findings

FlowSteer improves few-step image synthesis quality.

Online Trajectory Alignment reduces distribution mismatch.

Fixing the scheduler flaw enhances inference performance.

Abstract

With the success of flow matching in visual generation, sampling efficiency remains a critical bottleneck for its practical application. Among flow models' accelerating methods, ReFlow has been somehow overlooked although it has theoretical consistency with flow matching. This is primarily due to its suboptimal performance in practical scenarios compared to consistency distillation and score distillation. In this work, we investigate this issue within the ReFlow framework and propose FlowSteer, a method unlocks the potential of ReFlow-based distillation by guiding the student along teacher's authentic generation trajectories. We first identify that Piecewised ReFlow's performance is hampered by a critical distribution mismatch during the training and propose Online Trajectory Alignment(OTA) to resolve it. Then, we introduce a adversarial distillation objective applied directly on the ODE trajectory, improving the student's adherence to the teacher's generation trajectory. Furthermore, we find and fix a previously undiscovered flaw in the widely-used FlowMatchEulerDiscreteScheduler that largely degrades few-step inference quality. Our experiment result on SD3 demonstrates our method's efficacy.