Rectified-CFG++ for Flow Based Models

2025-10-08

AI Summary

Classifier-free guidance (CFG) in rectified flow (RF) models often induces out-of-manifold drift, resulting in visual artifacts, text misalignment, and generation instability. To address this, we propose Rectified-CFG++, an adaptive prediction-correction guidance mechanism operating within the deterministic RF framework. It jointly leverages weighted corrections from conditional and unconditional velocity fields, augmented by theoretically grounded boundary constraints and geometry-aware guidance rules that enforce trajectory confinement to the data manifold’s neighborhood while preserving marginal consistency. This is the first method enabling stable, in-manifold generation under strong CFG guidance. Extensive evaluation on Flux, Stable Diffusion 3/3.5, and Lumina demonstrates consistent superiority over standard CFG across MS-COCO, LAION-Aesthetic, and T2I-CompBench benchmarks—significantly improving generation fidelity, text-image alignment, and robustness.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Safety and Robustness

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Classifier-free guidance (CFG) is the workhorse for steering large diffusion models toward text-conditioned targets, yet its native application to rectified flow (RF) based models provokes severe off-manifold drift, yielding visual artifacts, text misalignment, and brittle behaviour. We present Rectified-CFG++, an adaptive predictor-corrector guidance that couples the deterministic efficiency of rectified flows with a geometry-aware conditioning rule. Each inference step first executes a conditional RF update that anchors the sample near the learned transport path, then applies a weighted conditional correction that interpolates between conditional and unconditional velocity fields. We prove that the resulting velocity field is marginally consistent and that its trajectories remain within a bounded tubular neighbourhood of the data manifold, ensuring stability across a wide range of guidance strengths. Extensive experiments on large-scale text-to-image models (Flux, Stable Diffusion 3/3.5, Lumina) show that Rectified-CFG++ consistently outperforms standard CFG on benchmark datasets such as MS-COCO, LAION-Aesthetic, and T2I-CompBench. Project page: https://rectified-cfgpp.github.io/

Problem

• Addresses off-manifold drift in rectified flow models with CFG
• Ensures stable trajectories near data manifold during generation
• Improves text-image alignment and reduces visual artifacts

Innovation

• Adaptive predictor-corrector guidance for rectified flows
• Geometry-aware conditioning rule ensures manifold proximity
• Interpolates conditional and unconditional velocity fields