RecTok: Reconstruction Distillation along Rectified Flow

TL;DR

RecTok introduces a novel approach to high-dimensional visual tokenizers by leveraging flow semantic distillation and reconstruction alignment, resulting in improved image reconstruction and generation quality while maintaining semantic richness.

Contribution

The paper proposes RecTok, a method that overcomes the limitations of high-dimensional visual tokenizers through flow semantic distillation and reconstruction--alignment distillation, enhancing semantic richness and performance.

Findings

Achieves state-of-the-art results on gFID-50K.

Improves image reconstruction and generation quality.

Maintains semantic richness in latent space.

Abstract

Visual tokenizers play a crucial role in diffusion models. The dimensionality of latent space governs both reconstruction fidelity and the semantic expressiveness of the latent feature. However, a fundamental trade-off is inherent between dimensionality and generation quality, constraining existing methods to low-dimensional latent spaces. Although recent works have leveraged vision foundation models to enrich the semantics of visual tokenizers and accelerate convergence, high-dimensional tokenizers still underperform their low-dimensional counterparts. In this work, we propose RecTok, which overcomes the limitations of high-dimensional visual tokenizers through two key innovations: flow semantic distillation and reconstruction--alignment distillation. Our key insight is to make the forward flow in flow matching semantically rich, which serves as the training space of diffusion transformers, rather than focusing on the latent space as in previous works. Specifically, our method distills the semantic information in VFMs into the forward flow trajectories in flow matching. And we further enhance the semantics by introducing a masked feature reconstruction loss. Our RecTok achieves superior image reconstruction, generation quality, and discriminative performance. It achieves state-of-the-art results on the gFID-50K under both with and without classifier-free guidance settings, while maintaining a semantically rich latent space structure. Furthermore, as the latent dimensionality increases, we observe consistent improvements. Code and model are available at https://shi-qingyu.github.io/rectok.github.io.