README.md

---

license: apache-2.0
---

<div align="center">

# L2P: Unlocking Latent Potential for Pixel Generation

<p>
  <a href="https://nju-pcalab.github.io/projects/L2P/">
    <img src="https://img.shields.io/badge/Project-Page-1f72ff?style=for-the-badge&logo=githubpages&logoColor=white" alt="Project Page">
  </a>
  <a href="https://arxiv.org/abs/2605.12013">
    <img src="https://img.shields.io/badge/arXiv-2605.12013-b31b1b?style=for-the-badge&logo=arxiv&logoColor=white" alt="arXiv">
  </a>
</p>

<p>
  <em>An efficient transfer paradigm enabling high-quality, end-to-end pixel-space diffusion with minimal computational overhead and data requirements.</em>
</p>
 
Pixel diffusion models have recently regained attention for visual generation. However, training advanced pixel-space models from scratch demands prohibitive computational and data resources. To address this, we propose the Latent-to-Pixel (L2P) transfer paradigm, an efficient framework that directly harnesses the rich knowledge of pre-trained LDMs to build powerful pixel-space models. Specifically, L2P discards the VAE in favor of large-patch tokenization and freezes the source LDM's intermediate layers, exclusively training shallow layers to learn the latent-to-pixel transformation. By utilizing LDM-generated synthetic images as the sole training corpus, L2P fits an already smooth data manifold, enabling rapid convergence with zero real-data collection. This strategy allows L2P to seamlessly migrate massive latent priors to the pixel space using only 8 GPUs. Furthermore, eliminating the VAE memory bottleneck unlocks native 4K ultra-high resolution generation. Extensive experiments across mainstream LDM architectures show that L2P incurs negligible training overhead, yet performs on par with the source LDM on DPG-Bench and reaches 93% performance on GenEval.