Mage-Flow: An Efficient Native-Resolution Foundation Model for Image Generation and Editing

📄 arXiv: 2607.19064v1 📥 PDF

作者: Xinjie Zhang, Peng Zhang, Shicheng Zheng, Jinghao Guo, Zhaoyang Jia, Yifei Shen, Xun Guo, Yuxuan Luo, Jiahao Li, Wenxuan Xie, Fanyi Pu, Xiaoyi Zhang, Kaichen Zhang, Zongyu Guo, Tianci Bi, Dongnan Gui, Zhening Liu, Zimo Wen, Zihan Zheng, Senqiao Yang, Xiao Li, Jinglu Wang, Bin Li, Yan Lu

分类: cs.CV, cs.AI, cs.LG, cs.MM, eess.IV

发布日期: 2026-07-21

💡 一句话要点

提出Mage-Flow以实现高效的图像生成与编辑

🎯 匹配领域: 支柱二：RL算法与架构 (RL & Architecture) 支柱九：具身大模型 (Embodied Foundation Models)

关键词: 图像生成 图像编辑 多模态扩散 潜在标记器 高效模型 实时应用 CUDA内核融合

📋 核心要点

  1. 现有的大规模视觉生成模型在训练和部署上成本高，效率低，难以满足实时应用需求。
  2. Mage-Flow通过结合Mage-VAE和原生分辨率扩散变换器，提供了一种高效的图像生成和编辑解决方案。
  3. 实验结果表明，Mage-Flow在标准基准测试中表现出竞争力，Turbo变体在高分辨率生成和编辑中实现了显著的速度提升。

📝 摘要（中文）

随着大规模视觉生成模型的能力不断提升，其训练、微调和部署的成本也随之增加。本文提出了Mage-Flow，一个紧凑的4B规模生成模型，旨在高效地进行文本到图像的生成和基于指令的图像编辑。该模型由两个共同设计的组件构成：Mage-VAE，一个轻量级的高保真潜在标记器，以及一个使用修正流匹配训练的原生分辨率多模态扩散变换器。Mage-VAE通过一步扩散风格的编码和解码，结合锚点潜在正则化，显著降低了标记化成本，同时保持了重建质量。该模型在高分辨率生成和编辑方面表现出色，尤其是在交互式应用中具有实际价值。

🔬 方法详解

问题定义：当前大规模视觉生成模型在训练和推理时面临高成本和低效率的问题，限制了其在实时应用中的使用。

核心思路：Mage-Flow通过设计轻量级的Mage-VAE和原生分辨率的多模态扩散变换器，旨在降低标记化成本并提高生成效率。

技术框架：Mage-Flow的整体架构包括Mage-VAE作为潜在标记器，结合原生分辨率的扩散变换器，支持灵活的分辨率训练和高效的CUDA内核融合。

关键创新：Mage-VAE通过一步扩散风格的编码和解码，结合锚点潜在正则化，显著提升了重建质量并降低了计算成本，形成了与现有方法的本质区别。

关键设计：在模型设计中，采用了高效的损失函数和网络结构，确保了在保持小内存占用的同时，实现了高分辨率生成和编辑的能力。

🖼️ 关键图片

fig_0
fig_1
fig_2

📊 实验亮点

Mage-Flow-Turbo在单个NVIDIA A100 GPU上以0.59秒生成1024²分辨率的图像，Mage-Flow-Edit-Turbo以1.02秒编辑图像，显示出显著的速度提升和小内存占用，证明了其在高效生成和编辑中的优势。

🎯 应用场景

Mage-Flow在图像生成和编辑领域具有广泛的应用潜力，尤其适用于需要实时反馈的交互式应用，如游戏开发、虚拟现实和在线内容创作。其高效的性能和较低的资源消耗使得高分辨率图像处理变得更加可行。

📄 摘要（原文）

Large-scale visual generators are increasingly capable but costly to train, fine-tune, and deploy. We introduce Mage-Flow, a compact 4B-scale generative stack for efficient text-to-image generation and instruction-based image editing. The stack is built from two co-designed components: Mage-VAE, a lightweight high-fidelity latent tokenizer, and a Native-Resolution Multimodal Diffusion Transformer trained with rectified flow matching. Mage-VAE uses one-step diffusion-style encoding and decoding with anchor-latent regularization, preserving the reconstruction quality of strong public VAEs while reducing tokenization cost by more than an order of magnitude. Together with native-resolution packing and stack-level CUDA kernel fusion, the stack supports flexible-resolution training and improves end-to-end training throughput by about $2.5\times$. Built on this foundation, we develop a complete model family with Base, RL-aligned, and Turbo variants for both generation and editing. Diffusion-NFT improves prompt following, text rendering, aesthetic quality, and editing fidelity, while few-step distillation with adversarial perceptual guidance produces 4-step Turbo models for low-latency inference. Despite its compact scale, Mage-Flow and Mage-Flow-Edit achieves competitive performance across standard generation and editing benchmarks. More importantly, the Turbo variants make high-resolution generation and editing practical for interactive use: at $1024^2$ resolution on a single NVIDIA A100 GPU, Mage-Flow-Turbo generates an image in 0.59s, and Mage-Flow-Edit-Turbo edits an image in 1.02s, while maintaining a small memory footprint. These results show that careful tokenizer--backbone--system co-design can deliver strong high-resolution generation and editing within an efficient 4B model family.