Moebius：0.2十亿像素的轻量级图像修复框架，性能达到100亿像素级别
Moebius: 0.2B Lightweight Image Inpainting Framework with 10B-Level Performance
摘要
虽然10B级别的工业级基础模型已经突破了图像修复技术的极限，但其极高的计算成本却严重限制了其实际应用。构建一种高度优化的、针对特定任务的专用模型则是一种可行的解决方案；不过，极端的结构压缩方式必然会导致表示能力受限的问题。为了解决这一问题，我们提出了Moebius——一个高效且轻量的图像修复框架。我们通过引入Local-λ Mix Interaction（LλMI）模块来系统地重建扩散模型的核心结构。该模块由Local-λ和Interactive-λ模块组成，能够将空间上下文和全局语义信息转化为固定大小的线性矩阵，从而保留复杂的潜在关系，同时大幅减少参数数量。此外，为了充分发挥这种紧凑架构的表示能力，我们还将它与自适应多粒度蒸馏策略相结合。这种策略在潜空间中进行操作，避免了昂贵的像素空间解码过程，能够动态平衡多种基于梯度的损失项，从而实现高保真度的修复效果。在各种自然图像和肖像图像数据集上的实验表明，这种优化策略使得Moebius能够与10B级别的工业级FLUX.1-Fill-Dev模型相媲美甚至超越其生成质量。值得注意的是，Moebius仅使用了不到2%的参数数量（0.22B对比11.9B），同时还能将整体推理时间缩短15倍，从而创造了高保真度图像修复的新效率标准。项目页面地址：https://hustvl.github.io/Moebius。
English Abstract
While 10B-level industrial foundation models have pushed the boundaries of image inpainting, their prohibitive computational costs severely hinder practical deployment. Constructing a highly optimized task-specific specialist offers a promising solution; however, extreme structural compression inevitably triggers a severe representation bottleneck. To conquer this, we propose Moebius, a highly efficient lightweight inpainting framework. We systematically reconstruct the diffusion backbone by introducing the Local-λ Mix Interaction (LλMI) block. Comprising Local-λ and Interactive-λ modules, it elegantly summarizes spatial contexts and global semantic priors into fixed-size linear matrices, preserving complex latent interactions while drastically shedding parameters. Furthermore, to unlock the full representational capacity of this highly compact architecture, we synergistically pair it with an adaptive multi-granularity distillation strategy. Operating strictly within the latent space to avoid expensive pixel-space decoding, this strategy dynamically balances multiple gradient-based losses to achieve high-fidelity alignment. Extensive experiments across natural and portrait benchmarks demonstrate that this optimal synergy enables Moebius to rival or even surpass the generation quality of the 10B-level industrial generalist FLUX.1-Fill-Dev. Remarkably, Moebius achieves this using less than 2\% of the parameters (0.22B vs. 11.9B) while delivering a >15times acceleration in total inference time, setting a new efficiency standard for high-fidelity inpainting. Project page at https://hustvl.github.io/Moebius.