arXiv：cs.LG· Julian D. Santamaria, Kai Wang, Jes\'us Malo, Javier Vazquez-Corral, Alexandra G\'omez-Villa·· 4 小时前AI 评分44
VAE 潜空间中的颜色对齐及其应用：从 SD1.5 到 FLUX.2 与 Z-Image
On Color Alignment in VAE Latent Spaces and Its Applications
AI 导读
研究者发现文本到图像模型的 VAE 潜空间中普遍存在一个与亮度和对立色对齐的颜色子空间，该子空间在 SD1.5 到 FLUX.2 和 Z-Image 等广泛 VAE 中一致存在。基于此提出三项应用：ColorTuning 在 GenColorBench 的 CSS3/X11 细粒度色彩系统上实现 SOTA 精确数值颜色生成，以及饱和度控制和颜色迁移。代码与模型已公开。
正文
Abstract:Variational autoencoders (VAEs) are a key part of modern text-to-image models, which generate images within their latent space. VAEs are known to disentangle the main factors of variation in the data, and color is known to be one of the most structured of these in natural images: decorrelating it yields one luminance axis and two opponent-color axes. Color should therefore be expected to emerge as a distinct factor in the VAE latent space. Yet how these latent spaces represent color remains largely unexplored. In this work, we show that the VAEs of text-to-image models share a color subspace aligned with brightness and opponent-colors. Through a linear approximation of the encoder and targeted latent steering, we find this subspace consistently across a broad range of VAEs, from SD1.5 to FLUX.2 and Z-Image. Building on this characterization, we propose three applications: ColorTuning, which achieves state-of-the-art in precise numerical color generation on the fine-grained CSS3/X11 system of GenColorBench, saturation control, to adjust the global chromatic intensity, and color transfer, to change the palette to match a reference. The code and models are publicly available at this https URL
| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG) |
| Cite as: | arXiv:2610.07072 [cs.CV] |
| (or arXiv:2610.07072v1 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2610.07072 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Julian David Santamaria Julian D. Santamaria [view email]
[v1]
Mon, 5 Oct 2026 08:53:58 UTC (41,659 KB)
来源：arXiv：cs.LG · arxiv.org