LayeringDiff: Layered Image Synthesis via Generation, then Disassembly with Generative Knowledge

Kyoungkook Kang1, Gyujin Sim1, Geonung Kim1,2, Donguk Kim3, Seungho Nam3, Sunghyun Cho1,2
1 POSTECH CSE     2 POSTECH GSAI     3 SHIFT UP


arXiv 2025

LayeringDiff generates a composite image and disassembles it into foreground and background layers.

Abstract

Layers have become indispensable tools for professional artists, allowing them to build a hierarchical structure that enables independent control over individual visual elements. In this paper, we propose LayeringDiff, a novel pipeline for the synthesis of layered images, which begins by generating a composite image using an off-the-shelf image generative model, followed by disassembling the image into its constituent foreground and background layers.

By extracting layers from a composite image, rather than generating them from scratch, LayeringDiff bypasses the need for large-scale training to develop generative capabilities for individual layers. Furthermore, by utilizing a pretrained off-the-shelf generative model, our method can produce diverse contents and object scales in synthesized layers. For effective layer decomposition, we adapt a large-scale pretrained generative prior to estimate foreground and background layers. We also propose high-frequency alignment modules to refine the fine details of the estimated layers. Our comprehensive experiments demonstrate that our approach effectively synthesizes layered images and supports various practical applications.

Pipeline

Initial Image Generation

An off-the-shelf text-to-image model synthesizes a diverse, high-quality composite image without task-specific fine-tuning.

Foreground Determination

The foreground prompt guides detection, segmentation, and matting to estimate the foreground region and alpha mask.

Layering

The FBDD and high-frequency alignment modules recover detailed foreground and background layers using a pretrained generative prior.

BibTeX

@article{kang2025layeringdiff,
  title   = {LayeringDiff: Layered Image Synthesis via Generation,
             then Disassembly with Generative Knowledge},
  author  = {Kang, Kyoungkook and Sim, Gyujin and Kim, Geonung and
             Kim, Donguk and Nam, Seungho and Cho, Sunghyun},
  journal = {arXiv preprint arXiv:2501.01197},
  year    = {2025}
}