Abstract
Graphic design is a creative and innovative process that plays a crucial role in applications such as e-commerce and advertising. However, developing an automated design system that can faithfully translate user intentions into editable design files remains an open challenge. Although recent studies have leveraged powerful text-to-image models and MLLMs to assist graphic design, they typically simplify professional workflows, resulting in limited flexibility and intuitiveness. To address these limitations, we propose PSDesigner, an automated graphic design system that emulates the creative workflow of human designers. Building upon multiple specialized components, PSDesigner collects theme-related assets based on user instructions, and autonomously infers and executes tool calls to manipulate design files, such as integrating new assets or refining inferior elements. To endow the system with strong tool-use capabilities, we construct a design dataset, CreativePSD, which contains a large amount of high-quality PSD design files annotated with operation traces across a wide range of design scenarios and artistic styles, enabling models to learn expert design procedures. Extensive experiments demonstrate that PSDesigner outperforms existing methods across diverse graphic design tasks, empowering non-specialists to conveniently create production-quality designs.
Challenges in Automatic Graphic Design
Existing methods struggle to create professional graphic design file (eg., Adobe Photoshop Document, PSD) with complicated layer structure, hindering their applications. The following example illustrates the complexity of the PSD format.
Methodology
Construction of PSD-based dataset CreativePSD: We first present CreativePSD, a large-scale collection of PSD-format design files with annotated operation traces, enabling our PSDesigner to learn professional design processes from human designers. In particular, it consists of two subsets, designed for training the model for asset integration and layer refinement, respectively. The following figure illustrates the construction pipeline of the CreativePSD: We first collect high-quality PSD files, while grouping the layers based on their underlying visual concepts. Then, we parse the PSD files and extract essential information, such as raw assets, metadata, and intermediate renders. Finally, we use the extracted data to construct the training data for asset integration and layer refinement.
Design workflows from human experts and PSDesigner: Similarly, both of human designers and PSDesigner begin by collecting theme-related assets based on the user instructions. Next, they iteratively integrate these assets, where a bottom-up traversal is performed on the nested hierarchy, first at the group level and then at the asset level. In particular, each step consists of planning (①) and inserting (②) the current asset, then identifying deficiencies (③) and performing refinements (④). The above steps are repeated until all assets are integrated into the design file.
The architecture of PSDesigner: As indicated in the above figure, PSDesigner consists of following components. Specifically, AssetCollector first collects theme-related assets based on user instructions. Then, GraphicPlanner, trained on our CreativePSD, predicts tool calls based on the current design. In particular, it performs on two design modes for incorporating new assets and refining the inferior layers, respectively. Finally, ToolExecutor performs these tool calls to manipulate the PSD file.
Experiments
To demonstrate the effectiveness of our method, we conduct the following experiments. (1). We first evaluate the model’s ability to directly translate user intentions into final designs. (2). We further assess the model’s capability to perform graphic design composition based on the given assets. Specifically, we use the test data from Crello-v5 to evaluate the model performance in simple design scenarios. We further evaluate our method on copyright-free PSD files as a complement, featuring complex layer hierarchies.
Performance in graphic design.
Performance in graphic composition (Crello-v5).
Performance in graphic composition (PSD benchmark).
BibTeX
@inproceedings{shuai2026psdesigner,
title={PSDesigner: Automated Graphic Design with a Human-Like Creative Workflow},
author={Shuai, Xincheng and Tang, Song and Huang, Yutong and Ding, Henghui and Tao, Dacheng},
booktitle={CVPR},
year={2026}
}