Illustrator's Depth: Monocular Layer Index Prediction for Image Decomposition

2025-11-21

AI Summary

This work addresses the challenge of decomposing planar raster images into editable, ordered layers—a fundamental limitation in digital content creation. We propose “Illustrator’s Depth,” a创作-oriented abstraction of depth that redefines traditional geometric depth as a pixel-wise layer ordering index. Our method trains a neural network on a layered vector graphics dataset to predict globally consistent, discretized layer orders directly from single raster inputs, incorporating interpretability mechanisms to facilitate user editing. Crucially, we are the first to model depth not as a geometric metric but as a creative, hierarchical indexing scheme. Experiments demonstrate substantial improvements over state-of-the-art methods in image vectorization, while enabling unified support for high-fidelity text-to-vector generation, 2D-to-3D relief modeling, and interactive depth-aware editing—establishing a novel paradigm for author-centric digital content creation.

Technology Category

• Computer Vision: Computational Photography, Image & Video Synthesis
• Machine Learning: Deep Neural Architectures and Foundation Models
• Cognitive Modeling & Cognitive Systems: Computational Creativity

Application Category

• Web Mining and Content Analysis: Large pretrained models with web data
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

We introduce Illustrator's Depth, a novel definition of depth that addresses a key challenge in digital content creation: decomposing flat images into editable, ordered layers. Inspired by an artist's compositional process, illustrator's depth infers a layer index to each pixel, forming an interpretable image decomposition through a discrete, globally consistent ordering of elements optimized for editability. We also propose and train a neural network using a curated dataset of layered vector graphics to predict layering directly from raster inputs. Our layer index inference unlocks a range of powerful downstream applications. In particular, it significantly outperforms state-of-the-art baselines for image vectorization while also enabling high-fidelity text-to-vector-graphics generation, automatic 3D relief generation from 2D images, and intuitive depth-aware editing. By reframing depth from a physical quantity to a creative abstraction, illustrator's depth prediction offers a new foundation for editable image decomposition.

Problem

• Decomposes flat images into editable ordered layers
• Predicts layer indices from raster inputs using neural networks
• Enables image vectorization and depth-aware editing applications

Innovation

• Predicts layer indices from raster images
• Uses neural network trained on vector graphics
• Enables editable image decomposition and vectorization