ART: Anonymous Region Transformer for Variable Multi-Layer Transparent Image Generation

2025-02-25

AI Summary

This paper addresses key challenges in multi-layer transparent image generation—strong semantic layout dependency, high computational cost, and severe inter-layer conflicts. To this end, we propose an end-to-end generative framework leveraging global text prompts and anonymous region layouts. Our core contributions are threefold: (1) replacing conventional semantic-labeled layouts with anonymous region layouts to decouple spatial structure from semantics; (2) introducing a layer-aware visual token pruning mechanism and the Anonymous Region Transformer to reduce cross-layer attention complexity; and (3) designing a joint encoder-decoder autoencoder for multi-layer transparent images to jointly model transparency and content across layers. Experiments demonstrate that our method efficiently generates over 50 layers, achieves >12× faster inference than full-attention baselines, significantly mitigates inter-layer interference, and supports high-fidelity image synthesis and interactive layer editing.

Technology Category

• Computer Vision: Interpretability, Explainability, and Transparency
• Natural Language Processing: Generation
• Machine Learning: Transparent, Interpretable, Explainable ML

Application Category

• Security and Privacy: Data transparency and provenance
• Responsible Web: Algorithmic accountability and transparency on the web
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG

Abstract

Multi-layer image generation is a fundamental task that enables users to isolate, select, and edit specific image layers, thereby revolutionizing interactions with generative models. In this paper, we introduce the Anonymous Region Transformer (ART), which facilitates the direct generation of variable multi-layer transparent images based on a global text prompt and an anonymous region layout. Inspired by Schema theory suggests that knowledge is organized in frameworks (schemas) that enable people to interpret and learn from new information by linking it to prior knowledge.}, this anonymous region layout allows the generative model to autonomously determine which set of visual tokens should align with which text tokens, which is in contrast to the previously dominant semantic layout for the image generation task. In addition, the layer-wise region crop mechanism, which only selects the visual tokens belonging to each anonymous region, significantly reduces attention computation costs and enables the efficient generation of images with numerous distinct layers (e.g., 50+). When compared to the full attention approach, our method is over 12 times faster and exhibits fewer layer conflicts. Furthermore, we propose a high-quality multi-layer transparent image autoencoder that supports the direct encoding and decoding of the transparency of variable multi-layer images in a joint manner. By enabling precise control and scalable layer generation, ART establishes a new paradigm for interactive content creation.

Problem

• Generates variable multi-layer transparent images
• Reduces attention computation costs efficiently
• Enables precise control and scalable layer generation

Innovation

• Anonymous Region Transformer enables transparent image generation
• Layer-wise region crop reduces attention computation costs
• Multi-layer transparent image autoencoder supports joint encoding