Form and Void: Entangled Composition through an Autonomous AI Agent

2026-10-01

AI Summary

This study addresses the challenge of achieving semantic coherence in generating visual compositions involving both positive and negative space. To this end, we propose FaV-A, a multimodal agent framework that introduces a progressive generation pipeline. The method first constructs foundational objects, subsequently identifies negative-space semantics through shape analysis, and finally generates precise instructions to drive image synthesis. By deeply integrating multimodal large language models with text-to-image generation techniques, FaV-A enables a phased, collaborative workflow. Experimental results demonstrate that the proposed framework significantly outperforms zero-shot baselines in terms of visual coherence and semantic alignment, offering an effective new paradigm for the generation of positive and negative space compositions.

Abstract

Positive and negative space is a fundamental principle in visual composition, supporting visually coherent forms and layered semantic relationships. Generating such compositions is challenging because it requires coordinated control over two semantic concepts that share a common boundary. Although recent text-to-image models and multimodal large language models (MLLMs) have achieved strong performance in image generation and visual understanding, positive-negative space generation remains difficult, particularly under direct single-pass prompting. In this work, we present the \textbf{F}orm \textbf{a}nd \textbf{V}oid \textbf{A}gent (\textbf{FaV-A}), a multimodal agent designed for staged positive-negative space generation. FaV-A follows a progressive workflow: it first generates a base object, then analyzes its shape and spatial structure to identify candidate negative-space semantics, and finally produces compositional instructions for the final image generation stage. Experimental results and ablation analyses suggest that FaV-A provides a more effective framework than direct zero-shot MLLM baselines for producing visually coherent and semantically aligned positive-negative space compositions.

Problem

• positive-negative space
• visual composition
• text-to-image generation
• multimodal large language models

Innovation

• Positive-Negative Space
• Multimodal AI Agent
• Progressive Generation
• Visual Composition
• Text-to-Image