Evolve to Inspire: Novelty Search for Diverse Image Generation

2025-11-01

AI Summary

Text-to-image diffusion models often suffer from limited output diversity, hindering their utility in creative exploration tasks. To address this, we propose WANDER—a novel framework that introduces novelty search to text-to-image generation for the first time. WANDER employs a *launcher* mechanism to steer the model toward distinct semantic regions, leverages large language models (LLMs) for semantic evolution of prompts, and quantifies image novelty via CLIP embeddings to guide diverse generation. Integrated with FLUX-DEV and augmented by GPT-4o-mini for mutation assistance, WANDER achieves statistically significant improvements over existing evolutionary baselines across standard diversity metrics. Ablation studies confirm the launcher’s critical role in enhancing both semantic coverage and distributional breadth. This work establishes a new paradigm for controllable, interpretable, and highly diverse text-to-image synthesis.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Generation

Application Category

• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work

Abstract

Text-to-image diffusion models, while proficient at generating high-fidelity im- ages, often suffer from limited output diversity, hindering their application in exploratory and ideation tasks. Existing prompt optimization techniques typically target aesthetic fitness or are ill-suited to the creative visual domain. To address this shortcoming, we introduce WANDER, a novelty search-based approach to generating diverse sets of images from a single input prompt. WANDER operates directly on natural language prompts, employing a Large Language Model (LLM) for semantic evolution of diverse sets of images, and using CLIP embeddings to quantify novelty. We additionally apply emitters to guide the search into distinct regions of the prompt space, and demonstrate that they boost the diversity of the generated images. Empirical evaluations using FLUX-DEV for generation and GPT-4o-mini for mutation demonstrate that WANDER significantly outperforms existing evolutionary prompt optimization baselines in diversity metrics. Ablation studies confirm the efficacy of emitters.

Problem

• Address limited diversity in text-to-image diffusion models
• Generate diverse images from single prompts using novelty search
• Enhance visual creativity through semantic evolution and CLIP embeddings

Innovation

• Uses LLM for semantic evolution of prompts
• Employs CLIP embeddings to quantify novelty
• Applies emitters to guide search diversity