VeilPlays

VeilPlays offers a novel approach to evaluating generative models by transforming the assessment process into an interactive gaming experience. Key features include: * Play games created by different generative models * Blind comparison of game concepts and visual designs * Evaluate coding quality and overall game performance * Provide direct player feedback for model ranking * Benchmark various models across diverse game genres This platform allows users to engage with outputs from various generative models without prior knowledge of their origins. By playing games, users can provide authentic feedback on aspects like game concept originality, aesthetic appeal, and technical execution. This blind evaluation ensures unbiased assessments, leading to more accurate and reliable performance benchmarks for models. Immerse yourself in a variety of game genres, including Dungeon Explorer, Enhanced Arkanoid, Card Battle, Mini FPS, Survival Defense, and Side-scrolling Action, each designed to test different facets of model capabilities. VeilPlays is ideal for researchers, developers, and enthusiasts interested in understanding and comparing the practical performance of generative models in a fun and engaging way. It facilitates a community-driven benchmarking system, offering valuable insights into model strengths and weaknesses through real-world application and user interaction. Elevate your understanding of generative model capabilities through actual gameplay and contribute to a robust evaluation framework.