Title:Advances in 4D Generation: A Survey

Abstract:Generative artificial intelligence (AI) has made significant progress across various domains in recent years. Building on the rapid advancements in 2D, video, and 3D content generation fields, 4D generation has emerged as a novel and rapidly evolving research area, attracting growing attention. 4D generation focuses on creating dynamic 3D assets with spatiotemporal consistency based on user input, offering greater creative freedom and richer immersive experiences. This paper presents a comprehensive survey of the 4D generation field, systematically summarizing its core technologies, developmental trajectory, key challenges, and practical applications, while also exploring potential future research directions. The survey begins by introducing various fundamental 4D representation models, followed by a review of 4D generation frameworks built upon these representations and the key technologies that incorporate motion and geometry priors into 4D assets. We summarize five major challenges of 4D generation: consistency, controllability, diversity, efficiency, and fidelity, accompanied by an outline of existing solutions to address these issues. We systematically analyze applications of 4D generation, spanning dynamic object generation, scene generation, digital human synthesis, 4D editing, and autonomous driving. Finally, we provide an in-depth discussion of the obstacles currently hindering the development of the 4D generation. This survey offers a clear and comprehensive overview of 4D generation, aiming to stimulate further exploration and innovation in this rapidly evolving field. Our code is publicly available at: this https URL.