Text to Video Generation
Text-to-video generation aims to create videos from textual descriptions, bridging the gap between human language and visual media. Current research heavily utilizes diffusion models, often incorporating 3D U-Nets or transformer architectures, and focuses on improving video quality, temporal consistency, controllability (including camera movement and object manipulation), and compositional capabilities—the ability to synthesize videos with multiple interacting elements. These advancements hold significant implications for various fields, including film production, animation, and virtual reality, by automating video creation and enabling more precise control over generated content.
Papers
AIGV-Assessor: Benchmarking and Evaluating the Perceptual Quality of Text-to-Video Generation with LMM
Jiarui Wang, Huiyu Duan, Guangtao Zhai, Juntong Wang, Xiongkuo Min
Free$^2$Guide: Gradient-Free Path Integral Control for Enhancing Text-to-Video Generation with Large Vision-Language Models
Jaemin Kim, Bryan S Kim, Jong Chul Ye