Ruoyu Feng | ViLab https://vilab.team/author/ruoyu-feng/ Ruoyu Feng Hugo Blox Builder (https://hugoblox.com)en-usSun, 16 Jun 2024 00:00:00 +0000 https://vilab.team/media/icon_hu2896232876136423579.png Ruoyu Feng https://vilab.team/author/ruoyu-feng/ Microcinema: A divide-and-conquer approach for text-to-video generation https://vilab.team/publication/microcinema-a-divide-and-conquer-approach-for-text-to-video-/ Sun, 16 Jun 2024 00:00:00 +0000 https://vilab.team/publication/microcinema-a-divide-and-conquer-approach-for-text-to-video-/ <p>本文提出 MicroCinema，一种简洁而有效的文本生成视频框架。它采用分而治之策略，将任务分解为文本生成图像和图像与文本联合生成视频两个阶段，从而充分利用现有文本到图像模型的强大能力，生成逼真且细节丰富的图像，并让视频模型更专注于运动动态的学习。为高效实现该策略，文章设计了外观注入网络和外观噪声先验，以增强外观保持和视频连贯性，在多个基准上取得了优越性能。</p>