Temporally Consistent Relighting for Portrait Videos
Temporally Consistent Relighting for Portrait Videos
复制标题
DOI：
10.1109/wacvw54805.2022.00079
复制
发表时间：
2022-01
期刊：
影响因子：
--
通讯作者：
Sreenithy Chandran;Yannick Hold-Geoffroy;Kalyan Sunkavalli;Zhixin Shu;Suren Jayasuriya
中科院分区：
文献类型：
--
作者：
Sreenithy Chandran;Yannick Hold-Geoffroy;Kalyan Sunkavalli;Zhixin Shu;Suren Jayasuriya
在录制人士视频时，确保理想的照明可能是一项艰巨的任务，需要受控的环境和昂贵的设备。最近，提出了对静止图像进行肖像重新保证的方法，从而实现事后照明增强。但是，在每个框架上天真地应用这些方法会独立地产生困扰着闪烁的文物的视频。在这项工作中，我们提出了执行时间一致的视频肖像重新保证的第一种方法。为此，我们的方法共同优化了所需的照明和时间一致性。我们在培训期间不需要地面真相照明注释，使我们能够利用互联网上已经可用的大量肖像视频。我们证明，我们的方法在平衡许多现实世界肖像视频上的准确重新确定和时间一致性方面优于先前的工作。
Ensuring ideal lighting when recording videos of people can be a daunting task requiring a controlled environment and expensive equipment. Methods were recently proposed to perform portrait relighting for still images, enabling after-the-fact lighting enhancement. However, naively applying these methods on each frame independently yields videos plagued with flickering artifacts. In this work, we propose the first method to perform temporally consistent video portrait relighting. To achieve this, our method optimizes end-to-end both desired lighting and temporal consistency jointly. We do not require ground truth lighting annotations during training, allowing us to take advantage of the large corpus of portrait videos already available on the internet. We demonstrate that our method outperforms previous work in balancing accurate relighting and temporal consistency on a number of real-world portrait videos.