search

TMElyralab / MuseV

MuseV: Infinite-length and High Fidelity Virtual Human Video Generation with Visual Conditioned Parallel Denoising
Other
2.1k stars 214 forks source link

有shutterstock水印 #119

Open QwenchC opened 1 month ago

QwenchC commented 1 month ago

生成的视频中下方有若隐若现的shutterstock水印，低分辨率的浅色视频比较显眼，这是为什么？

xzqjack commented 1 month ago

@QwenchC 因为musev目前开源的模型都是在webvid训练的哈，webvid数据集有水印，所以生成视频也容易出水印。相关讨论可以参考 #41