search

LizhenWangT / NormalGAN

NormalGAN: Learning Detailed 3D Human from a Single RGB-D Image (ECCV 2020)
72 stars 13 forks source link

about the depth information #2

Closed jeffreyking23 closed 3 years ago

jeffreyking23 commented 3 years ago

Hello, amazing work, really love it, because i don't have a kinect, can I use the result of sigle view depth estimation methods instead of kinectV2 depth?

LizhenWangT commented 3 years ago

Thank you for your interest on our work! We haven't test our method on depth estimation methods yet. But I think if you can generate high-quality predicted depth maps, NormalGAN might still work. Actually, FACSMILE and Moulding Humans shows the ability to generate full body from a single RGB image. However, it might be much more difficult to inference the invisible geometric details.