Open Zvyozdo4ka opened 1 month ago
Hi, do you mean reconstrucing 3D hair from video/image and then extracting mesh or point cloud from the output? If so, the answer is yes as our strand output is basically a connected point cloud.
Hi, do you mean reconstrucing 3D hair from video/image and then extracting mesh or point cloud from the output? If so, the answer is yes as our strand output is basically a connected point cloud.
thank you for your answer!
is the strand output a ply file? is it possible to see reconstructed hair output and corresponding video/image?
Hi, sorry for my late reply. For example, for this input image, I uploaded the reconstructed body mesh from DELTA and reconstructed hair strands from our method to https://drive.google.com/drive/folders/1CN1xHUsfYkmBSz0dG7VZWeRAuEsybn6q?usp=drive_link. You can import these files into software like Blender to visualize them.
@c-he Thank you for your response.
But could you explain how to get 3d mesh of the input image?
As I understood from github page i should choose the target, which is with the closest geometry to the input photo?
is it possible to get 3d output without knowing what the target is, just to reconstruct based on photo, like hairstep does?
And what is the 'seeds'?
We do support single-view hair reconstruction, see Section C.5 of our supplemental. However, we haven't released this part of code yet due to license issues.
Can you share some of the reconstructed hair outputs? What would it look like?
And is it possible to extract mesh or point cloud from running your project? or only video?