search

cleardusk / 3DDFA_V2

The official PyTorch implementation of Towards Fast, Accurate and Stable 3D Dense Face Alignment, ECCV 2020.
MIT License
2.9k stars 514 forks source link

Ear, Neck and Landmark Visibility #10

Closed RegisWu closed 4 years ago

RegisWu commented 4 years ago

Hi,

Thanks for sharing the great work!

Here I have a few questions: 1) In the old 3DDFA (3DDFA_V1), the reconstructed vertices cover the ears and neck regions. But it is excluded in the new 3DDFA_V2. However, in some applications we would like to get the details around those regions, may I know is there an easy way to adapt the new 3DDFA_V2 model to cover the ears and neck regions?

2) In the old 3DDFA paper (CVPR ver.), the authors visualized the visibility of each landmark. I am wondering how to get the landmark visibility from the reconstructed vertices, could you please give me some hints about it?

Looking forward to your reply! Best regards.

cleardusk commented 4 years ago
  1. I will consider releasing the version with neck regions.
  2. You can calculate the normal vector or use z-buffer to judge it.
RegisWu commented 4 years ago

Noted with thanks!

cleardusk commented 4 years ago

@RegisWu The version with neck: BaiduYun, passwod: vd2t

RegisWu commented 4 years ago

@cleardusk Many thanks! Will test it later!

IQ17 commented 4 years ago

Hi, for the 68 landmark case, how is this z-buffer calculated? Would you mind show an example ? Thanks!

cleardusk commented 4 years ago

@IQ17 You can define the topological mesh for 68 landmarks, then do z-buffer to check the visibility. I think the solution is obvious and you can dig into the z-buffer code to implement it.

cleardusk commented 4 years ago

For anyone using the bfm version without neck, you should replace the tri by bfm.tri since tri is for no-neck.

ghost commented 3 years ago

@cleardusk how to generate the 3D render picture with color?

kvrooman commented 3 years ago

Came across this neck model and was interested in trying it out. Would you happen to have a non-Baidu file as Baidu states that registration / sign-up is currently not supported outside mainland China...

TouqeerAhmad commented 1 year ago

@cleardusk Can you please direct to a reference code for getting normal map from shape emerging from 3DFFA/3DFF_V2. For example in this paper, authors comparing against Normal Maps emerging from 3DDFA, but not clear how they generated normal maps from shape as their repo is not complete. https://openaccess.thecvf.com/content_CVPR_2020/papers/Abrevaya_Cross-Modal_Deep_Face_Normals_With_Deactivable_Skip_Connections_CVPR_2020_paper.pdf

chenjun2hao commented 1 year ago

@RegisWu The version with neck: BaiduYun, passwod: vd2t

could you share the code producessed the bfm.pkl file. thanks!