will open the training code?

[foocker]

统一拓扑结构得到了约2000个不同性别、年龄、种族的人脸mesh，在此基础上，构建了一个全头人脸3DMM并命名为HIFI3D++. all these data preprocess script will be release?

[czh-98]

Thanks for your interest. Unfortunately, we have no plans to release the process scripts at present.

[RujingD]

Thanks for your interest. Unfortunately, we have no plans to release the process scripts at present.

Hey，have you released the code for converting the HIFI3D++ .mat file to .obj ?

[czh-98]

Thanks for your interest. Unfortunately, we have no plans to release the process scripts at present.

Hey，have you released the code for converting the HIFI3D++ .mat file to .obj ?

Please refer to the HIFI3D repo about generating and saving the shape: https://github.com/tencent-ailab/hifi3dface/blob/82eb80bcb08ecf7e022b9382f41b11e6e38ddf57/3DMM/scripts/test_basis_io.py#L185.

You can replace "AI-NExT-Shape.mat" with our "HIFI3D++.mat" to generate the new shape according to the above script.

[JiejiangWu]

Thanks for your interest. Unfortunately, we have no plans to release the process scripts at present.

Hey，have you released the code for converting the HIFI3D++ .mat file to .obj ?

Please refer to the HIFI3D repo about generating and saving the shape: https://github.com/tencent-ailab/hifi3dface/blob/82eb80bcb08ecf7e022b9382f41b11e6e38ddf57/3DMM/scripts/test_basis_io.py#L185.

You can replace "AI-NExT-Shape.mat" with our "HIFI3D++.mat" to generate the new shape according to the above script.

Hi, I tested the code, and found that ["vt_list"], ["tri"] and ["tri_vt"] are not in the keys of basis3dmm, which is loaded from HIFI3D++. So the script line 229

write_obj(
"rand_shape.obj",
ver_shape,
basis3dmm["vt_list"],
basis3dmm["tri"],
basis3dmm["tri_vt"],
"face.mtl",
)
returns error

[czh-98]

Thanks for your interest. Unfortunately, we have no plans to release the process scripts at present.

Hey，have you released the code for converting the HIFI3D++ .mat file to .obj ?

Please refer to the HIFI3D repo about generating and saving the shape: https://github.com/tencent-ailab/hifi3dface/blob/82eb80bcb08ecf7e022b9382f41b11e6e38ddf57/3DMM/scripts/test_basis_io.py#L185.
You can replace "AI-NExT-Shape.mat" with our "HIFI3D++.mat" to generate the new shape according to the above script.

Hi, I tested the code, and found that ["vt_list"], ["tri"] and ["tri_vt"] are not in the keys of basis3dmm, which is loaded from HIFI3D++. So the script line 229

write_obj( "rand_shape.obj", ver_shape, basis3dmm["vt_list"], basis3dmm["tri"], basis3dmm["tri_vt"], "face.mtl", ) returns error

Thanks for your reminder. The missing information has been updated. Please download the new version of HIFI3D++.mat. Note: remove the line: "face.mtl" to ensure it can run successfully.

[GrigorySamokhin]

Hello! Thanks for ypur work, im interesting, is it possible to use HIFI3D++ with HIFI3D code optimization?

[haoxianzGit]

Hello! Thanks for ypur work, im interesting, is it possible to use HIFI3D++ with HIFI3D code optimization?

yes, but you may need some adaptation modifications.
please refer to #4.