search

ascust / 3DMM-Fitting-Pytorch

A 3DMM fitting framework using Pytorch.
600 stars 95 forks source link

why the 3DMM fitting face does not look like the input image? #18

Closed Randa-NACHET closed 2 years ago

Randa-NACHET commented 2 years ago

Thank you for sharing your work, please which parameters exactly correspond to obtain a good result ( preserved identity) and on what basis should I modify), because the output image does not look like the input (very smooth).

ascust commented 2 years ago

Hi, are you talking about the texture? or the geometry of the mesh? Since the optimized mesh is still a linear combination of the basis meshes, you would not expect an identical one.

lvZic commented 5 months ago

Hi, are you talking about the texture? or the geometry of the mesh? Since the optimized mesh is still a linear combination of the basis meshes, you would not expect an identical one.

hi, is there anything that could improve the identical?

ascust commented 5 months ago

@lvZic The quality is highly limited by the 3DMM model itself, which means if we cannot find a similar texture in the texture space, or shape in the shape space, we might not get a face that assembles the face in the image.