TimoBolkart

Results 84 comments of TimoBolkart

FLAME in the public form has no dedicated blendshapes to add eyeblinks but by optimizing the expression parameters, FLAME is able to blink. Commonly, a too high regularization of the...

@pcgreat Neither BFM nor ICT are our models, so we cannot provide anything to convert between these models. What we do have and what we could share is something to...

@xiaolin-mt I am not sure if we are allowed to share BFM meshes. I will check

Sorry, I missed that. Deep3DFace uses BFM as underlying model. Do you have the vertex indices of those cropped faces?

How do you get the custom basel face model? By excluding / masking some vertices from the original BFM model?

Instead of the cropped face of NextFace, use its predicted parameters and reconstruct the full, non-cropped BFM2017 head mesh. This you can then directly convert with the provided demo code

What exactly do you intend to do? The col_to_tex.py takes the albedo space of BFM (represented as colors per vertex) and converts them to the FLAME texture layout. The result...

The converted texture space has no color difference in the face and it looks like this: ![image](https://user-images.githubusercontent.com/26549953/232477520-40554192-8241-4e57-8abe-f5f7d926a773.png) What color difference are you referring to here?

That is the visualization of the `col_to_tex.py` script that produces the FLAME_albedo_from_BFM.npz. I am not sure why the DECA output has only the face colored.