search

chiehwangs / gaussian-head

Official repository for 'GaussianHead: High-fidelity Head Avatars with Learnable Gaussian Derivation'
MIT License
253 stars 16 forks source link

Some questions about the code. #26

Closed gaga0714 closed 2 months ago

gaga0714 commented 6 months ago

Hi, when I use your model for training, there report an error: The size of tensor a (9) must match the size of tensor b (3) at non-singleton dimension 1 image I wonder if this line of code may be changed from if input.shape[1] != 3: to if input.shape[0] != 3:? image I really hope you can give me an answer, thank you!

ZardYuan commented 4 months ago

I guess the shape of render-output is (c,h,w), while the shape of vgg_loss input is (n,c,h,w). Maybe you can unsqueeze the render-output before cal vgg_loss.

chiehwangs commented 2 months ago

Many people have successfully reproduced the rendering process, so I think there is nothing wrong with the code.