search

Anttwo / SuGaR

[CVPR 2024] Official PyTorch implementation of SuGaR: Surface-Aligned Gaussian Splatting for Efficient 3D Mesh Reconstruction and High-Quality Mesh Rendering
https://anttwo.github.io/sugar/
Other
2.36k stars 182 forks source link

question to to Eq.4 #16

Open yuedajiong opened 11 months ago

yuedajiong commented 11 months ago

"diagonal matrix with only one nonzero value on the diagonal (the other two being much smaller)" @ https://github.com/Anttwo/SuGaR/issues/2

‘flat’: only-one nonzero OR “at-most-two nonzero” ?

at-most-two nonzero includes: 2-nonzero: flat / plane 1-nonzero: thin / line

for example, a single Gaussian point, the scaling is x_y_z: [0.1, 0.2, 0.3] In Eq.4, does the s_g mean 0.1, and n_g mean 0 : x-axis?

Thanks

XOURTNEY commented 11 months ago

I think the author means Sg is really small, then 1/Sg would be bigger than the others. When considering to the scale matix, 1/sg would be the only number while others are ≈0. So the Sg means the scale along z-axis and Sg is 1e-5 or even smaller, ng means z-axis, and ng would be the 3rd roll of rotation matirx. not sure if i get the author right.