Hello, thank you for your open-source contribution.
This work has been very enlightening for me. While studying the project, I noticed that for certain 2D mask groups lacking corresponding Gaussian points, the code initializes primitive Gaussian points at (0,0,0). What is the purpose of these points? To my understanding, the L_obj loss fundamentally masks off gradient optimization for Gaussian points outside the mask. I hope you can correct me if I'm mistaken.
Hello, thank you for your open-source contribution.
This work has been very enlightening for me. While studying the project, I noticed that for certain 2D mask groups lacking corresponding Gaussian points, the code initializes primitive Gaussian points at (0,0,0). What is the purpose of these points? To my understanding, the L_obj loss fundamentally masks off gradient optimization for Gaussian points outside the mask. I hope you can correct me if I'm mistaken.