search

JokerYan / MS-GS-rasterizer

The rasterizer submodule for MS-GS
Other
0 stars 0 forks source link

Question about Implementation of Gaussian Aggregation #1

Open tommycwh opened 1 month ago

tommycwh commented 1 month ago

Thank you very much for your fascinating work! I am studying the implementation about Gaussian aggregation, which is described in the appendix of the paper "10. Theoretical Anti-aliasing Effectiveness of Gaussian Aggregation for 1D Signals".

I found that the level set and the pixel size of a Gaussian, as described in the excerpt below, is computed here: https://github.com/JokerYan/MS-GS-rasterizer/blob/e7a2b4f04c0980d462264f52032e6eba1abf4cd6/cuda_rasterizer/forward.cu#L248-L260

image

Furthermore, a little bit earlier in the same section, it is mentioned that multiple Gaussians are aggregated, but I could not find where this step is implemented.

image image

Would you mind showing me some hints about the related parts in the code?

Thank you again for your work.

JokerYan commented 1 month ago

Thank you for your interest, but please refer to the main repo for this part of the code: https://github.com/JokerYan/MS-GS/tree/main. The part is called in train.py L334 This repo is just a rasterizer submodule.