Question for implement details

[yillis]

Sorry to bother you again. I encountered some confusion in the process of replication. I would like to ask about the generation strategy of the occlusion volumes mentioned in the paper. It seems that the paper does not detail how the occlusion volumes are generated, such as how many there are, whether they are uniformly distributed or generated close to the surface of the object...

[lzhnb]

Please refer to Plenoxel that place SH coefficients regularly in 3D space. We set bound = 1.5, res = 64 and bound = 16.0, res = 256 for TensoIR and Mip-NeRF-360 dataset respectively.
In details, we maintain a sparse structure (valid IDs + coefficients matrix) to implement occlusion volumes. The valid volumes are determined by the quantized position of Gaussians, and we dilate the IDs to fill up empty volumes.
The baking and recovery details are illustraed in Section 4.2 and supplmentary.

[yillis]

Please refer to Plenoxel that place SH coefficients regularly in 3D space. We set bound = 1.5, res = 64 and bound = 16.0, res = 256 for TensoIR and Mip-NeRF-360 dataset respectively. In details, we maintain a sparse structure (valid IDs + coefficients matrix) to implement occlusion volumes. The valid volumes are determined by the quantized position of Gaussians, and we dilate the IDs to fill up empty volumes. The baking and recovery details are illustraed in Section 4.2 and supplmentary.

OK，I'll try，Thank you！