How did you determine how many Gaussians require which SH band?

[yannik-p]

In your paper, you stated that "For our main model, the point distribution over bands was 89%, 0.1%, 2.7%, 8.2% for 0, 1, 2, 3 bands respectively."
However I cannot find any part in the files that deals with evaluating the SH-band distribution.

[PanagiotisP]

Hello!
That's just us reporting the results. The way the culling happens is with the thresholds. It happened that throughout all the datasets, the SH band distribution was the one we mention in the paper