A question on the point sampling strategy

[Totoro97]

Hi thanks for this excellent project! It is really impressive.

I have a dumb question on the point sampling strategy. In the fox dataset, I set the aabb_scale as 128, which is a really large value.
And as shown in the visualization the occupancy grids are very dense (the image below). However, the sampled points number per ray is amazingly low (~80 points/ray).

For reference I also see that the samples per ray where aabb_scale=4 is ~70.

What confused me is that dramatically increasing the aabb_scale does not really affect the number of point samples. Is there any other point sampling strategy besides exponential stepping and occupancy grid skipping?

Thanks in advance!

[Tom94]

Hi there, yes, there is an additional termination of rays once they become opaque (i.e. transmittance gets close to zero). Cheers!

[Totoro97]

Thanks!