Hi there! When I visualize the 'pos' sampled by the density_grid_sampler in line 70 in runner.py, I found that whatever aabb_scale is, pos ranges in [0, 1]. And when I set 'const_dt=True', pos ranges even smaller when aabb_scale is larger! It seems that there is something wrong with the aabb_scale or the density_grid_sampler. @Gword
Here are some pictures. The yellow cube is a unit cube. The white points are rays_o. The red/black points are pos.
👆aabb_scale=1
Hi there! When I visualize the 'pos' sampled by the density_grid_sampler in line 70 in runner.py, I found that whatever aabb_scale is, pos ranges in [0, 1]. And when I set 'const_dt=True', pos ranges even smaller when aabb_scale is larger! It seems that there is something wrong with the aabb_scale or the density_grid_sampler. @Gword
Here are some pictures. The yellow cube is a unit cube. The white points are rays_o. The red/black points are pos.
👆aabb_scale=1
👆aabb_scale=2
👆aabb_scale=16
👆aabb_scale=1, constdt=True
👆aabb_scale=8, constdt=True