[spencerflem]

To be fair, if they first vocalize it they could just use Sparse Voxels which is my favorite differentiable rendering technique https://github.com/NVlabs/svraster