hi, i wonder are there any future plans to develop voxel rendering further ? if we could control the voxel size on a per particle basis using channels that would be very very useful 
I was throwing around the idea of variable voxel density by distance, I swear its probably a ton of work but if you could define voxels size by how close it is to, say your camera, even if it where somewhat rudimentary, distant voxels being larger calculate/draw faster and reserve the overhead for closer to camera voxels where you want more detail with a higher voxel density. Basically like a Fume wavelet by distance.
We were in fact discussing the future of Voxel rendering this week. I cannot share any details about that discussion.
But I remember from reading papers years ago that there was a nice concept of a frustum-aligned voxel grid where the camera’s cone is subdivided into non-cubic voxels that get larger with the distance. This way close up voxels would be as small as needed, and farther voxels would become larger and larger, while ensuring that there is no gap between the voxels anywhere.
We cannot do arbitrarily sized voxels though because there are certain rules about how a grid can be subdivided. We could (and we do in PRT Volume and Repopulation for example) subdivide voxels into sub-regions using equally spaced subdivisions (1,2,3 etc.), but we cannot have a voxel with size of 1.0 sitting next to a voxel of size 1.2 for example…