I didn't quite get what is being shown here. I think that different rows correspond to different path lengths, but are these samples from bidirectional sampling?
Is the data from different path lengths used together to render an image, or is one path length chosen at random to use?
Could you clarify what is happening here? The rows are different path lengths, and the columns is how many paths we take from tracing from the camera vs light source?
Why are images arranged into a pyramid shape? Is there a specific meaning attached to this?