Comment by exortaz

this is insane - what’s your thinking on how this improves model grounding and efficiency vs single pov outputs?