so let's fix a particular 2d null subalgebra of the split octonions, and consider the 2d g2 schubert variety based at it, and try to make explicit a nice map down to the space of axises through the rolling ball... or something like that...