What do the GMI think about that makes them succeed at drawing more consistently and frequently that the NGMI fundamentally lacks?
Basically how do I build a step by step GMI neural network as a NGMI?

As an example, now I'm making this thread, I could be drawing but I suck, I don't know what to draw and how.
What am I supposed to do, just draw and feel bad? Already been there, should I just keep proceeding until I eventually like what I draw?