It's probably just a matter of rerolling a few times. I was able to get it around 25% of the time.

https://imgur.com/a/H9gH3Zy

that's pretty good. I was using a cartoon girl as an example of a dance move for kids.

https://g.co/gemini/share/0e0de0d42029

I guess the vast majority of images have the palms the other way, that this biases the output. It's like how we misinterpret images to generate optical illusions, because we're expecting valid 3D structures (Escher's staircases, say).

Yes - it's the same reason generating a 5-leaf clover fails - massive amounts of training data that predisposes the model against it.