The problem is that the sheep can’t really stand on two legs. It looks really… ridiculous.
So you’re basically limited to four legged animations.
I’m not too sure about combined animations either. While yes, technically it would be possible to organise them… rather easily even thanks to the AI system, I’m not quite sure how well that would work. It would be a lot of work to animate and tune it, that’s for certain.
Basically, you would need a system that organises these things, i.e. sheeps can enter a pool to form a pyramid and once they’ve found enough members they gather at a location, perform the stunt and then go back to their normal behaviour.
I don’t think it’s very ravey either. I was thinking about giving them silly hats or glasses (such as the one I was wearing in the original picture) or maybe some sort of glowstick. But then again they would need to carry those in their mouth and I’m not sure how well you could create fancy looking glasses here.
Transparency would be the big issue; the glasses would likely need to be rendered with a different shader or something, while not completely impossible would require another entity, which needs to be re-positioned to the bone and… I’m not sure if those kind of attachments are already implemented, or if certain models can be rendered with a different shader. I doubt both of it, the latter more than the former.
So, the whole thing has lots of potential I guess, but not at this stage yet.