I've now made a width plan for the cloud layers so that I'll have something to refer to as I make them (width is calculated by distance from camera and speed of movement):
I then made a base colour layer for the sky ("sky background" in the above graph):
And the first (smallest/most distant) cloud layer (rendered on a white background - the final will be semi-translucent):
I've also made the silhouettes for 70 different buildings to be set on one "foreground" layer (these buildings will briefly obscure the main character) and three "background" layers. Each layer's buildings are a slightly different size and will pass by the camera at different speeds (in a similar fashion to the mountain and cloud layers. The following are just a few examples from the "first" background layer (the one closest to the camera without actually being in front of the character - those ones are more detailed, but the roofs are often outside the frame so they're less interesting to look at):
And last but not least, I made a detailed shot plan for the entire animation (many details of which can be dropped along the way, as the core focus of the rider runs throughout and is the only component that's strictly necessary, but this plan represents the full extent of the animation if it has time to be fully realised). I also finally settles on what song to use (Fatboy Slim's 'Love Island'), and have built this animation plan around it (cuts to reflect musical queues, movements
matching the beat, etc.):
