So much of 'how' you do things depends on your design, goals and location.
For me, I have a fairly large yard and two homes, so I get the controllers out as close to the big display items as possible, to minimize extension cord lengths. So in that design, I must run long power cords to the controllers, and then try to co-locate the controllers near things that use a lot of channels and then spider web out from those de-centralized controller locations. I also spread the load so that most controllers can be fed with a single extension cord, or two cords from the same circuit. I also put stuff on the roof, so by running just a few power cords up to roof-based controllers ... I limit my roof items to short ext. cords on the roof and very little hanging over the edges. I've found this to be a good system for me. I typically have been using 16 channels units like LOR or LE. Keep in mind you can also go wireless with LE, so thats another design option.
Anyways, my color changing has been the low-tech and cheap way. Its by taking red, green and clear 100ct minis and putting them on virtually everything. Then I can turn each item red, green or clear or a combination of the three. The interesting thing about RGB is that you can go with string level control (whole string any color you want) ... or pixel level control (every pixel can be programmed independently).
The first is powerful and easier to sequence. The second is more powerful but way more to sequence as well.
For me, I'd benefit by eventually having just string level RGB, because I could install just one RGB string and then get any color I want out of it. Of course, doing any color still isn't the same as having a ton of clear and a ton of red on at the same time ... but its a choice ... and there are pros and cons of course.
Anyways, I'd keep in mind placement of controllers, extension cord use (cause they are expensive), power distribution and your design goals for your show and expansion 1,2,3+ years into the future. THe more planning you do now, the less you'll have to redo later (in theory).
So capability, cost, complexity ... are all things to consider.
Good luck,
Scott