In the theoretical game I imagine being made with all these pictures, you're meant to assemble a team pokemon-style. I'm doing multiple variations ("skins") when I can, but this raises issues I have with all games where multiple character skins are available.
Firstly it messes with character identification - silhouettes and colour are the easiest ways to ID a character and messing with that takes a toll. It's one of the many things I don't understand about dota-clones: in a game with 50+ different characters, why is anything that makes it harder to ID your opponent a good idea?
Secondly it makes groups look like an explosion in a paint factory. Everyone clashes. Compare TF2 at launch to TF2 now. Pokemon itself doesn't have a big problem here because each pokemon has a very limited number of colours and they all draw on a fairly limited palette. Also you don't see more than 2 of your pokemon on screen at once.
And I know this game will never be made and it will never be an issue, but it's a fun problem to address.
So, minimise a player's ability to customise their team. Just make it impossible to field Mr Pink, Mr Yellow, Mr Purple and Mr Green together. Drawback is that this reduces a player's ability to come up with cool-looking but unorthodox combos.
Another part of the problem is that with only three skins per figure, odds are good you can't find one that goes with your existing colour scheme.
So make more skins, or just more colour variations for each skin? That involves lots of work, multiplied across every skin I ever make. I could make it easier by homogenising the palettes, so I have a "the green palette" and a "the red palette" where each colour has a clear equivalent and I can just swap them about with no creative thought necessary. But from prior experience I know this won't work - requiring a colour pattern to look good with every available set of hues is (to put it lightly) not easy.
My frontrunner plan at the moment is colour themes. You get to choose the colour theme for your guild and every character skin is tagged to indicate what themes it is compatible with and you only get given skins compatible with your theme. For example yellow, orange, red and brown skins will all be tagged compatible with the "autumn" theme. But yellow would also be compatible with the "spring" theme. I could create new themes in the future and assess existing skins for compatibility much more easily than creating whole new skins.
This strikes me as something that will be harder in practice than it is in theory, but I'd like to at least try it out. A subtle problem with retconning incompatible skins out of the world is that it messes with character rarity - but since that's a system I've barely even thought about, I'm not fussed. There will be ways.
Big issue I see is if it means there are certain characters without a compatible skin for certain themes. To a certain extent I can make sure to create a few extra "night" or "winter"-compatible skins here and there if those themes need more love, but it's still an aesthetic choice with strong gameplay implications.
I'm not sure how I feel about that. First instinct is to run with it and see. I've a feeling it might end up being an interesting feature rather than a problem.