The large scenes looks like about 4 screens wide? But they're not full height - looks like about 2/3, so let's say ~24KB total including color data. I don't think it should be a problem. The walk + scroll is slow enough that if you had to (and I don't think you do) you ought to be able to time things so you can load the next screen while the player is walking.

Similarly, e.g. slow down the door animation, and a fade, and you ought to have enough time for a decent fast loader to load the next screen (~2-3 seconds assuming you're loading 2/3 of the screen)

You really benefit from the low amount of action on screen here.

If you want to actually compress the data to reduce the number of floppies, you'd slow it down quite a bit. If you were doing it for a real C64 or cartridges constrained to what was viable at the time, that might well be preferable to more floppies. If you're doing it for a modern cartridge or an emulator, it won't matter.