We’ve limited ourselves to a maximum of 100x100 tiles, though in practice most of our maps are 50x50.
I do disable/enable game objects depending on whether or not they are on screen. That is making some difference.
I have been considering just exporting the ground layer (this is the layer with the most tiles obviously, and doesn’t need any live z-position manipulation) and placing that in one big atlas. That way we’d save 2,500 (50x50) game objects and sprite components. I ran a test of this and performance was way better. Any reason why this would be a bad idea?