Every object in a scene is rendered individually, in every engine, yes. Objects are Culled (not graphically rendered) when outside of all of the layer camera’s current view, though.
Sure, if you have enough of them on screen at once, or enough logic impacting those off screen at once.
I don’t understand the question here “as a whole”, pixels are pixels. The more pixels you have to render, the more performance it has to take. It doesn’t matter if the pixels are part of object A or object B. So the number of objects don’t matter so much as the amount of logic that impacts or stores them (positioning, event logic, etc).
Tilemaps improve performance in some cases because the map is made up of reusable tiles from a single tileset rather than a bunch of huge images. It is a memory reduction, not necessarily an outright performance improvement.
As far as your piece of decoration being 600x600, the biggest issue is that means your decoration is 1/2rd the size of a 1920x1080 screen in height. Is that intended? (for example, you should not be designing your assets to be larger sized than shrink them down in the scene editor. There is no graphical improvement, and all it does is increase memory and cpu usage. If you want to design them at a larger size, that’s fine, but make sure you export them from your graphic system at the intended output size, not the original larger esize.)
So again, high level, objects themselves aren’t really a big ‘performance’ thing outside of memory (with caveats that if you have thousands on screen at once, it will impact performance). Objects aren’t rendered when they’re not visible in the current window (meaning all layer camera’s current view). What you’re doing with the objects, however, as far as events, will impact performance. Most of your optimization should be focused on events. If these are just background objects that aren’t touched/targeted by any events, they shouldn’t affect performance much.