Changing Image.Sprite at runtime vs having one gameobject per sprite.

I'm making a HUD for currently active weapon and I'm wondering what would be more performance efficient: having a one GameObject with Image component and changing its sprite via script (`Image.sprite`) at runtime or having one GameObject per image and activating/deactivating them.

 

Neither will have any real significant impact on performance. I'd personally go with just one GO with an array or list of possible sprites you can swap to, as it's more straight forward on how to add or remove ones during development.