Jonathan, thanks again for the help. I really don't understand I'm sorry..
Would this piece of code be my add remove tiles? Maybe I should post some (semi) code.
RenderingEngine extends Sprite
Tile extends Sprite
RenderingEngine has a loop where it creates tiles to be added to the screen, and tiles to be removed from the screen.
RenderingEngine create an instance of Tile which is empty and place it on the stage (if the tile enters the screen)
RenderingEngine destroys the instances of Tile which are leaving the screen
RenderingEngine passes the frame and texture to instances of Tile (the ones that are entering the screen) which attach to itself an image or movieclip depending on its type, by calling a static asset class. It does it this way because Tile is linked to an AI which controls the movieclip, behavior of the Tile instance and also because I have a lot of tiles, which means I have to unload from memory as much as possible (long story short). But, the image and texture is cached in my Asset static class (stored into a dictionary and fetch from it if it exists, or create and store)
Now, in your example, image would get the position of the sprite, but if i put it in sprite, then it would have an offset. Considering my situation/architecture, would it be possible to use Quadbatch, or even should I try to use quadbatch?
Thanks for your help.