Image to image vs text to image
Text to image starts from nothing, so the model invents every detail. Image to image starts from your picture, so faces, products, poses and layout carry over and you only describe the difference. That makes it the right tool whenever the result has to look like something real: your product, your room, your pet or yourself.
When you need a brand-new scene, start with text to image and switch to image to image to refine the result.