AI 3D Pop-Out Effect

Full body works best. The effect puts your subject standing on a phone screen, so the AI needs legs and feet to plant on the glass. With a headshot it has to invent the rest of the body, and that is where proportions go wrong. Use a JPG or PNG with the whole person in frame, feet included, and a fairly plain background so the outline is easy to separate.

Yes. A standing dog, a sneaker or a bottle shot on a plain background all work, and product shots are the easiest inputs. The subject gets shrunk to fit on the phone, so small labels and logos can blur or get redrawn; check them before posting. You can generally use the image in ads and social posts, but you are responsible for the rights to your photo and to any brands or people in it.

The hand and phone are generated fresh every time, and hands are still a weak spot for image models: an extra finger, a bent phone edge, or a subject that floats instead of casting a shadow on the screen. Run it again, or switch between GPT Image 2.5 and Nano Banana 2.1. A photo where the feet or the bottom of the object are clearly visible helps the model place the shadow correctly.

The Background Style menu has four presets: Neutral Clean (the default), Office Environment, Modern Interior and Gradient Background. You can also type your own scene of up to 50 characters, such as "beach at sunset" or "neon city street". The background sits behind the hand and phone and does not replace them. Every result shows a hand holding a phone, and the output is a still image, not an animation.

GPT Image 2.5 is the default and costs 8 credits per image; it always returns a 1:1 square. Nano Banana 2.1 costs 4 credits and is a cheap way to check whether a photo works before spending more. New accounts get 10 free credits at signup, plus 2 to 6 credits from the daily check-in (those expire after 3 days), so you can try a few photos without paying.