AI 3D Trompe L'Oeil Art Generator

The Interface Style menu has Instagram Interface (the default), TikTok Interface, Facebook Interface and Generic Social Media. You can also type your own value of up to 50 characters, such as "YouTube" or "Xiaohongshu", and the phone screen is drawn in that app's look. It only imitates the layout: Dreamega isn't affiliated with Instagram, TikTok, Facebook or Meta, and if you put a platform-like screen in an ad, clearing that is your responsibility.

No. There is no field for a username, caption or numbers. The AI draws the like, comment and share buttons and the follower, view or reaction counts on its own, so the text is usually invented, sometimes misspelled, and won't match your account. If you need your real handle or a real milestone number on the screen, add it afterwards in an editor like Canva or Photoshop.

In the result your subject steps out of the phone with one leg past the frame, so a full-body photo, or at least one that reaches the knees, works best. With a head-and-shoulders selfie the AI has to invent the legs and clothes, and that's where odd proportions creep in. Upload a sharp JPG or PNG (WebP isn't accepted) with one person, even lighting and nothing blocking the body.

Pets and products work: a dog climbing out of the screen or a sneaker breaking past the frame both come out well. The prompt describes a person's pose, though, so with an object the AI improvises and results vary more. Group photos are unreliable. The layout is built around one figure, and two or three people often get merged, cropped or shrunk. Crop to a single subject for each run.

GPT Image 2.5 is the default at 8 credits per image. It always returns a square 1:1 picture and is generally better with small text and icons. Nano Banana 2.1 costs 4 credits and is a cheap way to test a photo first. New accounts get 10 free credits, plus 2–6 from the daily check-in. If the hearts, emoji faces or numbers come out smeared, run it again; each run is a fresh draw, and switching models often helps.