How to Create AI Selfie Videos with Celebrities That Look Shockingly Real

AI-generated selfie videos with celebrities are everywhere right now. You’ve probably seen people casually taking selfies with famous actors, singers, or athletes as if they were old friends. At first glance, it looks impossible. But once you understand the process, you’ll realize it’s actually very doable, even for beginners.

What makes this trend so exciting is how realistic the results can be. We’re not talking about stiff AI images that scream “fake.” These selfies look natural, cinematic, and genuinely convincing. The lighting feels real, the expressions look natural, and the whole scene feels like it was captured behind the scenes of a real production.

If you’ve ever thought, “I want to try this too,” you’re in the right place. In this guide, I’ll walk you through the exact steps used in the tutorial video. From generating a high-quality AI selfie image to turning it into a smooth video, everything is broken down in a simple and practical way.

Step 1: Generate the AI Selfie Image

The first thing you need is a solid AI-generated selfie image. This image will later become the foundation of your video, so don’t rush this step.

1. Use the Master Image Prompt

Start by taking the master prompt below. This prompt is designed to produce ultra-realistic, cinematic selfie images with famous people. The best part is that you only need to fill in one thing: the famous person’s name.

You are an expert cinematic image prompt engineer.

Famous person name: Fill here

TASK:

Generate a hyper-realistic image generation prompt.

The user will ONLY provide {FAMOUS_PERSON_NAME}.

You must AUTOMATICALLY choose the most iconic, realistic, and contextually appropriate filming location for that famous person

(e.g. movie set, concert stage, sports arena, fashion show backstage, interview set, red carpet, studio, or on-location shoot).

IDENTITY REFERENCE:

Use the uploaded image as a STRICT identity reference.

Preserve the exact facial features, face structure, skin tone, hairstyle,

and overall appearance of the person from the reference photo.

NO changes allowed.

ASPECT RATIO:

9:16 (vertical).

CAMERA & POV:

True smartphone front-camera POV,

as if the viewer IS the phone camera.

No smartphone visible.

No hands visible.

No reflections.

Natural selfie distance (arm-length),

slight wide-angle distortion typical of front-facing phone cameras.

SCENE (AUTO-GENERATED BY YOU):

The selfie takes place on a professional filming or production set

that logically fits {FAMOUS_PERSON_NAME}’s real-world persona and career.

The environment must feel authentic, not generic.

In the background (slightly blurred):

• professional film or production crew

• cinema cameras on rigs

• lighting stands, softboxes, reflectors

• cables, monitors, clapperboards

• strong behind-the-scenes atmosphere

MAIN SUBJECTS:

• The main person (from the reference image)

• {FAMOUS_PERSON_NAME}

They are standing very close together,

framed exactly like a real casual selfie,

smiling naturally,

looking directly into the camera lens.

RULES (STRICT):

• The main person must look EXACTLY like the reference image

• No face modification

• No beautification

• No age change

• Same hairstyle and face proportions

• Same casual modern outfit across all images

• Eye contact must be directed to the camera lens

• Faces fully visible, unobstructed

PHOTOGRAPHY STYLE:

Ultra-realistic smartphone selfie photography

Cinematic, professional film-set lighting

Sharp focus on faces

Shallow depth of field

Natural skin texture

High realism

No AI artifacts

FINAL OUTPUT:

Return ONLY the final polished image-generation prompt.

Do NOT include explanations.

Do NOT include alternatives.

Do NOT ask questions.

2. Fill in the Famous Person’s Name

Next, replace the Famous person name with anyone you like. For example, you can use Robert Downey Jr. Feel free to experiment later with other celebrities, actors, or public figures.

3. Copy the Entire Prompt

Once the name is filled in, select the entire prompt and copy it. Make sure nothing is missing, because every detail helps the AI produce a realistic result.

4. Generate the Final Image Prompt in ChatGPT

Open ChatGPT, paste the prompt, and hit send. ChatGPT will return a final, polished image-generation prompt. This is the prompt you’ll actually use to create the image.

Copy that output carefully.

5. Create the Image Using Google Gemini

Now head over to Google Gemini and choose the Create Image feature.

First, upload a clear photo of yourself. This photo will be used as a strict identity reference, so pick one where your face is fully visible and well-lit.

After that, paste the prompt you copied from ChatGPT and hit Send.

Within a few seconds, Gemini will generate a realistic selfie image of you standing next to the famous person you chose. The result usually looks incredibly natural, especially if the input photo is good.

6. Repeat for Other Celebrities

You can repeat all the steps above to create multiple selfie images with different famous people. There’s no limit here. The more images you have, the more creative your final video can be.

Step 2: Turn the Images into an AI Video

Once you’ve collected all your AI selfie images, it’s time to bring them to life with motion.

1. Open an AI Video Generator

You can use any AI video generator you like. Popular options include Google Flow, Dreamina AI, Kling, or Higgsfield. In this tutorial, Dreamina AI is used.

2. Create a New AI Video

Open Dreamina AI and tap the Create menu. Then select AI Video.

Choose the Video 3.5 Pro module, which is currently the latest and offers better motion and realism.

3. Set Video Parameters

Set the aspect ratio to 9:16, which is perfect for Shorts, Reels, and TikTok.

Choose your video duration. You can select 5 seconds, 10 seconds, or 12 seconds. In this example, 10 seconds works best.

4. Upload First and Last Frames

Tap the plus button and upload your first selfie image as the first frame.

Then upload another selfie image with a different famous person as the last frame.

This tells the AI how the video should start and how it should end.

5. Enter the Video Prompt

Paste the following video prompt into the prompt field:

in the first frame, the person takes a selfie with his idol. He then runs toward another set, as seen in the final frame. He meets his another idol and takes a picture with her/him. A handheld camera follows him throughout the entire sequence.

This prompt guides the AI to create a smooth transition between two selfie moments.

6. Generate the Video

Tap Send and wait a few minutes. The processing time depends on the platform and server load.

Once it’s done, your AI selfie video with famous people will be ready. The movement, camera shake, and transitions usually look surprisingly realistic.

7. Combine and Edit

If you generate multiple videos, you can edit and merge them using a video editing app like CapCut. Add music, text, or effects, and you’ll have a polished final result ready for social media.

Video Tutorial Reference

If you prefer learning by watching, you can follow the full video tutorial here:

Watching the workflow in action can make everything even clearer.

Conclusion

Creating AI selfie videos with celebrities is no longer something reserved for professionals or big studios. With the right prompts, a good reference photo, and modern AI tools, anyone can do it.

The key is to be patient and follow each step carefully. Focus on generating high-quality images first, because great videos always start with strong visuals. Once you master the process, you can experiment with different celebrities, scenes, and story ideas.

Most importantly, have fun with it. This trend is all about creativity and imagination. And who knows? Your next AI selfie video might be the one that goes viral.