How to make a realistic AI video of yourself
The photo, the light and the length that keep your face and build, the frames where fakes show, and what went wrong in our own tests.
Checked on
All answersDrakeAIVideo📌 Short answer
Start from one sharp, evenly lit photo of you alone, waist up or full length, with no beauty filter; keep the clip short; then check the frames where AI gives itself away: the first second, every turn, the hands and anything you hold. The face comes from the photo, so the photo decides most of the realism. On this site one photo becomes 15 seconds of the Burning Bridges party with you in the lead, in about 4 to 15 minutes.
Your seat is open
One photo is enough. Watch a free preview of yourself at the table before you pay anything.
The photo stays in this browser until you press the button at the seating desk.
What makes it look like you
| Part | Comes from | Keep it right by |
|---|---|---|
| Face | Your photo | Sharp eyes, facing the camera, no filter |
| Hair and build | Your photo | Showing your shoulders, ideally your whole body |
| Clothes | Your photo | Wearing what you want to wear in the video |
| Moves and timing | The reference clip | A reference with one clear lead |
| Light and skin | The model, matched to the scene | Soft, even light in the photo, so it relights cleanly |
How this site keeps your likeness
Your photo is first redrawn as one standing, full-length portrait, which gives the model your build and clothes from head to toe. The video model then gets two images: the portrait for the body and your original photo for the face, with the instruction to keep the shape of your face, your eyes and your beard as they are in your photo, not the original performer's. If the portrait step fails, the clip is made from your photo alone.
What went wrong in our own tests, and the fix
Our paid tests showed where likeness holds and where it slips. In a 10-second test on 16 October 2026, with a man in a plaid shirt, the face, the shirt and the cup all held, even when he turned his back to the camera. The misses, and what we did about them:
| What we saw | When | Why | What changed |
|---|---|---|---|
| A head wrap for under a second at the start | Paid test, 15 Oct 2026 | The reference had one and our instructions named it | We stopped naming it; the subject wears nothing on the head |
| A leather jacket over the shirt | Same test | The same reason | Only the clothes in the photo; the next test kept the shirt throughout |
| A wristband the photo did not have | Second test, same day | The model filled in an accessory | Not fixed: small items can still appear |
| A mouth that did not follow the words | Mid-October | No song went to the model | Since 16 October the song goes in with the moves |
Check these frames before you post
- 1
The first second: the original performer's hair, hat or outfit can flash in.
- 2
Every turn: profiles and the back of the head are invented, so look for a new hairline.
- 3
The hands: count the fingers and check that rings and watches match yours.
- 4
Anything held: cups and phones can melt into the hand in fast moves.
- 5
The last second: faces drift most toward the end of a clip.
Realistic, and still labelled
The more real it looks, the more it matters that viewers know. Switch on the platform's AI label, use only your own face or an adult who agreed, and leave public figures out (this site blocks them). The rules in more detail: is it allowed to post a Drake AI video? Try it with the free 2-second preview; a full clip is 10 credits at 480p or 22 at 720p.
Your chair is still warm
Take the seatFAQ
Questions guests ask
Can I create an AI video of myself?
Yes. Upload a clear photo to a photo-to-video or recast tool. Here one photo makes a 15-second Burning Bridges clip with you in the lead.
How do you make realistic human AI videos?
Start from a sharp, evenly lit photo, keep the clip short, let a real reference clip drive the motion, and redo clips where the face or hands drift.
Why does the AI video not look like me?
Usually because of the photo: too small, filtered, side-on or with other people in it. Try a sharp, front-facing photo of you alone.
Do I need to record a video of myself?
Not here: one photo is enough. Talking-avatar tools often ask for a short recording instead.