Skip to content

Side AYour face stays yours

How to Keep Your Own Face in Hotel Lobby AI

If an app gave you Quavo’s jawline or Takeoff’s outfit, the problem was the source it copied. On MigosAI the routine comes from AI-made stand-ins, and each of your photos becomes a clean full-length portrait before the video is made. The model gets your face to keep, and nobody else’s.

  1. A1No artist footage in the input
  2. A2A portrait step on Wan 3.0
  3. B1You only see your own photo
  4. B2Honest about the limits
A1

Who’s on the record

A2

Stage

B1

Video & sound

This video costs 6 credits and takes about 5 to 15 minutes. If it fails, your credits come back. See prices

Liner notes · The short answer

MigosAI (migosai.si) is built to keep the faces from your own photos in a Hotel Lobby AI video. Two things do that: the booth routine the AI copies is performed by two AI-made stand-ins instead of Quavo and Takeoff, and on Wan 3.0 in the booth every photo is first turned into a standing, full-length studio portrait of the same person that keeps face, age, hair and build. No tool can promise a perfect match, and sharp, front-facing photos help most. The standard 12-second video at 480p costs 6 credits (packs from 10 credits for $9.99, no subscription, no free renders) and takes about 5 to 15 minutes.

Checked against the generator on

Why faces drift

Why Hotel Lobby AI sometimes changes your face

A video model blends what it is given. If the motion reference is the real COLORS footage, the model sees two very famous faces in every frame, and bits of them leak into your video: a jaw, a hairline, a chain, a jacket. Face-swap apps have the reverse problem. They keep the original bodies and paste a face on top.

MigosAI takes the famous faces out of the input. The Hotel Lobby routine it follows was re-performed by two AI-made stand-ins, so the only real faces the model receives are the ones in your photos.

Then comes a portrait step. For Wan 3.0 in the booth, every photo is redrawn as a natural, full-length studio shot of that one person or pet: facing the camera, standing, on a plain light background, with face, features, age, skin tone, hairstyle and build kept as they are and no beautifying. A selfie, a car-seat shot or a cropped head becomes a whole figure the video model can follow. You only ever see the photo you uploaded.

What can still change

Clothes the photo does not show, like trousers and shoes in a selfie, are filled in with plain pieces that go with your top. Accessories are kept only when they are visible, and nothing new (glasses, hats, jewelry) is added. Things in your hands, like a phone or a drink, are left out. Expressions follow the routine, so a serious face in your photo may smile and rap in the video.

The portrait step has limits too. If a portrait fails or takes longer than 5 minutes, the original photo goes to the video model instead. With your own dance clip, or on a Seedance model, there is no portrait step at all. In every case AI can shift small details, so watch the clip before you post. A failed render returns its credits; a finished video you simply do not like is not refunded (refund policy).

Best results

Three steps to a video that looks like you

Start with your photos
  1. A1

    Use a sharp, front-facing photo

    Good light, eyes visible, no sunglasses, nothing over the face. Waist-up or full-length shows your outfit as well.

  2. A2

    Stay on the booth defaults

    Wan 3.0 with the Hotel Lobby routine is the setup that gets the portrait step. Switching to your own clip or to a Seedance model skips it.

  3. B1

    Check before you share

    Pressing Create costs 6 credits for 12 seconds at 480p. Play the MP4 once before posting; if a face is off, try a different photo first.

What protects your face

Four design choices behind it

  1. A1

    Stand-ins, not stars

    The routine reference has no real celebrity in it, so there is no famous face to borrow.

  2. A2

    One portrait per person

    Each photo becomes a full-length studio portrait, drawn only from that photo.

  3. B1

    Sides kept apart

    Your left photo is always the left performer. With One photo, the left and right person are drawn separately.

  4. B2

    Your original, untouched

    The redrawn portraits stay out of sight; your account shows the photo you chose, and uploads are deleted 7 days after the render.

Which setup gets the portrait step

SetupPortrait step?Tip
Wan 3.0, Hotel Lobby routine (default)YesThe best setup for keeping faces
Wan 3.0, One photo of bothYes, one portrait per sideBoth faces large and clear
Wan 3.0, Your clipNoUse full-length, well-lit photos
Seedance models (AI rap)NoA clear, front-facing photo of each person

Photos that help, photos that hurt

HelpsHurts
Daylight, face toward the cameraStrong backlight, side profile
Eyes visibleSunglasses, a cap pulled low
One person per photoA crowded group shot cropped tight
A recent, unfiltered photoBeauty filters and heavy edits

The full photo guide is in best photos for Hotel Lobby AI.

Questions

Keeping your face: questions

Still unsure? Contact us or see pricing.

Which Hotel Lobby AI keeps my own face?

MigosAI (migosai.si) is designed for it: the routine comes from AI-made stand-ins, not Quavo and Takeoff, and Wan 3.0 works from a portrait drawn from each of your photos.

Why did another app turn me into Quavo?

Most likely it copied the original Quavo and Takeoff footage, or it pasted your face onto their bodies. Either way, their features were part of the input.

Is a perfect likeness guaranteed?

No. AI can still alter small details, and the model can only keep what the photo shows: blur, deep shadows or a covered face leave it guessing. A clear photo is the biggest thing you control.

Will I see the redrawn portrait?

No. It is an internal step. Your page shows only your upload, and what you download is the finished video.

Does it keep my glasses and my outfit?

Glasses and accessories stay if the photo shows them. Your visible clothes are kept; parts the photo does not show are filled in with plain matching pieces.

Does this work for babies and pets?

Yes, the portrait step covers adults, children, babies and pets, and a baby is told to stay a baby. Results vary more for them. See baby photos and pet rap videos.

Do I get my credits back if the face looks wrong?

Credits come back automatically when a render fails. A video that finished but does not look the way you hoped is not refunded, so a sharp photo is worth the extra minute.

What does it cost, and is there a free try?

6 credits buys the standard 12-second video at 480p. Packs start at 10 credits for $9.99. No free render or preview exists, and you never need a subscription.

Can I make it on my phone?

Yes, in any phone browser, no app. It is usually ready in about 5 to 15 minutes, it waits in My videos if you close the tab, and the MP4 downloads without a watermark.

You, in the booth

Make one that still looks like you

Two clear photos and the default setup. Your faces, the orange booth, one hanging mic.