Jump to main content
Hotel Lobby Video

Best Photos for a Hotel Lobby AI Duet: Casting Checklist

The best photos for Hotel Lobby AI show one person each, facing the lens in decent light, ideally full length and standing, with eyes uncovered and hands empty. On Wan 3.0, every portrait is first redrawn into a standing studio portrait of the same person, and anything your photo leaves out has to be made up by the model.

Treat it like casting: the camera only gets what you hand it. The checklist comes from how the studio processes photos and from quirks the video models are known to show. A side-by-side comparison of good and weak photos is not published yet.

Last revised

A selection of clear adult portrait prints, with a loupe and pencil.

HOTEL LOBBY AI · PHOTO GUIDE

Pick photos that keep the face

  • One per photo
  • Head to toe
  • Eyes visible
AI concept illustration · Hotel Lobby Video

The five-question casting test

  1. Is there exactly one person in the frame? If a friend’s shoulder sneaks in, crop it out.
  2. Can you see both eyes clearly, with no sunglasses, mask or low cap?
  3. Does the shot reach at least the waist, ideally the shoes?
  4. Is the person standing, mouth closed or softly smiling?
  5. Is this the original file rather than a screenshot or a compressed chat copy?

Five yeses and the portrait is ready for the orange set. Every no is a gap the model will fill with a guess.

Two clear reference portraits of fictional adults. A photo-selection diagram, not a comparison test of generated results.
Two clear reference portraits of fictional adults. A photo-selection diagram, not a comparison test of generated results.Our illustration, not a real video frame · 2026-10-05

Casting done?

Put your pair on the orange set

One face per slot: the left takes the verse, the right backs it up. The button shows the credits before anything is charged.

1

Your two faces

A single face per picture. Left leads, right backs up.

2

Set

3

Take and sound

Price: 6 credits. Done in roughly 5 to 15 min; if it fails, the credits come back. See prices

Swap these shots for better ones

If your photo isTry insteadThe reason
A group shotA crop around one personA second face forces the model to choose who to copy.
A side profile or backlitFace-on, even lightThe face is the part that must survive the redraw and the motion.
A tight selfieWaist-up or full lengthMissing body and clothes get filled with plain pieces that match the top.
Seated or lying downStandingEveryone is stood up, but legs and posture have to be guessed.
Tongue out or mouth wide openClosed mouth or a soft smileWide-open mouths tend to look strange; the redraw now requests a closed mouth.
Holding a phone, drink or bouquetEmpty handsHeld objects and some small jewelry get dropped.
Eyes hiddenEyes uncoveredCovered eyes leave nothing to copy.
A tiny or blurry copyThe biggest original you haveUnder 300 pixels per side is refused; blur costs detail.

Inside the studio: what happens to each portrait

  1. Your browser resaves the photo as a JPG with the longest side at 2,048 pixels and removes its location data.
  2. On Wan 3.0 with the Hotel Lobby set, each photo becomes a full-length studio portrait: same face, hair, age, skin or fur and build, standing with relaxed arms and a closed mouth on a plain gray backdrop. Pets are stood on their hind legs.
  3. Anything the photo does not show is filled with simple clothes that match what is visible. Glasses, hats and jewelry are not invented.
  4. The video model then runs the duet routine with the two portraits.
  5. The portraits are never shown to you and are deleted with your photos 7 days after the render. If a portrait fails or takes over 5 minutes, your original photo is used instead.

There is no redraw when you upload Your clip or use a Seedance model. Your photos go in untouched, so full-length shots matter even more there.

One photo mode: one photo of you both, left raps and right hypes. A very wide photo is refused with a message to crop it closer.
One photo mode: one photo of you both, left raps and right hypes. A very wide photo is refused with a message to crop it closer.Actual interface, English · 2026-10-05

One photo of you both, or a pair of photos

  • A pair of photos (Two photos) gives you the most control: one face per slot, the left face leads the verse, the right face is the sidekick. The shots can come from different days and places.
  • A single shot (One photo) works when both of you stand side by side with clear faces; whoever is on the left of the picture leads. On Wan 3.0 it is split into a left and a right portrait. Very wide photos are refused.
  • The single-photo route has not been shown in a published sample yet. If separate shots exist, use them.

Quirks no photo can prevent

  • Likeness is not guaranteed. Faces can drift, most often during quick head turns.
  • Scale is not kept: a small dog can stand as tall as its owner.
  • Paws can look hand-shaped when the routine points or gestures.
  • Babies keep their faces but can look a little older once stood upright.
  • The same two photos give a slightly different take every time.
  • Moves, camera and timing come from the routine. A higher resolution sharpens the picture; nothing shows it makes a face more faithful.
  • Each retry with a better photo costs credits again. Failed renders are refunded automatically.

Why these happen and what the studio does about them is explained in Hotel Lobby AI video problems.

Consent first

Only cast people who said yes. TikTok does not allow AI content of anyone under 18, or of private adults without their permission, and wants realistic AI content labeled (TikTok). The trend has also stirred debate about AI versions of Takeoff, who died in 2022; asked about it, Quavo told the AP on October 2, 2026: “I’m proud of it.” (AP via WSLS).

Create your Hotel Lobby video

FAQ

I only have a selfie from the shoulders up. Will it still work?

Yes, if the face is sharp and looking at the camera. The body and outfit will be invented, so a waist-up or full-length shot keeps more of the real person.

Can I use one photo where we are both in it?

Yes, with One photo, as long as both faces are clear and you stand side by side. On Wan 3.0 it is split into two portraits. Separate shots still give more control.

My dad wears sunglasses in every photo. Is that a problem?

The photo is accepted, but the eyes behind the lenses cannot be recovered. A shot with his eyes showing gives a closer face.

Why is my outfit different in the video?

The redraw keeps what the photo shows and fills the rest with plain matching clothes. Things in your hands and some accessories are left out.

Can I put my toddler in a duet?

The studio accepts the photo, and baby faces are generally kept. Only do it with a parent’s or guardian’s consent, and keep it off TikTok, which bans AI content showing anyone under 18.

Is a 48-megapixel photo better than a normal one?

Not really. Every photo is scaled to 2,048 pixels on the longest side in your browser, so sharpness and light matter more than file size.

My dog is lying down in every photo. Will that work?

On Wan 3.0 with the Hotel Lobby set, the redraw stands pets on their hind legs. A standing or sitting shot with the face clear still leaves less to guess.

References

Up next

Hotel Lobby Video is an independent app with no ties to Quavo, Takeoff, Migos, Quality Control Music, Motown or COLORS.