Best Photos for the Hotel Lobby AI Video: What to Upload (2026)

Short answer: for the Hotel Lobby AI video, upload one clear, front-facing photo per person, chest up or waist up, in soft even light, wearing the outfit you want on camera. No sunglasses, no filters, no group shots. On the original clip you tap the rapper each photo replaces, and a 20 second clip comes back in 3 to 8 minutes. Checked on October 31, 2026.

Why the Photo Matters So Much Here

The Hotel Lobby clip is close. Two rappers stand in the orange COLORS booth under a hanging microphone, and the camera stays on their faces and upper bodies for the whole 20 seconds. There is no wide shot to hide in. Whatever the model reads from your photo, the face, the hair, the build and the clothes, is what fills the frame from the first bar to the last.

That is good news: a strong photo gives a result your friends recognise at once. It also means a weak photo shows. Most results that look off on this trend trace back to the picture, not the render.

The Photo That Works Best

• Front facing. Look straight at the camera, or nearly. A three quarter angle is fine; a full profile is not.
• Chest up or waist up. The booth shows shoulders, arms and hands, so a photo that shows them gives the model your build and your clothes. A tight face crop works, but the outfit is then a guess.
• Soft, even light. Daylight from a window or outdoor shade is ideal. Hard flash, a strong backlight or one side of the face in shadow make the face harder to read.
• Sharp and recent. A clear phone photo from this year beats an old, compressed picture saved from a group chat.
• Face fully visible. No sunglasses, no cap pulled low, no hand, drink or phone in front of the face, no hair across the eyes.
• No filters. Beauty filters and face smoothing change the face the model sees, and the video ends up looking like the filter.

A plain background helps a little, but it matters far less than light and angle. The booth replaces it anyway.

Which Photo Becomes Which Rapper

On the original clip, the generator shows a still of the booth with a circle on each rapper: the man in the striped shirt on the left and the man in the orange shirt on the right. Tap the circle of the one you want to replace and add the photo for him; do the same for your friend. If you only add one photo, the other rapper stays as in the clip. On the other examples, the slots labelled 1 · You and 2 · Your duo (optional) fill in the order you add the photos.

The two photos do not need to be taken together, on the same day or in the same place. Separate photos are actually better than one picture of both of you: one face per photo, so the model never mixes the two people up. If you want to say exactly who stands where, switch on Prompt and write it, for example "the woman in the red jacket on the left".

Outfits: What You Wear Is What You Get

The clothes in the photo are the clothes in the video. That makes the outfit a choice worth a minute:

• Match the vibe. Puffer jackets, chains, caps and bright colours read as the original. A suit and tie or a cardigan makes it a joke, which also works.
• Coordinate the duo. Matching colours, matching jerseys or deliberately opposite looks (one in a blazer, one in a hoodie) make the clip read as a pair.
• Avoid busy logos and printed text. Large slogans on a shirt can come out warped in motion.

If you like the photo but not the outfit, keep the photo and describe the clothes with Prompt on, such as "both in matching orange tracksuits". The face still comes from the photo.

Five Photo Mistakes That Make It Look Off

1. Sunglasses or a hat brim over the eyes. The model has to invent the eyes, and they rarely look like yours.
2. A group photo in one slot. With two or more faces in one picture, the wrong person can end up in the booth. Crop to one face, or use another photo.
3. A far away full-body shot. The face is a few pixels wide, so the close-ups in the booth have nothing sharp to work from.
4. A heavy filter or an old screenshot. Smoothing, blur and compression all wash out the features that make someone recognisable.
5. Someone who did not agree. Use your own photo, or an adult's who said yes. No children, and no celebrity photos.

Quick Checklist Before You Tap Generate

• One face per photo, looking at the camera.
• Chest up or waist up, in the outfit you want on camera.
• Even light, no filter, nothing covering the face.
• Each photo sits on the rapper you tapped for it.
• Everyone in the video agreed to be in it.

How to Make It, in Short

Open the Hotel Lobby AI page. The original clip is loaded in the generator; to start from another take, such as the Leather Couple or the Blazer and Flannel example, tap Recreate this on it and its clip loads instead. Add your photo, and your friend's if you want both spots, then tap Generate. Checkout shows the price before anything is charged. In 3 to 8 minutes the MP4 is ready to download, vertical, with sound and no watermark. The full walkthrough with screenshots is in How to Make the Hotel Lobby AI Video, and ideas for who takes which spot are in Hotel Lobby AI duo ideas.

What It Costs

One Hotel Lobby AI video costs $9.99, whether you replace one rapper or both. The 3-video pack is $24.99 ($8.33 a video) and the 7-video pack $49.99 ($7.14 a video). Packs are one-time purchases with no subscription, the videos in them work on every trend, and a failed render gives the video back automatically.

FAQ

Can I use a selfie? Yes. A clear, front-facing selfie in even light works well, because the clip stays close on the faces. Hold the phone a little further away so your shoulders and outfit are in the frame too.

Do both photos need the same background or lighting? No. Each photo is read on its own, and both people are placed in the orange booth. Similar lighting helps the two faces match in tone, but it is not required.

Can I use a photo of a pet? The page has a Dapper Animal Duet example that puts two animals in the booth. Tap Recreate this on it and add a clear photo of each pet, face visible.

My result does not look like me. What should I change first? The photo. Swap in a sharper, front-facing picture with no sunglasses and soft light, and try again. If the wrong person changed, check that each slot holds one face. See also why an AI video face looks off.

How long is the video? The original clip runs 20 seconds and comes back as a vertical 720p MP4 with sound, ready for TikTok, Reels and Shorts.

See every trend on the trends page, or start in the generator.