What the Trending Clip Actually Shows
One static camera, full body, and one hanging microphone in the middle of a warm orange set. Two performers trade verses: the person on the left takes the first line while the other nods on the beat, then they switch. That is the whole format. Every useful prompt instruction exists to keep those four things from breaking.
The appeal comes from the contrast between a very recognizable stage and an unexpected cast. The booth stays simple so viewers can notice the faces, the pause before the handoff, and the reaction from the person who is waiting. If the camera wanders, the microphone disappears, or both performers move at once, the format loses its clean rhythm. A good remake therefore starts with composition and timing before it adds style words.
What You Need Before You Start
One photo or two?
One photo gives you the solo spotlight version. Two photos give you the duo, and the two photos should be uploaded separately, one per subject. A single group photo gives the model less control over who belongs on each side, which is where face blending starts.
Photo rules that decide the result
Use a clear, unobstructed face, even lighting, and shoulders or full body in frame. Comparable lighting between the two photos helps with a duo. Decide the left and right positions before you open the generator — swapping them in the prompt is far easier than fixing a mismatched render.
The Prompt Has Seven Slots
The live generator keeps these instructions on the server, so you only upload two photos. The slots below explain what the fixed prompt protects; they are reference material for understanding the render, not a prompt box you need to fill in.
Hotel Lobby AI Video Prompt Examples
These examples make the hidden instruction slots concrete for advanced readers. The live generator already applies a fixed two-photo prompt, so you do not need to paste or edit any of these examples to create a video.
1. Classic Duo Booth
Two consenting adultsTurn TWO separate photos of two consenting adults into a 16:9 reference-format orange-booth duet. Keep Person A full body on the LEFT and Person B full body on the RIGHT, with a single black microphone hanging at center frame. Person A performs the first line while Person B nods on the beat; then they switch, and A reacts while B performs. Preserve both faces, hairstyles and outfits exactly as they appear in the source photos. Static centered camera, full body with feet visible, soft even studio light, restrained natural gestures only. No face blending, no left-right swap, no cuts, zooms or pans, no extra people, no on-screen text, no logos, no watermarks. Keep the take continuous and let each performer finish one clear beat before the other starts.
If a first render changes a face, edit the Input or Position slot only and keep the exchange instruction unchanged.
2. Pet + Owner
A pet and owner, or two petsUse TWO separate pet or pet-and-owner photos to build a playful 16:9 reference-format hotel-lobby-style performance. Keep Subject A on the LEFT and Subject B on the RIGHT in a matte-orange studio, with one microphone hanging between them. Alternate small head turns, ear flicks, and natural gestures, and keep each subject's movement independent. Keep fur patterns, markings, faces, and body proportions believable. Static full-body framing with paws or feet visible, soft even light. No extra animals, no merged or duplicated limbs, no face swapping, no text, no logos, no watermarks. Let the owner or larger subject start the first beat, then give the pet a small reaction without forcing human-like speech or anatomy. Keep the animal's gaze natural and avoid synchronized mouth movement.
If the pet looks too human, change the Exchange slot to a smaller reaction and leave the Scene and Camera slots alone.
3. Siblings / Family
Consenting adult family membersCreate a 15-second 16:9 performance from TWO separate photos of consenting adult siblings in a clean terracotta studio with one centered hanging microphone. Keep Sibling A on the LEFT and Sibling B on the RIGHT, preserving both identities, outfits, and relative height. Let A take the first beat while B smiles and reacts, then B takes over as A nods. Locked full-body camera, soft studio lighting. No synchronized motion, no overlapping hands, no side swap, no cuts or pans, no captions, no brand marks, no watermarks. Keep the siblings clearly separate in screen space and use only small hands or head movements so the family resemblance does not turn into face blending. Let the background remain empty so both identities stay primary.
If the two subjects drift together, change the Position slot once and rerun before changing the timing or lighting language.
4. Solo Spotlight
One consenting adult photoTurn ONE photo of a consenting adult into a 16:9 reference-format solo performance in the same orange-booth set. Keep the subject centered and full body with feet visible, with one hanging microphone in frame. Give them one continuous take: a head nod on the beat, a single hand gesture on the second line, and no walking. Preserve the face, hair, and outfit from the source photo. Static camera, soft even studio light. No second person, no crowd, no camera movement, no cuts, no text, no logos, no watermarks. Hold the subject in the center third of the frame and keep the gesture below the microphone so the face remains the visual anchor from the opening frame to the last.
If the solo subject walks or leaves frame, change only the Camera slot to reinforce a locked full-body take.
5. Original Character Duo
Illustrated or original charactersTurn TWO illustrated or original characters into a 16:9 reference-format booth duet. Keep Character A on the LEFT and Character B on the RIGHT, with one hanging microphone at center. Match the illustration style of each source image and hold both designs consistent for the whole clip: outline weight, colour palette, and proportions must not drift. Alternate one line each, with the non-performing character reacting on the beat. Static centered camera, flat even lighting. No photoreal faces, no real-person likeness, no copyrighted characters, no extra figures, no text, no logos, no watermarks. Preserve each original design as a distinct character, and use a simple two-beat exchange instead of adding a crowd, branded costume, or recognizable franchise detail. Use neutral staging so the original designs remain easy to recognize.
If the style drifts, change the Scene slot to repeat the source illustration style and keep the Character and Exchange slots unchanged.
The Negative Constraint Block
Keep this block at the end of your prompt when you need to protect the composition:
No face blending, no left/right swap, no cuts, no zooms, no pans, no extra people, no on-screen text, no logos, no watermarks, no morphing hands or duplicate limbs.
These rules are not filler. Each one maps to a failure that can make the result look artificial, and negative constraints are the part new users most often leave out. Treat them as guardrails rather than as a replacement for a clear positive description: the model still needs to know who stands where, how the camera is framed, and when each person moves.
Common Mistakes That Make It Look Fake
Using a group photo instead of separate photos can blend faces. Making both people move at once makes the clip look stiff. Adding camera movement makes continuity harder to hold. Compressing the export in another app can make faces and edges look worse. Using the wrong aspect ratio can crop out feet or the microphone. Finally, changing three instructions at once makes it impossible to know which change fixed or broke the result.
Run a short diagnosis before you render again. If the identity is wrong, replace the source photo or revise the Input slot. If the identity is right but the side is wrong, revise Position. If the scene is stable but the action is chaotic, revise Exchange. One targeted change preserves what already worked and gives you a useful comparison.
How to Make It on This Site
Go back to the hotel lobby AI video generator, upload two separate photos, and start the paid render. The site adds the private reference video, fixed prompt, audio direction, and Seedance settings on the server. The default result is a 15-second 720p landscape preview while the KIE task is processing.
Use the generator for the first render, then return to this guide only when you want to understand why a source photo or a fixed instruction affects the composition. The tool handles uploads, task polling, and the private preview link; you do not need to write or paste a prompt.
FAQ
These answers cover the practical questions that come up after someone sees the trend and wants to recreate the fixed-camera performance. They also clarify the difference between a social video generator and game-building tools that happen to share the words hotel lobby.
What is the hotel lobby AI trend?
It is a fixed-camera orange booth performance format that spread across TikTok, Reels, and Instagram. The duet version keeps two subjects on separate sides and lets them trade the first beat.
How many photos do I need?
The solo version needs one photo. The duet version needs two separate photos, one per subject. Do not use a group photo for the duet.
How do I pay for a Seedance render?
Open the generator, upload the two source photos, review the Seedance render price, and complete payment before the task begins.
Why does my video look fake?
Start by checking the three most common causes: separate portraits were not used, the camera was allowed to move, or both performers were told to move at the same time.
Is there a hotel lobby AI app?
The mobile browser version performs the same workflow, so you do not need to install an app.
Is this the same as a hotel lobby generator for Minecraft or Roblox?
No. This page is for an AI performance video. Game builders generate 3D rooms, maps, or assets and serve a different intent.