Hotel Lobby AIPrepare Two Photos
One clear, front-facing subject per image. Use good lighting and similar framing.
Avoid group photos, blur, extreme side profiles, and faces covered by glasses or hands.
How to Make a Hotel Lobby AI Video
Prepare two reference photos, assign left and right, generate a stable orange-booth performance, and add licensed audio.
You need two clear references, a simple performance brief, and a video tool or editor. Start with a short test clip before spending time or credits on a longer version.
Build your Hotel Lobby AI prompt ↗How to Make a Hotel Lobby AI Video
Assign LEFT and RIGHT
LEFT
RIGHT
RIGHTKeep each reference on one side to help reduce swapped identities, merged faces and drift.
Choose a Workflow
Start with a guided tool, or take control with a custom prompt.
Generate and Add Audio
Review the visual first. Then add licensed or platform-provided sound in TikTok, Instagram Reels or CapCut.
Keep only clips with stable faces and clean movement.
What you need
- Two photos you have permission to use, with a clearly visible face in each.
- A reference-capable AI video tool, or an editor for a manual version.
- A distinct description for each subject and an orange-studio composition.
- A licensed or platform-provided sound for the final edit.
Choose your method
Method 1: the easiest workflow
Look for a guided workflow that explicitly accepts the number of references you need. Confirm that its preview places two subjects in the scene. A one-click label is not a guarantee of identity accuracy; check what references and motion it actually supports in our generator comparison.
Method 2: a free workflow
Check whether your existing tool offers usable free credits. If not, arrange your own two-person footage against an orange backdrop and edit it manually. This can create the format without paying for AI, though a photo montage will not look like a generated performance. Follow the free workflow guide.
Method 3: an advanced AI prompt workflow
Prepare an orange-booth start image containing both subjects. Use it in a supported image-to-video mode, or supply separate references if the selected model allows them. Describe a single continuous shot, low-amplitude movement and stable identities. Use our prompt generator to structure the brief.
Prepare the two photos
Choose front-facing portraits with even lighting, similar crop and no heavy filters. Avoid sunglasses, hands over faces, group photos and strong side profiles. Give the subjects distinct clothing and leave room between them in a combined composition.
If a model accepts only one image, combine both subjects into the intended scene first using an editor. A collage is not automatically a coherent start frame: align the scale, perspective and lighting before animation.
Set LEFT and RIGHT before generating
Assign one reference to the left and one to the right. Repeat that assignment in the prompt. Keep the microphone centered, avoid crossing paths, and name distinguishing details such as a cream jacket versus a navy sweater.
Prompt recommendations
Describe the scene, each subject, motion, camera, lighting and constraints in that order. Ask both subjects to make small but different movements. Start with a static camera. Negative instructions can help clarify intent, but they cannot repair a model that does not support your reference setup.
Generate a short test and review every frame
Select a supported aspect ratio and short duration in your chosen tool. Inspect the first, middle and final frames for face changes, outfit transfer, extra limbs, microphone movement and missing subjects. Change one variable at a time: references first, then movement, then camera.
Trim a weak ending or regenerate a short section where your editor supports it. Do not increase duration until both subjects remain consistent in the test.
Add audio and export
Bring the accepted clip into your editor. Add sound you are licensed to use, align a few visible movements to the beat, and export a vertical version for your intended platform. Check the platform’s sound permissions for your account and intended use; access to a sound is not a universal license.
Fix Common Hotel Lobby AI Problems
Why do the two faces swap?+
Two-person generation can lose track of identity. Use separate clear reference images when supported, explicitly assign LEFT and RIGHT, and avoid ambiguous group photos. If the model accepts one image, prepare a two-person composition with distinct outfits first.
Why do both people become the same person?+
Subject merging happens when the model treats two descriptions as one identity. Describe distinctive hair, clothing and facial details for each subject. Simplify the action and keep space between them.
Why does only one person move?+
Some models handle multiple moving subjects less reliably. Ask for a small head bob from the left subject and a gentle shoulder movement from the right. Start with a short clip and one simple action per subject.
Why does the face change halfway through?+
This is temporal identity drift. Shorten the shot, reduce camera motion, keep faces visible and avoid turns or occlusions. Inspect the entire clip; a good opening frame does not guarantee a consistent ending.
Why are the hands distorted?+
Hands and overlapping limbs remain difficult for generative video. Use small gestures, keep hands apart and avoid crossing the microphone. Regenerate or trim the affected portion instead of relying on stronger wording alone.
Why is there no music?+
Many workflows generate visuals separately. Even tools with generated audio do not automatically include rights to a commercial song. Add licensed or platform-provided sound in your editor or publishing app.
Why doesn’t it look like the viral version?+
Focus on the visual structure: minimalist orange studio, exactly two subjects, centered microphone, simple camera, matching framing and stable left/right positions. You do not need to copy an existing video to establish the concept.
Tutorial FAQ
Do I need two separate uploads?+
Only if your chosen model supports them. A single, prepared two-person start image is an alternative for image-to-video workflows.
Can I guarantee matching motion?+
No. A text prompt guides movement but cannot guarantee choreography or identity. Use a supported motion reference you have rights to, or edit your own footage for precise timing.