Two-person performance videos are difficult to reproduce from scratch because every element must stay coordinated: positions, gestures, camera angles, timing, props, lighting, and audio. The hotel lobby swap AI video generator on imageat simplifies that production into a reference-led workflow. You upload exactly two character sheets, assign one performer to the left role and one to the right role, then generate a landscape performance that retains the source choreography and synchronized soundtrack.

The concept works for creator collaborations, fictional duos, artist announcements, brand-safe character campaigns, and playful friend edits. The key is not writing a complicated motion prompt. It is preparing two unambiguous character sheets and placing them in the correct order.

Two-person hotel lobby performance video example created with imageat

What the hotel lobby swap effect creates

The effect replaces both people in a pre-structured performance while preserving the elements that make the original take coherent. The source supplies:

Choreography and body movement
Camera position, angles, and movement
Performance timing
Microphone placement and interaction
Lighting
The orange studio environment
The original synchronized soundtrack

Your uploads supply the two characters’ identities, body builds, hair, gender presentation, and outfits. The result is a landscape 16:9 video available in 15-second or 30-second length and 480p or 720p resolution.

This division of responsibilities is important. The workflow is a character replacement, not a tool for redesigning the choreography or relocating the performance. If your goal is to preserve recognizable timing while changing who appears on screen, the fixed source structure is an advantage.

What you need: exactly two character sheets

Prepare exactly two character sheets, one for each role. Do not upload a group image, a third person, a product photo, or several separate files for each angle.

The mapping is fixed:

First character sheet: the left singer
Second character sheet: the right dancer

Each sheet should show one consistent person and include all three of these views:

Front portrait for facial identity and hair detail
Front full-body view for proportions, outfit, and shoes
Back full-body view for rear hair, clothing, and silhouette continuity

Think of each character sheet as one composite reference image with three panels. The workflow expects two sheet files total, not six independent images. Keeping the views together helps the model read each performer as one coherent character.

How to build a strong character sheet

A reliable sheet prioritizes consistency. Use the same person, hairstyle, clothing, and accessories across all three views.

Front portrait

Choose a sharp, unobstructed face looking forward in even light. Avoid sunglasses, masks, heavy filters, and hair covering key features.

Front full-body view

Show the person from head to toe, including shoes. Keep arms slightly away from the torso so the outline remains readable. Use a natural standing posture.

Back full-body view

Show the complete rear silhouette, including shoes. Match the front image’s hair, clothing, and accessories.

Layout and technical checks

Use one view per panel on a plain background. Do not overlap figures, add ornate frames, or place labels across bodies.

Before exporting each sheet, verify:

Only one person appears in all three panels.
The same outfit is visible from front and back.
Face, hair, and skin tone are consistent.
The entire body and footwear remain inside the frame.
No watermark or caption covers the character.
The sheet has enough resolution to inspect facial and clothing detail.

Plan the left and right roles before uploading

Upload order is part of the creative direction. The first character sheet always maps to the left singer, who performs with the microphone. The second always maps to the right dancer. Renaming your files before upload can prevent an avoidable mix-up:

01-left-singer-character-sheet
02-right-dancer-character-sheet

Match each person to the role deliberately. If both wear similar colours, distinguish them through silhouette, texture, footwear, or accessories. Because the environment is orange, check that warm-toned outfits do not disappear into the set.

Step-by-step workflow in imageat

1. Open the dedicated generator

Go to the hotel lobby swap effect. Review the example composition so you understand the left and right positions before assigning your characters.

2. Upload the left singer sheet first

Select the first composite character sheet. Confirm that it includes a front portrait, front full-body view, and back full-body view of the same person.

3. Upload the right dancer sheet second

Add the second composite sheet with the same three-view structure. Double-check that you have not reversed the intended performers.

4. Choose a duration

Select 15 seconds for a compact post or identity test. Choose 30 seconds for more of the synchronized performance. Credit requirements can change, so check the interface.

5. Choose a resolution

Use 480p to test sheet quality and role order. Use 720p for a sharper final output.

6. Generate the performance

Start the job. The workflow replaces both performers while retaining the source choreography, camera, timing, microphone, lighting, orange studio, and original synchronized soundtrack.

7. Review both characters separately

Watch each role separately. Check facial stability, outfit continuity, microphone interaction, foot placement, and whether identities cross between roles.

8. Download the clean landscape master

Save the strongest output before adding text or reframing it. A clean 16:9 master gives you more flexibility for thumbnails, cutdowns, reaction layouts, and platform-specific exports.

Creative ideas for a brand-safe performance

Give the pair a clear concept. Try one of these directions:

Creator collaboration: Put two consenting creators into defined singer and dancer roles.
Original fictional duo: Pair original characters with complementary silhouettes.
Mascot pairing: Turn approved humanized mascots into a performance team while keeping the visual treatment consistent.
Fashion comparison: Use two looks from the same collection as a motion lookbook.
Team announcement: Cast two team members and add a final title card for a launch, event, or collaboration.
Before-and-after pairing: Use two authorized versions of an original character, such as everyday styling and stage styling.
Friend performance: Cast two friends who consent to the edit.

Do not imply endorsement without authorization. Use your own likeness, original characters, licensed assets, or consenting people, and check soundtrack rights and platform rules before commercial posting.

Incomplete prompt starters for adjacent custom workflows

The dedicated effect preserves a fixed performance, so it does not need an elaborate prompt. If you later adapt the concept in a general workflow, these fragments can help you plan. They are intentionally incomplete and need your own authorized details:

“Two original characters in an orange studio, left role wearing …”
“Preserve synchronized partner timing while the right performer …”
“Microphone remains with the left singer; camera follows …”
“Contrasting wardrobe silhouettes, consistent identity through …”
“Landscape performance with negative space for campaign text at …”

Do not use a prompt to contradict the reference take. Requests for different choreography, another set, a new camera path, and exact source timing all at once can work against one another.

How to edit and post the finished video

Preserve synchronization. Avoid speed ramps, frame removal, or audio offsets that make a gesture or microphone beat feel late.

For a polished post:

Keep the original synchronized soundtrack attached to the picture.
Add titles before or after the performance rather than covering faces.
Use subtitles only where they serve your audience and you can verify the wording.
Keep text inside safe margins for platform interfaces.
Export directly from the clean master to limit compression damage.
Select a thumbnail where both faces are visible and bodies are not blurred.
For vertical feeds, place the 16:9 video in a designed 9:16 canvas or use a careful crop that keeps both performers visible.
Add an AI-media disclosure when required by the platform, campaign, or jurisdiction.

If both characters cannot fit in a vertical crop, use the landscape clip with designed space above and below instead of stretching it.

Troubleshooting common issues

The performers are reversed: Re-upload the intended left singer as the first sheet and the intended right dancer as the second.

One identity is unstable: Improve the front portrait and ensure all three panels show the same hair, outfit, and person.

The rear view looks inconsistent: Use a full-body rear view that matches the front outfit.

Shoes or legs change: Include uncropped footwear in both full-body views.

The two characters look blended: Give each person a separate sheet, remove all group imagery, and make their outfit silhouettes more distinct.

Microphone interaction looks weak: Improve the left character’s full-body clarity. The microphone comes from the source, not the sheet.

The vertical edit cuts someone out: Return to the 16:9 master and use a letterboxed vertical layout or a wider crop.

Build a small series from the same approved character references:

Subway freeze AI video generator for a one-person cinematic product story.
Fan blade outfit change AI video for a fashion-focused vertical transformation.
Smash and grab AI video for a fast object-driven action sequence.

These formats use different inputs and outputs, so prepare each upload for its specific workflow rather than reusing a character sheet where a portrait or product image is requested.

FAQ

How many character sheets do I need?

Exactly two. The first maps to the left singer, and the second maps to the right dancer.

What should each character sheet contain?

Each sheet should contain one front portrait, one front full-body view, and one back full-body view of the same person in a consistent outfit.

Can I upload six separate angle photos?

The intended input is two composite character-sheet files, one per performer. Combine the three required views into each person’s sheet before uploading.

What remains unchanged in the result?

The workflow preserves the source choreography, camera, timing, microphone, lighting, orange studio environment, and synchronized soundtrack while replacing the two performers.

Can I change which character sings?

Role mapping depends on upload order. Put the intended singer’s sheet first for the left role and the intended dancer’s sheet second for the right role.

What video lengths and resolutions are available?

Choose a 15-second or 30-second landscape 16:9 video at 480p or 720p.

Does the finished video have sound?

Yes. It includes the original synchronized soundtrack associated with the source performance.

Which settings should I use for a first test?

A 15-second 480p draft is practical for checking sheets and role order. Move to 720p for a sharper final. Pricing may change, so review the current interface.

Can I use celebrity or brand characters?

Use only likenesses, characters, trademarks, and assets you own or have permission to use. Do not suggest endorsement or participation without authorization.

Create your two-person performance video

Build two clean three-view character sheets, put the left singer first and right dancer second, and choose the duration and resolution that suit your edit. Create a two-person hotel lobby performance video with imageat, then review both roles, preserve the synchronized soundtrack, and export a platform-ready version from the landscape master.