How to do the Hotel Lobby AI trend with two photos
Prepare one clear portrait for each consenting adult, select both photos on the homepage, and check availability and credits before generating. This guide takes you from photo selection to reviewing and sharing a duo performance.
1. Prepare one clear portrait for each performer
Choose two separate photos, each showing one adult. JPG, PNG and WebP are supported, with a maximum of 10 MB per image. Ask each participant for permission to use their photo in an AI-generated performance and to share the result. A photo being publicly visible does not make it yours to use.
Look for a face that is sharp when you view the original image at normal size. Even lighting should show both eyes and the shape of the mouth. Keep the head inside the frame and avoid sunglasses, strong beauty filters, deep shadows or a hand covering the face. These are practical input checks, not a guarantee that generation will preserve every detail.
If your only source is a group picture, make a separate crop for each person and inspect the crops before continuing. A distant face can become blurry after cropping. In that case, ask for a better portrait rather than enlarging the low-detail image. For a balanced pair, use references with reasonably similar sharpness and face angles.
Before moving on: you should have two files, one identifiable person in each, and permission from both participants. The filter and photo guide provides a shorter reference checklist.
2. Select the two photos and inspect the slots
Open the Hotel Lobby AI generator. Select the first photo in the first performer slot and the second photo in the other slot. The form displays the selected images so you can check that the right files are paired. Replace the wrong file or swap the photos before submitting.
This stage is local preparation: choosing files does not create a performance video. A preview of the photographs cannot show future motion, face consistency or lip movement. If you are looking for a “Hotel Lobby AI template free,” the preset can guide your preparation, but a completed rendered video is a separate operation.
Our homepage uses a built-in scene instruction. It does not provide a custom prompt box, source-video upload or song-upload control. You do not need to copy a long prompt from another tool into the form. For an explanation of these different workflows, see the Hotel Lobby trend overview.
3. Check availability and credit cost, then submit once
Read the generator's current status before you start. If generation is unavailable, you can prepare your references, but the form cannot produce a clip yet. When available, sign in and check the displayed credit cost and your balance. Account registration alone does not promise a free generated video.
Treat the price shown by the live form as the current price. Before buying credits, confirm that you can use the generation feature and that the offer meets your needs. The free and credit guide explains which preparation steps cost nothing and which operation uses credits.
After submitting, allow the request to move into task tracking. Repeatedly clicking generate or creating a second job is not a way to speed up the first one. Keep the task reference if the form provides it; it helps identify the exact request if you later need support.
4. Follow the task status instead of guessing from the wait
A processing task has not finished yet. If the form offers a status refresh, use it to check the existing task. Generation time can vary, so there is no guaranteed completion time in this tutorial. Plan around a finished result rather than a timer estimate.
A temporary status-check error also does not establish that the provider failed to generate the video. Check the same task again before starting another one. Confirmed provider failures return the generation credits; pending tasks need their status resolved first. See the account's task history or contact page for help with a specific task reference.
Before moving on: the task should show a completed result with a video you can play. If you only see the two original portraits or the illustrative stage image, you are still looking at preparation or sample artwork.
- Review both faces, motion and sound from beginning to end
Play the full clip once at normal speed. Then pause near the beginning, middle and end. Check each face separately, especially when a person turns, gestures across their face or moves behind the microphone. A good first frame does not tell you whether the rest of the performance is usable.
Use this review checklist:
- Appearance: do both people remain recognizable, without faces merging or changing unexpectedly?
- Movement: do hands, arms and microphone interactions look plausible enough for your intended use?
- Framing: are both performers visible, and will a later vertical crop cut off a head or a gesture?
- Sound: is there unexpected speech or audio that could misrepresent either participant?
- Context: is the clip clearly a fictional AI performance rather than evidence of a real event?
The preset guides a new duo scene. It does not promise the original choreography, an exact likeness or word-for-word lip synchronization. If the output is unsuitable, decide whether the issue is a weak input, a cropping problem or a model limitation before spending credits on another generation.
6. Download, choose your soundtrack and share with context
Save the completed clip before editing. Use your own editor for captions, trimming or reframing; those editing controls are not part of the homepage photo form. Review the exported version again, since a crop or text overlay can hide one performer or cover important facial details.
The generated clip may contain original AI audio. It does not include the licensed Hotel Lobby recording or its full lyrics. For a specific song, check the music permissions for your publishing platform and account type. Our audio guide explains the distinction between generated sound and a licensed soundtrack.
Use the destination platform's AI disclosure controls where required, and add context that makes the fictional performance understandable. TikTok explains its expectations in its official AI-generated content guide. Ask both participants to approve the final clip before sharing it with a wider audience.
Troubleshooting before spending credits on a retry
The image will not fit the upload requirements
Check its format and size first. Convert an unsupported file into JPG, PNG or WebP and keep it under 10 MB. Inspect the converted image for blur before selecting it again. A new file extension alone does not convert the actual image format.
One face becomes less recognizable
Compare that person's source with the other portrait. Replace a side profile, dark face or heavily filtered reference with a sharper front-facing photo. This gives the next attempt clearer information; it cannot guarantee a specific result.
The clip is complete, but the soundtrack is wrong
Decide whether replacing the audio in an editor solves the problem before rerendering the visuals. Changing a soundtrack will not correct lip movement already present in the clip. If exact singing is essential, assess whether this preset fits that requirement.
The task appears stuck
Check the same task's status and keep its reference. Do not treat a long wait as proof of failure or assume a new job will replace the old one. Use the contact page if you cannot resolve the task state.
Questions about making the Hotel Lobby trend
Can I use one photo containing both people?
This form asks for two separate files. If you have a clear shared portrait, prepare two individual crops and inspect each crop for enough detail. If either face becomes too small or blurry, use separate portraits instead.
Do I need to subscribe before choosing photos?
You can inspect the form and select photos without spending generation credits. Actual video creation requires the live feature to be available, sign-in and the displayed credit balance. Check the current purchase options rather than assuming that a subscription is required.
Will swapping the photos fix every bad result?
No. Swapping changes the reference order; it does not restore details that are missing from the files. Start by improving the weaker portrait and reviewing what went wrong. Each completed generation can vary even with the same pair.