How to Make a Hotel Lobby AI Video from Two Photos

What the Hotel Lobby AI template does

Hotel Lobby AI is a two-person reference template. Two adults stand side by side against an orange wall beneath a hanging microphone and perform the same coordinated hand and shoulder gestures. The scene is one continuous landscape 4:3 shot, 15 seconds long and silent, and it is built for the duo format rather than a single performer. ZoopFizz runs the template on Wan 3.0 in reference mode: your two photos supply the identities, and a fixed motion reference supplies the timing and choreography. The appeal comes from deadpan repetition between two people, so the clip reads best when both faces are clear and the two outfits are easy to tell apart.

The displayed sample was generated on ZoopFizz from two authorized adult photos and a user-created motion reference. It was requested at 480p, 15 seconds and 4:3, and delivered at 960×720 pixels. The showcase copy for that sample adds an audio track for display; generated template videos themselves are silent.

What you need before you start

Four things need to be ready before you add photos: two suitable adult photos, permission to use both images, a ZoopFizz account that can run the template, and sound you are licensed to publish. Photos may be PNG, JPEG or WebP up to 10 MB. The generation request is 15 seconds at 4:3 and silent, and the draft uses the SD (480p) or HD (720p) option you choose. A completed run consumes its credits; preparing a draft is free.

• Two photos, one consenting adult in each; no minors in either image.
• One clear face per photo, with the half body that the role needs.
• A ZoopFizz account with enough credits, an eligible plan or a template-video pack.
• A licensed track to add in your publishing app after the silent clip downloads.

Choose the two photos

Each photo replaces one role, so consistency matters more than a dramatic pose. Photo 1 replaces the person on the left and photo 2 replaces the person on the right. The prompt asks the model to preserve each adult's distinct face, hair and clothing without merging or swapping identities, which is easier when the two people look clearly different and wear different outfits. Choose evenly lit photos where the face is toward the camera and the expression is calm rather than theatrical, because the joke depends on the contrast between two serious faces and choreographed gestures.

• One adult per photo; the template does not accept a group shot, a child or an animal.
• Face clear and half body visible; cropping at the eyes or chin weakens identity retention.
• Different hair, clothing or accessories for each role so the model keeps the two people apart.
• Sharp, evenly lit images; a busy background adds detail the model can warp.

How the two roles are assigned

Role assignment is fixed by the template, not by a text box. The left role is the first photo slot and the right role is the second, and the prompt keeps both people visible and their hands coherent through the shared gestures. If you submit two photos that show the same person, the two roles collapse into one repeated identity and the result reads as a duplicate rather than a duo. If you swap the photos, the two identities change sides, so compare your line-up against the role hints before you create.

The fixed reference and what you cannot change

The movement, timing and camera framing come from a fixed reference video selected by ZoopFizz, which is why the template is predictable but not editable in the way a custom prompt is. You do not upload a reference clip and there is no text box to rewrite the choreography. What you choose is the two photos, the SD or HD resolution and, through your plan, the price. The style is called Duo: two friends rap side by side beneath a microphone, with an orange wall and synchronized hand gestures. One continuous 4:3 shot at 15 seconds is the result to expect, not a sequence of cuts.

Step 1 — add both photos and keep the draft local

Open the Hotel Lobby AI page and fill the two labelled slots: Photo 1 (adults only) for the left role and Photo 2 (adults only) for the right role. Selecting a file creates a local preview; nothing is uploaded until you submit. Check each preview against the role it fills, replace the file if it is cropped badly, and leave the page open while you compare versions. If you need to step away, the draft can stay in this browser and you can restore it later without spending credits.

• Photo 1 left, Photo 2 right: the role order is fixed.
• Confirm the permission statement and that no minors appear in either image.
• If you are signed out, the action reads Sign in to generate.

Step 2 — confirm settings, consent and price

Choose SD (480p) or HD (720p) for this run and read the account price before you submit. The cost is 750 or 1,500 credits depending on resolution, with the Starter, Pro and Unlimited rates applied to eligible plans, or one template video credit when your plan includes 480p template videos. Template video packs cover 480p generation: $7.99 for one, $19.99 for three or $29.99 for five. When using credits, template pricing is 750 credits at 480p or 1,500 at 720p HD; Starter and Pro pay 375 or 750, and Unlimited pays 225 or 450. The page confirms your account price before submission. If your balance is short, the payment window opens without discarding the photos.

Step 3 — review the 15-second result honestly

When the task finishes, watch the clip start to finish rather than judging the poster frame. Check the things two-person generative video usually breaks: a swapped or merged identity, hands that lose their shape during gestures, clothing that changes between people, a warped microphone or wall, and one person drifting out of frame. The prompt asks for coherent hands and two distinct identities, but it cannot enforce either. If one detail fails, rerun with clearer photos and stronger contrast between the two outfits instead of assuming the template cannot handle your pair. A confirmed failed or cancelled generation restores unexpired credits once, with the original expiry date; unknown outcomes are checked before any credits are returned.

Step 4 — add sound and publish the 4:3 clip

The template output is silent, so sound is a separate step. The generated clip is silent. Download it from your library, then add music or sound in your editing or publishing app using a track that is licensed for your account and intended use. The frame is landscape 4:3, which suits in-feed landscape slots and embedded players; if you publish to a vertical feed, check the crop so both faces and the microphone stay visible. Because the clip runs 15 seconds, choose a track whose hook lands early, and keep the downloaded file as a backup so you can repost without generating again.

Limits, privacy and common mistakes

The result will not match the reference frame for frame, and no completion time is promised because queue and model availability depend on the supplier. The validated sample used two adult photos and a user-created motion reference; it is not a guarantee that every pair will keep both identities intact. Keep the permission checklist strict: this person template is for consenting adults only. Photos stay local while you build the draft: selecting a file creates a preview in this browser, and continuing saves the draft locally. The image reaches the server and the generation provider only when you choose Create. Generated files are private to your account. Completed videos are stored for 30 days after saving with an email reminder three days before removal; uploads become eligible for deletion seven days after upload when no active video uses them.

• Two photos of the same person produce a repeated identity instead of a duo.
• Group photos, children or animals break the two-role format.
• Matching outfits make the two identities harder to separate.
• Posting a low-shoulder crop hides the gestures that carry the format.