PHOTO TO VIDEO · PRACTICAL GUIDE

How to Make a Hotel Lobby AI Video

Hotel Lobby AI recasts the orange-room duet: two performers, one suspended microphone, pointing gestures and a turn before the camera widens to the platform. This generator follows the complete thirty-second cat remix. It accepts two separate photos of adults, cats or dogs, so your cast can be two pets or a person-and-pet duo.

Start with the reference

This thirty-second cat remix is the motion reference. The staging traces to Quavo and Takeoff’s 2022 HOTEL LOBBY performance on A COLORS SHOW; it is not the full original music performance or our generated result.

The cat remix shared by @ogfeitoza on X ↗

What each photo does

Photo 1: the left performer

Upload one clear adult, cat or dog in a separate image. The first photo always takes the left role. Keep the face or muzzle, hair or coat, ears and body outline visible. The photo supplies appearance; the reference supplies the pose and orange-room stage.

Photo 2: the right performer

Use a separate individual photo for the right role, which points toward the lens and turns away and back. Choose a sharp image with visible identifying details. Two separate images work even if the subjects have never been photographed together. A single group photo is not a substitute for the two labeled slots.

Do you need a prompt?

You do not need to write a prompt. The workflow already assigns photo 1 to the left and photo 2 to the right and supplies the full performance. The room, hanging microphone, widescreen frame and thirty-second length are fixed. Writing a different song or backdrop does not customize this template.

Make your video, step by step

  1. Check the reference and actual sample

    Watch the point, right-performer turn and wide platform shots. Use the synchronized comparison on the effect page to compare the original gray-tabby and spaniel inputs with their real generated result.

  2. Assign the two photos

    Upload the left performer first and right performer second. Confirm that each image contains just its intended subject. Use photos of adults who consent, or pets you have permission to depict.

  3. Review the credit purchase

    The full duet uses 40 site credits. One Creator pack contains 40 credits for one complete output. Current prices and any active offer appear on the pricing page before applicable tax. Sign in and buy credits before starting generation.

  4. Generate the full duet

    Submit the task, then find its progress and final file in My videos. The workflow assembles the performance and continuous audio automatically; you do not have to stitch two halves or synchronize the recording yourself.

  5. Inspect the turn and the join

    Watch both identities in profile and in the wide shots. Check the middle of the video at fifteen seconds as well as the opening and ending. Download the MP4 after checking faces, pet paws, outfits and the single microphone.

Three results to watch for

The left and right identities merge or swap

Why it happens: A photo with several subjects or a weak face can make the model’s two role assignments harder to maintain through a cut.

What to do: Use one subject in each image and recheck the upload order. Inspect both subjects after every camera cut. The role directions guide the model; they cannot guarantee that every frame keeps the correct face.

A pet grows human fingers or loses its coat markings

Why it happens: Upright gestures can pull the generated anatomy toward human hands, especially when the original pet photo hides the forelegs.

What to do: Choose an unobstructed pet image with clear muzzle, markings and paws. Check fur continuity from foreleg to paw and during pointing. Treat bare skin, detached hands or a changed species as artifacts, not a promised feature.

The turn or fifteen-second join changes the performer

Why it happens: The right performer’s profile and the transition between generated sections can expose different identity or clothing interpretations.

What to do: Use sharp source photos with consistent visible clothing or fur. Inspect the turn, the midpoint and the full-body platform views. The current template has no manual continuity editor; judge the complete result before sharing.

Before you share

The output has 720 frames at 24 fps. Matching the reference length does not mean an exact gesture or likeness match. Confirmed failed jobs restore the full reserved credits. For sharing, use a recording authorized for your platform; the generator does not grant ownership of the underlying song.

One continuous music-and-vocal recording from the displayed cat template is included. Your photos do not clone or replace the recorded voices.