Hotel Lobby AI

Prompt guide

Hotel Lobby AI Prompt Guide: Copy, Adapt, and Fix Common Problems

A useful Hotel Lobby AI prompt does less than most people expect. It assigns two references clearly, preserves the left-right staging, and tells the model which visual details must stay stable. This guide gives you an original starting prompt and shows how to simplify it when a result goes wrong.

Published 7 min read

Quick answer

  • • Use one separate reference photo for each subject.
  • • Name the left and right performer explicitly.
  • • Keep the camera and scene instructions short on the first attempt.
  • • Hotel Lobby AI itself uses a prepared workflow and does not require a prompt.

On this page

A copy-paste Hotel Lobby AI prompt

Use the prompt below as a starting point in a third-party video tool that accepts two image references and a text instruction. Replace the bracketed labels with the actual reference names shown by that tool.

Replace the performer on the left with [LEFT SUBJECT] and the performer on the right with [RIGHT SUBJECT]. Keep both identities separate and consistent from the first frame to the last. Preserve the two-person staging, alternating performance, centered hanging microphone, warm orange studio, camera position, and timing. Match each subject's face, hair, clothing, and visible features to the corresponding reference. Do not swap sides, blend faces, add extra people, crop either performer, or add text and logos.

The prompt is intentionally specific about casting and restrained about style. If the source video or template already contains the movement, camera, and set, asking the model to redesign all of them creates more opportunities for drift.

How to keep two subjects from swapping sides

Upload order alone is not always a reliable casting instruction. Check the label attached to each reference and repeat the assignment in plain language: first reference on the left, second reference on the right. If a tool offers a reference picker, insert those references rather than typing filenames the model may not understand.

Choose source photos with one obvious subject each. A group photo, a tiny face, or two similar-looking people in one image makes the assignment harder. Comparable camera angles help, but the faces do not need identical backgrounds.

  • • Write left and right instead of first person and second person.
  • • Describe one or two stable traits for each subject when identities look similar.
  • • Ask for exactly two performers and no extra characters.
  • • Review the middle and final frames, not only the opening image.

Prompt variations for people, pets, and vertical video

For a person-and-pet version, state the species as part of the casting instruction and ask the tool to preserve markings, fur, ears, and body shape. For two pets, identify a stable visual trait for each one and request natural movement without humanizing the faces.

For a 9:16 result, add a framing instruction: keep both full subjects visible inside a vertical frame with space around the microphone. For 16:9, ask for balanced space on both sides. Do not add camera movement until the identities and side assignments are stable.

Fix the prompt one problem at a time

A longer prompt is not automatically a better prompt. When a result fails, remove unrelated creative directions and change one instruction at a time. That makes it possible to tell whether the reference, casting, framing, or motion caused the problem.

  • • Faces blend: repeat that the identities must remain distinct and use clearer portraits.
  • • Sides swap: use the tool's exact reference labels and restate left versus right.
  • • The background changes: remove extra location and lighting requests.
  • • A third person appears: specify exactly two performers and no background characters.
  • • One face drifts: request identity consistency through the final frame and reduce fast motion.

When you do not need a prompt

Prompt-based tools are useful when you want to control a reference clip, rebuild the scene, or experiment with unusual casting. They also require you to select the right model, references, settings, and generation cost.

Hotel Lobby AI is the shorter route when you only want the recognizable two-subject format. Upload two authorized JPG, PNG, or WebP photos, choose 9:16 or 16:9, and generate with the prepared scene. A submitted video costs 10 Credits and the completed result appears in your private History area.

Frequently asked questions

What is the best Hotel Lobby AI prompt?

The best starting prompt assigns one reference to the left performer and one to the right, keeps both identities separate, and preserves the source motion, camera, orange studio, and hanging microphone. Add detail only to correct a specific problem.

Why does the AI merge the two faces?

The references may be ambiguous or the prompt may not separate the cast clearly. Use one subject per photo, name the left and right references, and ask for two distinct identities from the first frame through the last.

Can I use the prompt for a person and a pet?

Yes, if the selected tool accepts those references. State which side contains the person and which contains the pet, then ask it to preserve the animal's species, markings, face, and body shape.

Does Hotel Lobby AI require this prompt?

No. Hotel Lobby AI uses a prepared two-photo workflow, so you can upload the left and right subjects without writing a text prompt.

Prepared two-photo workflow

Create without building the setup yourself

Upload one authorized photo for each subject, choose 9:16 or 16:9, and generate the prepared Hotel Lobby AI scene. One submitted video costs 10 Credits.

Open the studioBrowse more guides