LobbyBooth applies a tested Hotel Lobby prompt to every render. This page explains what that prompt asks for and why the wording matters.
10s
10-second widescreen clip
16:9 / 720P
Ready for short-form posts
1
Technical failures are refunded
This guide is designed to help you choose inputs and avoid common identity problems. When you are ready, open the generator, upload two photos, and let the fixed booth prompt handle the scene direction.
Open the generatorThe prompt starts with role assignment rather than visual adjectives: Image 1 is the left performer and Image 2 is the right performer. It then asks the model to keep the room, the hanging microphone, the camera path, the lighting and the music unchanged, so the two photos stay the only variable. Movement and framing come from the fixed plate.
The prompt keeps a medium-wide camera, one hanging microphone, restrained gestures and warm orange studio light instead of inventing new ones. It avoids rapid cuts, spinning cameras, crowds and physical contact because those increase identity mixing.
A pet prompt should name the visual features that must remain stable: fur markings, breed, face shape, ear position, and body proportions. Ask for small rhythmic head and paw movements instead of human choreography. Add no logos and no text to avoid artifacts on the backdrop.
No prompt to write: these three steps are the whole input.
Start with two portraits that are similar in crop and lighting. Each photo should show one person or pet. A calm front-facing image gives the model a stronger identity reference than a busy group shot or a frame taken during fast motion.
Place the intended left performer first and the right performer second. The generator preserves that assignment in its internal prompt. Files are checked before generation, and technical failures return the consumed video automatically.
Upload two photos and LobbyBooth renders the clip in the background from a fixed orange-booth scene, then shows the live task state with a download action when it finishes.
The prompt states that the two reference photos define the performers, so the model copies faces, hair and outfits instead of inventing new people for the booth.
Backdrop, hanging microphone, lighting and camera movement are described as unchanged. That repetition is what keeps every render recognisable as the same booth rather than a new set each time.
Wording that asks for a face swap or a redesigned scene either gets refused by provider safety checks or quietly changes the set, so that language never enters the prompt.
LobbyBooth applies one tested prompt to every render, so users never need to paste or edit it. The structure below is still useful when comparing outputs or adapting the idea to another video tool: keep the role assignment and the keep-everything-the-same instruction, and change only the scene details.
Longer prompts are not automatically better. Video models prioritize clear subjects, a coherent scene, and compatible movement. A short hierarchy of identity, position, action, camera, and exclusions is easier to follow than a paragraph full of competing styles. The two source images still carry most of the visual information.
Reading and adapting the prompt examples is free. Each render uses one video from your pack, which covers the provider's processing cost.
Ready to make your own? Open the hotel lobby ai prompts and turn two photos into a ten-second clip.
No. The prompt is fixed and tuned for this effect, which is why two people following the same steps get the same kind of clip. You choose the photos, not the wording.
Open prompts are the fastest way to lose the booth: the model starts redesigning the scene instead of swapping the performers. Keeping the prompt fixed is what preserves the microphone, lighting and camera move.
Yes - through the photos. Framing, expression and what each person wears carry into the render. One clear subject per photo, facing the camera, gives the most predictable pairing.
