Fancy Cafe AI logoFancy Cafe AI
Prompt notes · Seedance 2.5

Fancy cafe AI prompt: the tiny-table gag, not an elegant café portrait

A good fancy cafe AI prompt names the joke outright: the subject sits on the tiny round table as if it were a chair, then climbs down when someone sits on the bench. Without that line, models drift to a chic café portrait. Copy the prompts below, or use our generator with no prompt.

Make my cafe clip

Updated

By the Fancy Cafe AI editorial team · How we check these pages

01 · Add your photos

One photo is enough. Add a second for a friend.

0/2

Photo 1 sits on the tiny table. Photo 2 walks in and sits on the bench; leave it empty and a fictional adult plays that part. Use one clear face per photo. Files can be JPG, PNG or WebP under 10 MB and 300 × 300 pixels or larger.

02 · Choose your video
ModelSeedance 2.0 Fast
Quality
Audio
Director’s notes Optional Clothing only. Start with “Outfit:”. 0/500

The template sound comes with your clip unless you switch to Silent, which leaves the audio track out.

This clip costs10 credits

A $6.99 pack has 10 credits, enough for this clip.

Seedance 2.0 Fast · about 15s · vertical MP4 · See pricing

What's new on this page

  • Solo and duo prompt starting points published.

Copy these fancy cafe prompts

Written starting points for image-to-video models. [Image 1] and [Image 2] are your uploaded photos. Results vary by model and run, and no prompt guarantees a likeness.

Solo: one photo

A one-take comedy POV in a bare minimalist concrete cafe, locked-off camera, vertical 9:16, photorealistic, about 15 seconds. Use [Image 1] only as the look and wardrobe reference. 0-5s: [Image 1] perches confidently on a tiny round metal table as if it were a stool, sipping a takeaway coffee. 5-9s: a stylish fictional adult who does not resemble [Image 1] walks in, sits on the low concrete bench and puts a cup on the table. 9-12s: [Image 1] notices, does a double take and freezes. 12-15s: [Image 1] awkwardly climbs down onto the bench, sets the cup on the table and pretends nothing happened. Preserve the exact face, hair, skin, age and outfit from the photo. No on-screen text, no logos, no extra limbs. Soft cafe ambience only, no music.

The climb-down at the end is the punchline. Keep it in even if you shorten the clip.

Duo: two photos

A one-take comedy POV in a bare minimalist concrete cafe, locked-off camera, vertical 9:16, photorealistic, about 15 seconds. Use each image only as the look and wardrobe reference, and keep each person distinct. 0-5s: [Image 1] perches on a tiny round metal table as if it were a stool, sipping coffee. 5-9s: [Image 2] walks in, sits on the low concrete bench and puts a cup on the table. 9-12s: [Image 1] looks at the cup, then at the table beneath them, and freezes. 12-15s: [Image 1] climbs down onto the bench beside [Image 2] and sits properly, trying to look casual. No on-screen text, no logos, no extra limbs. Soft cafe ambience only, no music.

Say which photo is which. Most mix-ups come from leaving this unclear.

Why prompts drift to a luxury café

“Fancy cafe” reads as an aesthetic, so models reach for marble, warm light and latte art. When we asked ChatGPT for a Seedance prompt with a photo on Nov 4, 2026, it offered a “natural, cinematic pose” in a luxury café. That is the opposite of the joke.

Fix it by adding one line anywhere in your prompt: “POV first time in a fancy cafe: the subject sits on the tiny table as a chair, then climbs down when someone sits on the bench. Not a glamorous coffee-shop photoshoot.”

Which settings to use

Use a vertical 9:16 ratio, a locked-off camera and about 15 seconds. Faster models cut the beats short, so if your tool caps at 5 or 8 seconds, drop the double take and keep the perch, the cup and the climb-down. Keep the room plain. Decorated cafes hide the fact that the table looks like a stool, and then the joke disappears.

If you use a reference video, check you have the right to use it. A photo with a clear beat sheet can work without one.

Do you need a prompt at all?

Not here. Our generator runs Seedance with the scene already written, so you add a photo and press one button. These prompts are for people who already pay for a model and want to work by hand.

Questions

Questions about this Fancy Cafe AI page

Describe a bare concrete cafe with a locked-off camera, the subject perching on a tiny round table as if it were a stool, someone walking in and sitting on the bench, a double take, and a climb down. Most creators use a template or generator, so no single official prompt exists.

If ChatGPT or Seedance assumes a chic café portrait, add this: “POV first time in a fancy cafe, the subject sits on the tiny table as a chair, then climbs down when someone sits on the bench, not a glamorous coffee-shop photoshoot.”

The beat sheet should translate to most image-to-video models, but timing and likeness vary. Shorter models need fewer beats.

No. These are public starting points written for this page. Our generator uses its own setup with a reference clip, which can change as we tune it.

Ask for soft cafe ambience and no music. Asking a model to recreate a song can trigger copyright filters, so add the trend sound when you post.

Guides and answers

Pick the question you came with

Each page answers one question about the fancy cafe POV and has the generator on it.

Fancy Cafe AI generator

Turn one photo into the ~15-second tiny-table POV.

How to make the fancy cafe AI video

The step-by-step guide, with CapCut and filming compared.

Fancy cafe AI from one photo

Which selfie works, and what the AI does with it.

Does Fancy Cafe AI keep the song?

The template sound, silent MP4s and what to know about music rights.

Can ChatGPT make this video?

What it did when we asked, and what it is good for.

Put yourself on the tiny table

Upload one photo and make your first time in a fancy cafe POV, with no filming and no editing app.

Make my cafe clipSee pricing