← Back to blog

Cafe Promo Video: Opening, New Menu and Seasonal Drink Clips with AI

How to make a cafe promo video for an opening, a new menu or a seasonal drink: photograph the real cup and plate, let AI build the light, steam and motion around them. The 8frame Food preset (Dish, Table, Style, optional Person; 3 scenes, 10 s, 9:16), prompts, fixed model prices in credits, and the rule that the drink must look like what you serve.

A good cafe promo video is short, warm and specific: one drink or one plate, made to look like the best version of what a customer will actually be handed, with your name and an address or a date at the end. You can make it without a shoot day. Photograph the real cup or plate in decent light, open the 8frame clip maker on the Food preset, add a photo of your counter or a table, and let the model add what a phone struggles with: steam, a slow pour, morning light moving across the cup. Keep the drink itself honest, since the person who orders it after watching will compare. This guide covers the three promos a cafe needs most (an opening, a new menu, a seasonal drink), the preset, prompts, prices in credits and where to post.

TL;DR

The three promos a cafe actually needs

Most cafes do not need a brand film. They need a reason for someone nearby to walk in this week. Three moments give that reason, and each one wants a slightly different clip.

The opening (or reopening, or new location)

The goal is to make a place exist in people's heads before they have been inside. The hero is the space as much as the coffee: the counter, the window seat, the first pour of the day. End on the date and the address. If you have not opened yet and the interior is unfinished, do not generate a finished interior that does not match the one people will walk into. Use the cup and the counter you have, and keep the space shots for after the fit-out.

The new menu

A menu change is several items, and a 10-second clip cannot show eight of them. Pick the two or three items you most want to sell, give each one a scene, and put the rest on a caption card. The Food preset's 3-scene default is built for exactly this.

The seasonal or limited drink

This is the cafe promo with the best return on effort, because it has urgency built in. One drink, three angles: the finished cup, the pour or the topping going on, the cup in a hand by the window. End on "until" and a date. Re-use the same structure each season with a new drink, and your feed starts to look planned.

Step by step in the Food preset

1. Photograph the real drink or plate

The Dish slot recommends three photos. For a drink: side on at cup height, from above to show the foam or latte art, and a three-quarter angle. For a pastry or plate: from above, at 45 degrees, and a close-up of the texture (the crumb, the glaze, the layers). Use window light rather than the overhead lights, turn off the flash, and wipe the rim of the cup. Make the drink exactly as the barista will make it for a customer, in the cup it is served in.

2. Add the Table and Style references

The Table slot takes one photo of where the drink sits: your counter, a wooden table by the window, a marble bar. This is what keeps the clip looking like your cafe rather than a stock one. The Style slot takes one frame whose colour and light you like, such as your best-performing post. The Person slot is optional: a barista's hands or a customer by the window adds life, but only use photos of someone who has agreed to appear.

3. Describe the clip and edit the script

Open the Food preset in the clip maker. Its question is "Which dish are we filming?". Describe the clip in one or two sentences, for example "our pumpkin spice latte on the window counter at 8am, steam rising, a slow pour of foam, ending on the cup in a hand". The clip maker drafts a script by scene. Edit any line that adds something you do not serve: a cinnamon stick you do not use, whipped cream on a drink that comes without it, a bigger cup. Words in the script shape the picture.

4. Render, check, re-shoot

The render button shows its credit price before you press it, and the clip maker renders all scenes as one video. Check each scene against the real drink: foam height, colour, garnish, cup. If one scene drifted, re-shoot it for a new take; a scene that fails can be retried on its own, and failed generations are refunded automatically.

5. Finish in the Video Editor

Add the text people need: the drink name, the price if you want it on screen, the date and the address. Put text in the middle of the frame if you will run it as a Reels ad (see the safe zones below), and type it in the editor rather than asking a model to draw it, since models misspell lettering. Captions come from a script you paste; there is no transcription. Keep the Warm and Kodak filter presets away from drinks whose colour matters, such as a matcha or a berry drink, or check the result against the cup. For a voiceover, ElevenLabs text to speech is 13 credits per started 1,000 characters; for background music, MiniMax Music 3 is 17 credits for 60 seconds.

Prompts for cafe clips

Use these with your own photos as references, inside the preset or on the board.

Prompt:

The latte from the reference photos sits on the wooden window counter at 8 in the morning, low sun from the side, a thin line of steam rising, slow push in from a medium shot to a close shot of the foam, the cup and the latte art exactly as in the photos

Prompt:

A barista's hands pour steamed milk into the iced drink from the reference photos in slow motion, the colours swirl and settle, camera at cup height, warm cafe light in the background, the glass and the drink's final colour unchanged from the reference

Prompt:

The almond croissant from the reference photos on a small plate on the marble counter, a hand breaks it in half, flakes fall, slow close shot of the layers, soft morning light, nothing added to the plate

Prompt:

Early morning, an empty cafe with chairs still on the tables, the lights switch on above the counter, the espresso machine lets out a burst of steam, slow wide shot toward the window, warm documentary grade, no visible text or logos

The last one is a mood shot for an opening or a reopening. Use it only if it looks like your cafe, or generate it from a photo of your real counter.

Which model for which shot

The clip maker gives you a finished multi-scene vertical clip. When you need one extra shot, or a still for the menu board, use a single node on the board. Prices are fixed per clip, in credits:

Model on the board Credits Good for The catch
Kling v3 Standard 86 for 5 s with sound, 57 silent Animating your photo of the cup: steam, a slow push in Needs a start image; 720p
Veo 3.1 Lite 54 for 8 s with sound A mood shot of the room with cafe ambience 720p; 16:9 or 9:16 only; from an image it is locked to 8 s
Veo 3.1 Fast 168 for 8 s with sound A richer shot with a person and room sound Same ratio and 8 s limits
Seedance 2.5 157 for 5 s at 720p A pour or a hand in motion from your start image Tops out at 720p
Nano Banana 2 (image) 11 per image at 1K A hero still of the drink on your counter for the menu board or a Story A still, so check it against the real cup just as strictly

No video model on 8frame outputs 4K and there is no video upscaler, so take sharp photos to begin with. If you join clips that have sound, do it on the Video Editor timeline: the Montage node (10 credits) joins the picture only, and its file has no audio track.

Keeping it honest

Food is the category where a promo and reality meet within minutes, at the counter. The UK Advertising Standards Authority's guidance on imagery is a useful rule even outside the UK: an ad is "likely to mislead" if it shows more than the customer gets, and you "should usually either show the item as customers will receive it" or make clear what the price covers. Delivery platforms apply the same idea to menu photos; DoorDash's merchant help says images "must reflect the actual item served, not an idealized or generic version". This is general information, not legal advice.

In practice that means three checks before you post: the cup and its size are the ones you use, the toppings and garnish match the recipe, and the colour of the drink is close to the real one. Generated steam, light and motion are fine; a generated extra layer of cream is not. If you run the clip as a TikTok ad, add the AIGC label or your own clear disclaimer when AI has changed it significantly, which TikTok's ads policy asks for. More on that in do you have to disclose AI-generated ads.

Where to post

If you also run a restaurant menu, see how to make a restaurant video with AI and the comparison of AI video generators for restaurants.

Rather have the clips made for you?

If you run a cafe and would rather spend the morning behind the counter than in an editor, the 8frame team can make the clips for you: 10 clips in 5 business days for $490, with one round of changes, made in the same Food preset this guide uses. Send photos of your drinks and a short brief on the done-for-you food clips page. To make them yourself instead, open the Food preset in the clip maker and follow the steps above.

FAQ

What should a cafe promo video show?

One clear thing: a drink, a dish or the opening of the space, then the reason to come (a date, a limited run, a new menu) and where you are. Ten to fifteen seconds is enough for three scenes and an end card.

Can AI make a coffee shop video from a photo of my drink?

Yes. Upload three photos of the real drink to the Food preset's Dish slot and one of your counter to the Table slot, and the clip maker builds the motion and light around them. Check every scene against the real cup before posting, since models can change foam, garnish or colour.

How long should a cafe video be for Google Business Profile?

Google Business Profile accepts videos up to 30 seconds, up to 75 MB, at 720p or higher. The Food preset's 10-second default fits, and you can raise it to 15 or 20 seconds.

What does a cafe promo cost on 8frame?

The clip maker shows its price on the render button before you render, and it depends on the scenes and length you set. Single shots on the board have fixed prices, such as Veo 3.1 Lite at 54 credits for 8 seconds with sound and Kling v3 Standard at 86 for 5 seconds with sound. Plans start at $19 a month for 1,000 credits; Creator is $49 for 3,000.

Do I need to say the video was made with AI?

On TikTok ads, significantly AI-edited content needs the AIGC label or your own clear disclaimer. Rules differ by platform and country; our guide on disclosing AI-generated ads covers the main ones.

Sources


Try it without signing up: the 8frame canvas opens on a real board. The canvas is free and unlimited; generation is paid from $19/month.

Written with AI assistance. Corrections are noted on the page. Editorial policy

Related articles

use caseFood Truck Promo Video: Where You Are Today, What You Serve, Why Come Nowuse caseHair Salon Promo Video: What to Film, What to Generate, How to Book Chairsuse caseCar Detailing Promo Video: Real Results, AI for Everything Around Them

Make it
move.

Stay in the loop

Be the first to hear about our launch and get product updates