Food and Product Ad Video Prompts in AI

A strong AI food ad video prompt names the surface, the light angle and the camera move, then locks the product so it stays rigid and readable — a generic prompt describes the food and leaves everything else to chance. On VdoBloom, the food product video tool and the UGC ad generator supply most of that structure for you, and the six prompts below show how to write the rest yourself.

Updated October 2026.

What separates a food or product ad prompt from a generic one

A generic prompt says “make a video of a burger”. An ad prompt answers the four questions that decide whether the clip is usable: what surface is the product on, where is the light coming from, how does the camera move, and what must stay unchanged. For food, that usually means a dark or matte surface, one soft side light at roughly 90 degrees to pick out texture, and a slow macro push-in. For a packaged product, the priority flips: the pack must stay centred and rigid and the front label, nutrition panel and brand mark must stay sharp for the whole clip, so a locked camera beats a dramatic one.

Length matters as much as wording. Most of the video models on VdoBloom render 4 to 15 seconds, so a single prompt should describe one clean beat rather than a whole commercial. Cramming a hook, a reveal and a call to action into one 8-second clip is why food ad prompts turn vague.

Two VdoBloom tools that carry the workflow

The food product video tool is a one-photo workflow built around its Gourmet Table template. Upload a single photo of a packaged food item and it turns through a smooth, continuous 360 degrees on a dark slate table, lit by a side key at 90 degrees plus a gentle backlight rim that picks out the wrapper grain or the gloss on a jar lid. Props sit far behind and out of focus so nothing competes with the pack, the camera stays locked slightly above the item, the rotation runs at constant speed, and the clip loops. The prompt behind the template keeps the front label, nutrition panel and brand mark sharp and legible through the whole revolution, with the contents unchanged from your photo. Output is an MP4 of roughly 4 to 30 seconds depending on the model you pick, and the underlying Product 360 collection ships 18 templates.

The UGC ad generator is the talking-ad path, and it is deliberately split into three steps rather than one button. Step one builds the creator as a still image with controls for gender, age band, free-text ethnicity, background, camera look, camera placement, framing and aspect ratio; the shipped styling note is “No plastic skin, natural lighting”, aimed at authenticity rather than a glossy stock look. You can also upload a real person’s photo behind a consent checkbox. Step two composites your product photo into that scene using an image-edit model that accepts two or more images. Step three takes your dialogue and renders the talking video.

The costs are real and shown before each run. The UGC render is the dominant charge: Seedance 2 image-to-video is 76 credits for 10 seconds at 480p and 163 at 720p, and Wan 2.7 image-to-video is 64 credits for 10 seconds at 720p. For a shorter talking clip, Seedance 1.5 Pro with audio is 3 credits per second at 720p. The creator image and the product composite are separate, much smaller image-model charges.

Six food and product ad prompts you can paste in

1. Packaged food 360 (food product video tool, Gourmet Table template, 16:9). “Input: one front-on photo of a matte black bag of coffee beans. Dark slate table, warm side key at 90 degrees, soft backlight rim picking out the foil texture, blurred props far behind. The pack turns a smooth, constant-speed 360 degrees in a single unbroken take and loops; camera locked slightly above the pack. Keep the front label, roast notes and brand mark sharp and legible throughout, contents unchanged.”
2. Plated dish ASMR macro (food-dish tab, Seedance 2 or Seedance 2.5, 9:16, 10s, 1080p). “Extreme close-up ASMR video of a ramen bowl: a soft-boiled egg is halved and lowered into the broth, steam curls up, and a ladle of golden stock is poured over so the surface ripples and catches the light. Very slow macro camera, shallow depth of field, dark matte tabletop, one soft side light. No people, no hands, no text — pure food cinematography.”
3. Talking UGC ad (UGC ad generator, Seedance 2, 720p, 9:16, 8s). Step 1: Young Woman, 20s, living room, iPhone-authentic camera, handheld, close-up, 9:16. Step 2: composite one clean photo of a granola bar. Step 3, dialogue: “Okay, I did not expect a protein bar to taste like this. Twelve grams of protein, no weird aftertaste, and I have one in my bag every single day.” Vibe: casual. Camera motion: slight movement. Eye direction: natural.
4. Beverage splash spin (beverage splash video, Kling 3.0, 1:1, 6s). “A chilled glass bottle of sparkling water makes one smooth, continuous 360-degree turn on a dark moody background while slow-motion water arcs and droplets swirl around it and fresh condensation beads cling to the glass. Cool backlight through the splash so the liquid glows, crisp rim light on the bottle edges. The water never covers, distorts or obscures the front label, and the logo stays sharp. Locked camera, constant rotation, one unbroken take, seamless loop.”
5. Product hero orbit (product-demo tab, Seedance 2, 16:9, 8s, 1080p). “A pair of wireless earbuds in a matte charging case, alone at the centre of a seamless neutral studio set. A wide establishing shot pushes in slowly, the camera arcs 45 degrees around the case as a band of light sweeps along the edge and a macro glide reads the hinge detail, then a slow pull-out returns to a clean centred hero frame. Soft key light with a gentle rim light; no people, product only.”
6. Sizzle-and-serve spot (WAN 3.0, 9:16, 12s, native audio). “A 12-second spot for a food-truck taco: a flat-top grill, tortillas pressed flat, flames flaring as the cook flips the filling, chili sauce spooned on fast, the finished taco handed into a paper wrapper. Dynamic handheld energy, bright warm colours, shallow depth of field. Native audio: sizzling, a quick spatula scrape, a cheer from a short queue.”

Pick models to match the beat. Seedance 2 renders at roughly 16.5 credits per second at 720p, so a 5-second clip is about 82 credits; Seedance 2.5 runs higher per second but reaches 30 seconds. Where a clip needs sound baked in, WAN 3.0 starts at 32 credits for 5 seconds at 720p and generates audio natively, while Kling 3.0’s standard tier is about 5.6 credits per second for silent output and Kling 3.0 Omni is 40 credits for 5 seconds at 720p with audio always on. Veo 3.1 Fast bills a flat 40 credits per generation if you want its reference-image mode for a recurring product shot.

Shot length, camera movement and product consistency

Keep each prompt to one camera move. Food reads best with a single slow macro push-in, a rack focus from the garnish to the whole plate, or a slow orbit while the dish stays still. Fast pans smear texture and make a dish look like a stock clip. For a 360 product spin, the opposite rule applies: lock the camera and let the item rotate, because any camera drift pulls the label out of focus and makes the text unreadable.

Consistency across a set of shots comes from reusing one input. Start from the same clean hero photo of the product for every cut, keep the lighting description identical, and do not restyle the product between prompts. The UGC flow does this for you by saving the creator image and letting you re-render with new dialogue on the same face; a food or product set works the same way when you hold the source photo and the surface and light lines fixed.

Budget-wise, draft cheap. Rendering at a lower resolution first, then re-running the winning prompt at your final settings, is the single biggest lever on cost. HappyHorse 1.1 Reference-to-Video, for instance, is about 9 credits per second at 720p, so a 4-second reference clip is 36 credits, and it can hold one product consistent across several generated scenes for a multi-shot ad.

Frequently asked questions

Can I make a food ad video from one photo?

Yes. The food product video tool needs one photo of the pack and supplies the surface, side lighting, backlight rim and out-of-focus props around it, then renders a continuous 360-degree turn. It works best on packaged goods such as cartons, jars, pouches and bottles, where a rigid object holds its shape through a rotation.

What should a food ad video prompt include?

Name the surface, the light angle, the camera move and what must stay unchanged. For packaged food, add that the front label and nutrition panel must stay sharp and legible and the contents unchanged; for a plated dish, add that there are no people, hands or text in frame.

Do I need a real person for a UGC-style ad?

No. The UGC ad generator builds the creator as an image from your description in the first step, composites the product in the second, and renders the talking video in the third. If you prefer a real person — a founder or an ambassador whose photo you have rights to — you can upload it behind a consent checkbox.

How much do these ad videos cost in credits?

It depends on the model and settings, and the price is shown before every generation. Examples from the credit table: Seedance 2 image-to-video is 76 credits for 10 seconds at 480p and 163 at 720p, Wan 2.7 image-to-video is 64 credits for 10 seconds at 720p, and Seedance 1.5 Pro with audio is 3 credits per second at 720p. New accounts start with 10 free credits; plan details are on the pricing page.

What aspect ratio should a food or product ad use?

Use 9:16 for Reels, Shorts and TikTok, 1:1 for feed posts, 16:9 for a website hero or YouTube pre-roll, and 1:1 or 4:3 for a marketplace listing gallery. Available ratios depend on the model you choose.

Start with the food product video tool when you already have a clean packshot, or the UGC ad generator when the ad needs a person talking to camera. For the wider workflow, see the food photography guide, the 360 product video walkthrough, and the ecommerce product ads guide; the UGC ads workflow covers the three-step creator flow in more detail.