Food video lives on one rule: ultra slow-motion plus a single appetizing moment — a pull, a pour, a burst, a tear. The 10 prompts below spell out one explicit action, 120fps slow motion, a single hard light, and a contrasting backdrop, so they hold up on Veo 3.1, Sora 2, and Kling without morphing. Use them for restaurant TVCs, delivery ads, and paid social creative.
TL;DR
- Write one action per clip (cheese pull or wine pour, never both), 120fps, single hard light, 4–6 seconds. Stacking actions or going past 8 seconds is what makes food “melt” on screen.
- For liquids, splashes, and slow-mo physics, Veo 3.1 (native 4K, 48kHz audio, launched January 2026) is the most reliable. Sora 2 holds static close-ups well. Kling O3 biases toward Chinese and Japanese dishes.
- Generate 3–5 short clips, one moment each, then cut them together with BGM and text — that beats coaxing one long prompt every time.
The six things every food prompt needs
| Element | Why it matters | Example wording |
|---|---|---|
| One explicit action | Multi-action prompts wobble and morph | cheese being pulled apart |
| 120fps slow motion | Triggers the slo-mo look reliably | 120fps food commercial |
| Backdrop contrast | Warm food pops on dark; fresh food on light | dark moody background |
| Single hard light | Hard light catches the gloss ads need | single warm side light |
| Texture words | Sells the appetite appeal | glistening, molten, steaming |
| Duration 4–6s | The stable window before food distorts | 5 seconds |
10 copy-ready prompt templates
1. Pizza cheese-pull
Best for: Fast-food or pizza brand ad
Ultra slow-motion close-up of two halves of a pizza being pulled apart, golden molten cheese stretching between them, slight steam rising, dark moody background, single warm top light, 120fps food commercial, 6 seconds2. Coffee slow-pour hero
Best for: Coffee brand hero video
Slow pour of dark espresso falling into a white ceramic cup, splash detail visible, marble counter, soft window light from camera-left, premium coffee brand commercial, 120fps, 5 seconds3. Soda crack-open mist
Best for: Beverage brand hook segment
Slow-motion close-up of a frosted soda bottle being uncapped, fine mist bursts out, condensation visible on glass, dark background with single hard top rim light, beverage commercial style, 120fps, 4 seconds4. Red-wine pour luxury
Best for: Luxury wine brand campaign
Slow elegant pour of red wine into a crystal glass on a dark wood table, single warm candle-flame side light, slight ripple on surface, luxury wine brand commercial, anamorphic 50mm, 6 seconds5. Strawberry cream-splash
Best for: Dessert or dairy brand ad
Slow-motion macro of fresh strawberries dropping into a pool of cream, splashes flying outward, hard side spotlight, plain white background, 120fps food commercial, 5 seconds6. Steak sizzle tracking
Best for: Fine dining or meat brand
Slow tracking shot along a sizzling steak on a hot pan, slight smoke rising, sea salt being sprinkled from above, single warm side light, dark moody food commercial, 120fps, 6 seconds7. Ramen chopstick-lift
Best for: Ramen or Japanese cuisine ad
A bowl of ramen with steam rising, slow chopsticks lifting noodles from the bowl, golden broth dripping, dark moody background, single warm overhead light, 120fps, 6 seconds8. Honey-drizzle pancakes
Best for: Breakfast or bakery brand
Slow pour of golden honey from a wooden dipper onto a stack of pancakes, slight slow-motion drip, soft warm side light, plain dark wood surface, food commercial, 120fps, 5 seconds9. Burger lid-lift reveal
Best for: Burger or fast-food brand ad
A burger's top bun lifts off slightly revealing melted cheese and toppings, slow motion lift, single dramatic top light, deep contrast against black background, fast-food ad commercial, 120fps, 4 seconds10. Sushi placement macro
Best for: Japanese or fine-dining ad
Slow close-up of sushi being placed on a wooden board by chef chopsticks, single hard side light catching the fish gloss, dark plain background, premium Japanese cuisine commercial, 120fps, 5 secondsWhich model handles food best (June 2026)
| Model | Access | Best for food | Watch out for |
|---|---|---|---|
| Veo 3.1 | Google AI Pro $19.99/mo (1,000 Flow credits) or AI Ultra $249.99/mo | Liquids, splashes, slow-mo pours; native 4K + 48kHz audio | Quality tier burns credits fast — draft on Lite/Fast first |
| Sora 2 | ChatGPT Plus $20/mo (limited) or Pro $200/mo (up to 25s, higher res) | Clean static close-ups, plating, hero stills-in-motion | Consumer Sora app was discontinued April 2026; generate inside ChatGPT |
| Kling O3 | Kuaishou subscription, web/app | Chinese and Japanese dishes, hot-wok and broth shots | Western fast food reads less natural |
A practical loop: draft on a cheap tier (Veo 3.1 Lite or Fast, Kling Turbo), pick the take that holds, then re-render the winner at high quality. For audio-synced sizzle or pour sound, Veo 3.1’s native 48kHz audio saves a sound-design pass.
Common mistakes
- Logos in the prompt — text and brand marks almost always render garbled. Composite real logos in post.
- Two actions at once — “pour and sprinkle” destabilizes the clip. One verb per prompt.
- Empty filler like
colorful delicious food— gives the model nothing to anchor on. - Soft, even lighting — food ads need a single hard light to catch the gloss; soft light reads flat.
- Over 8 seconds — past the 4–6s window food morphs and melts. Keep clips short and cut them together.
How to push results further
120fps slow motionis the keyword that reliably triggers the slo-mo look. Keep it in every prompt.- Splash-into-liquid moment: template 5 (strawberry + cream) — single hard light, white background, fast macro action.
- Out-of-the-pan moment: template 6 (steak) — warm light, rising smoke, dark backdrop.
- Fine Japanese look: single hard side light, wood board, dark plain background (templates 7 and 10).
- One dish per prompt. Cut three or four different 5-second clips together rather than asking for one long shot.
- On Veo 3.1, use Ingredients to Video (upload up to three reference images) to lock a specific product’s look across multiple clips.
FAQ
Q: Is food video harder than a food image?
A: Yes. Slow-mo fluids — splashing liquid, a cheese pull — still wobble in AI video. Short 4–6 second clips plus editing is the reliable path, not one long take.
Q: Why does food morph or melt in AI video?
A: It happens when the duration runs long or the prompt stacks actions. Keep clips at 6 seconds or under, simplify to a single action, and avoid layering multiple movements.
Q: Best video model for food as of June 2026?
A: Veo 3.1 handles liquids and slow-motion physics best (native 4K, 48kHz audio). Sora 2 is strong for clean static close-ups inside ChatGPT. Kling O3 biases toward Chinese and Japanese dishes.
Q: How much does it cost to make these?
A: Veo 3.1 comes with Google AI Pro at $19.99/month (1,000 Flow credits, roughly 10 Quality or 100 Lite clips). Sora 2 is in ChatGPT Plus at $20/month (limited) or Pro at $200/month for full 25-second access. Prices as of June 2026.
Q: How do I make a “homemade” Xiaohongshu-style clip instead of a glossy ad?
A: Swap the hard light for natural window light, use a beige or light backdrop, keep one simple action (a tear or a pour), and hold to 5 seconds or less.
Q: How do I cut a full ad from these?
A: Generate 3–5 clips of about 5 seconds each, one appetizing moment per clip, then composite with BGM and text into a 15–30 second TVC.