ASMR and textureFull vertical 9:16 · Reels, Stories, TikTok, ShortsRestaurants and cafés
Before · input
In-house food photo (AI-generated for the industry showcase), cropped to 9:16.AI-generated · made by us
After · resultAI-generated · made by us
Prompt for the model
ASMR food video from this photo of our own dish. Camera: [static macro close-up with a very slight, slow push-in]. Action: [a brass spoon slowly cuts into the tahini semolina halva], [lifts a portion, and a few crumbs and pistachio pieces fall back into the copper bowl]. The dish stays exactly as in the photo: same food, same bowl, same garnish; do not add ingredients. Sound: only close-up sounds - [the soft crumble of the halva and the spoon gently touching the copper bowl]; no music, no voice. Warm candlelight, natural speed.
Replace the gold [bracketed] parts to match your own product, space or brand.
Keep unchanged
The prompt tells the model not to change these. Check the result anyway: these are the parts that break most often.
The food itself
The bowl or plate
Garnish
No added ingredients
When not to use it
Only use it with a photo of a dish you actually sell: the cross-section, consistency and texture in the video are the model's guess; the real dessert cuts differently. Don't present it as "real footage" on your menu or on a food delivery platform.
If you'll use it in ads, disclose that the video was generated with AI. Misleading food visuals are the AI use that draws the most backlash for restaurants.
What it does and how to adapt it
Creates a close-up ASMR video, with sound, of a spoon or knife cutting into your own dessert or dish, from a photo of it. This is the most-watched short video format for patisseries, dessert shops and restaurants.
In the square brackets, change the cutting utensil, the action and the sound. Limit the sound with "only close-up sounds, no music or voice"; otherwise the model may add music. Don't add ingredients: don't write in sauce or cream that isn't in the photo.
The example was generated with Kling 3.0 via Higgsfield (5 sec, with sound).
Generation log
Model
Kling 3.0 (Kuaishou)
Date generated
October 2, 2026
Attempts
Got it on the first try
Audio
on (effects only)
Aspect ratio
9:16
Duration
5 sec
Quality
std
Generated via
Higgsfield API
[Fields] in the example
camera: static macro, slight push-in · action: brass spoon cuts the halva · sound: crumble + spoon on copper
Resolution
720p
Job ID
c503ee12-2c1a-40ed-a3fe-26992915e7ed
What happened across attempts: On the first attempt, the spoon cut into the halva and lifted a portion, crumbs fell, and the sound of the spoon touching the copper came through. The fingertip holding the spoon enters the frame for a moment. The texture of the halva's cross-section is the model's guess.
Even with the same prompt and the same settings, you won't get an identical result: models generate something different every time. The example shows what kind of result this prompt produces; it is not a guarantee.
We use cookies to improve the site and measure advertising performance. Analytics and advertising cookies run only if you accept them. Details are in our Privacy Policy and Data Protection Notice.
