Ramen: the pillow-pack still; overlay cook minutes, skip the broth VO
Identity of the brick pack. Minutes are burned type. Nutrition stays off a talking head.
Identity of the brick pack. Minutes are burned type. Nutrition stays off a talking head.
A ramen listing that ships is this pillow pack: the brick, the printed face, the flavour name a buyer will search. Cook minutes are a string you typed. Nutrition is not a constructed host walking sodium over a steaming bowl. Photograph the pack. Animate that still. Overlay the minutes. Do not buy a broth voiceover.
The pillow pack is the SKU
The frame that answers the SKU is the actual wrap: pillow geometry, the brick behind the film, the flavour lockup, the net weight a shelf will match. Inventing a prettier bowl does not answer “is this the spicy miso brick.” It answers “does this look like a movie about noodles.”
Photograph the actual pillow, square-on, reading distance, no glare across the face. The still is the contract. Motion is allowed only after the still is true. Do not generate a pack and present it as yours.
Image-to-video keeps the upload as the opening frame. The film, the seal, the printed face survive because they were already in the picture. The prompt is camera and physics: a slow orbit, a hold. Do not re-describe the flavour, the cook time, or the sodium.
Crop to the pack. A wide of a rustic kitchen hides the type you will overlay and invites a second brick. A generated bowl is not the SKU. If you needed a plated shot, photograph your bowl on a different day. Keep the pillow.
Cook minutes are burned type
Three minutes or four minutes is a number on a pack and a number in a claim. Do not ask a video model to draw it in the scene. On-screen glyphs in a generate have to be right every frame; they will not be. A minted “3 min” on a SKU that cooks in four is a claim legal cannot ship.
Add a text overlay burns one authored line — hook, headline, CTA, the cook minutes — for the whole clip. Position is top, center, or bottom. It does not transcribe. It does not listen to broth.
Write the string you will ship: “Cook 4 min.” One line, whole file. If the number has to change at timestamps for two SKUs, that is timed overlays. If you needed speech on a later take, that is captions. The minutes were never speech.
Generate a clean plate. Type the real minutes here. Copy lock can be later than picture lock. Do not prompt the minutes onto the wrap. Do not use auto-captions hoping the transcriber invents a number nobody said. Overlay is typesetting of a claim you already approved.
If the pack still already carries sharp printed minutes, you may not need an overlay. Proof the still. If the print is small or glare-hit, overlay the same number you printed. Do not let the model “enhance” the face.
Skip the broth VO
A talking host over a steaming bowl is a mouth. Nutrition talk — sodium, “a balanced meal,” “what you should eat after training” — is health-adjacent. YouTube’s third inauthentic-content bucket is AI-generated personas that present as human experts on health, legal issues, finances, or politics. A constructed chef prescribing a serving or explaining the broth is that shape. Labelling the face AI does not leave the bucket.
The pack still does not have to talk. Overlay the minutes. If a named founder must speak, that person is real, on a different day. Generation still does the pillow. It cannot be the nutritionist. Do not glue a talking nutrition card onto the pack file. Quoted sodium in the prompt does not become a legal panel. The kill is persona plus prescription.
Nearby kitchens are not this listing
Restaurants already have an audience page: the dish is the hook, a pour, a plate landing. That playbook fills tables. A ramen brand is CPG identity on a locked pillow. A kitchen can show noodles leaving the pass. A brick listing cannot show a model-minted cook time and call it the pack.
Ecommerce brands sit nearby as a storefront: PDP loops, a pack that has to survive a cart. Useful overlap: the real wrap. Do not rewrite that page as a broth film. UGC-style review of a generated bowl is still a generated bowl.
Pause muted on the face of the pack. Read the flavour. Read the minutes. If you cannot read the minutes, the overlay is missing. If you can, and it is not the number on the SKU sheet, you shipped a claim you did not write.
FAQ
Can a video model put “4 min” on the pack if I spell it in the prompt?
It can draw something that looks like minutes. It will not be a string legal can ship. Overlay the cook time after picture lock.
Why image-to-video instead of a text-to-video broth pour?
Text-to-video re-decides the pack every take. The wrap is the expensive failure. I2V starts from the still you already approved. The brick is the contract.
Does the overlay listen to the steam?
No. One authored line, whole clip, top / center / bottom. It does not transcribe. If nobody said the minutes, you type them.
Can we use a talking chef if we add “not nutritional advice”?
The face is still a constructed presenter. Broth talk is still health-adjacent. Hold the pack. If someone must speak, put a real named person on camera, or do not prescribe.