Defib-pad pouches overlay lot; AED OEM owns the unit
GPT Image 1 Edit is 2cr / 4K for a pouch-still cleanup, then overlay lot and expiry as type you typed; AED OEM already owns the defibrillator chassis, do not generate a rescue or a photoreal clinician.
GPT Image 1 Edit is 2cr / 4K for a pouch-still cleanup, then overlay lot and expiry as type you typed; AED OEM already owns the defibrillator chassis, do not generate a rescue or a photoreal clinician.
A defibrillation-pad listing a buyer can match is this pouch: foil, connector, the window that shows lot and expiry, the electrode SKU a biomedical buyer will match to a bid. Those pixels live in a photograph. Cleaning glare is an edit. Inventing a rescue is a patient you did not consent. Photograph the pouch. Point at the mess. Overlay the dates. Leave the chassis to the OEM post. Leave the body off the generate.
The pouch still is the leftover, not the chassis
AED OEM is the unit still; cabinets already own the wall already shipped the device: handle, status LED, serial, no rescue. AED cabinets: wall still plus overlay on this wall, no fake emergency already shipped the facility plate. This leftover is a different object: the foil you ship, the connector you stock, the date window a tech will read. Mixing those jobs is how a pouch still becomes a lobby rescue you did not consent.
Do not restage the unit still as a pad listing. Do not prompt “AED with pads ready on a hospital wall” and call it the pouch. That paragraph decides a chassis and a wall. Neither is this SKU.
The honest plate is the pouch on a bench. Square-on of the face. Empty of people. If the only photo includes a responder, crop them or shoot again. A generated patient is a hard stop. Do not ask a model to “open the pouch so buyers see the electrodes.” An opened pack is a second still you photographed, or it is a drawing of internals you never unsealed.
GPT Image 1 Edit is a 2-credit 4K cleanup
GPT Image 1 Edit is OpenAI's instruction-based image row: edit-image and image-to-image, image required, 4K ceiling, 2 credits. Describe the change in plain English; it should repaint only what you asked. That is not a second generate. If you re-prompt the pouch from scratch, you have a new still that happens to be on the same brief.
Use it when you already have the pouch photograph and one object has to move: glare on the window, a crop that keeps the connector, a bench reflection. Rolling a fresh text-to-image “defib pads in a modern clinic” throws away the lot window and invents a room.
Write the change, not a scene. “Kill the glare on the date window. Keep the foil, connector, and printed artwork identical. Do not add a person.” A prompt that re-describes the whole tray is how the model “helps” by inventing a different pouch.
The AI photo editor is the surface. Edit a photo with AI is the named agent job: attach the pouch still, describe the cleanup, pick an edit model. Cost is priced per edit model, shown before you confirm. There is no “edit this” without a this.
The record accepts up to six references of the same pouch. Two credits is the catalog price. Max output is 4K. Change the glare. Look. Then crop. Stacking five instructions in one generate is how the model redraws the artwork. Lot and expiry still do not belong in the edit prompt.
Overlay lot and expiry as type you typed
Lot and expiry are claims. A date that drifts for six frames is a mis-ship. Video and image models redraw letterforms every pass. Text you cannot afford to get wrong does not live inside the generate.
Do not ask GPT Image 1 Edit to typeset a new expiry onto the foil. Dates change. The pouch artwork should not. Regenerating the foil to roll a new lot is how the connector cousins.
Add a text overlay burns one authored line for the whole clip — top, center, or bottom — if the pouch later moves. It does not transcribe. Write the lot and expiry as they sit on the batch record. If the window is unreadable, overlay the string you typed. Do not “enhance” the digits. Prettier digits are a different lot.
A single persistent badge — lot, expiry, “for demonstration only” — is this overlay. Neither job belongs in the GPT Image prompt.
Do not generate a rescue or a photoreal clinician
A rescue in frame is a patient and a protocol. A photoreal clinician next to a pouch is a health persona. You did not cast either. Synthetic presenters on health and finance are a YPP kill: AI personas as human experts on health. A constructed responder telling a stranger how to pad or shock is that shape. Labelling the face AI does not leave the bucket.
Do not prompt hands opening the pouch onto a torso. Do not generate a collapsed extra or a dummy in scrubs. If someone must speak a protocol, that someone is a real named credentialed person on a different take. Generation can still do the foil. It cannot be the rescue.
The cabinet page owns the wall. The OEM page owns the unit. This page owns the pouch. None of them owns a patient. Photograph the pouch. Spend 2 credits at 4K on the glare. Overlay lot and expiry as type you typed.
FAQ
Can we generate a pad-placement demo so buyers see how the pouch opens?
No. That demo is anatomy and a protocol. Photograph the closed pouch. Overlay lot, expiry, and the IFU URL you already host. A real named applications person, on a different shoot, if you need a voice. Do not invent a body so the pouch has a scene.
Is this the same job as the AED-OEM unit post?
No. That post owns the defibrillator chassis: unit still, overlay SKU, no rescue. Cabinets already own the wall. This leftover is the pad pouch — 2-credit 4K cleanup, overlay lot and expiry. Do not restage the unit still as the consumable row.
Why GPT Image 1 Edit instead of a talking cinematic row?
Because the listing is a pouch still, not a mouth. GPT Image 1 Edit is 4K, 2 credits, edit-image / image-to-image, image required. A presenter row will talk over a rescue it invented.
Is a disclosed AI first-aid instructor acceptable on the product channel?
Disclosure does not move a constructed health expert out of YouTube's third bucket. Keep the pouch plate. Keep the overlay. If someone must speak, put a real named clinician on camera. Do not generate a rescue or a photoreal clinician so the mouth has a subject.