NICU isolette ads overlay servo mode on the empty hood

Photograph the closed, empty isolette and overlay servo mode as authored type; Luma Photon Flash (4K, 1cr) can hold that plate — a generated neonate or talking neonatologist is a YPP kill, so do not I2V a body into the hood.

A neonatal listing a buyer can trust is the hood on your floor: latch, portholes, the servo badge a biomed will match to a bid. It is not a constructed infant or a photoreal neonatologist walking temperature. Shoot the closed, empty unit. Burn the mode string after the picture exists. Leave the body off the generate.

The empty hood is the plate

The buyer has stood next to this isolette. They know the porthole spacing, the hinge colour, the order of the ports, the height of the canopy. Text-to-video of “a cinematic NICU with a sleeping newborn in a modern isolette” will invent a cousin hood in a prettier bay. That bay is a different hospital. That infant is a private figure you did not consent.

Photograph the unit with the hood closed and empty. Same lens height a tech uses on a walkthrough. Lights as they actually sit. No residual anatomy. No dummy. No extra in scrubs at the side. Crop a neighbouring warmer if it is not the SKU you sell. Hold the nameplate if the listing needs it. Occupied is a likeness problem and a health generate.

If the only photo includes a patient, crop them or shoot again. Do not prompt the infant out and hope the model rebuilds the mattress. The cheap place to lock the hood is the photograph. A generated neonate is a hard stop even if you never show a face.

Do not send “photoreal isolette in a hospital nursery, baby inside, ready for use” to a still model and call it the install. That paragraph decides a room and a body. The room is not yours. The body is not yours.

Photon Flash holds a still; it does not occupy the hood

Luma Photon Flash is the stills row: the speed tier of the Photon image family. Near-instant photorealistic stills at the lowest credit tier in the catalog. 1 credit. 4K max output. Categories: text-to-image and edit-image. Audio empty. Durations empty. It is not a video model.

Use it when you need a closed, empty SKU still of the hood you sell — door shut, mattress empty, no body — labeled as a product render, not a named NICU. Prompt furniture and light only. Closed canopy. Ports readable. If a face is in the brief, the brief is wrong.

The text-to-image door is the generate half of this row. The edit-image half is the other: when a 1-credit hero is almost right, edit the glare or the crop here instead of rolling a new identity.

Photon Flash will not orbit. Prompting “slow push, 8 seconds” returns a detailed frame of a push that never happened. Approve that frame. Motion on the empty hood you photographed is a later image-to-video job on that file — camera and physics only. Do not I2V a body into the hood. Re-describing “a neonate in servo mode” is how the model starts drawing a patient.

A 4K invented nursery is still the wrong wall.

Overlay servo mode as type you wrote

Servo mode, air mode, the model string, “empty unit / no patient” if legal asked for it — those lines already exist on a nameplate and a datasheet. A still model asked to letter them into the canopy will mint cousin glyphs. That is a false spec.

Add a text overlay to a video burns one authored line for the whole clip once you have a locked plate: servo mode, SKU, a CTA. You write the line. Position is top, center, or bottom. Font size, color, and an optional background block are controllable. It does not transcribe. Photon Flash has no soundtrack, so there is nothing to listen to.

If you only have a still, typeset on the still in an editor, or animate the empty hood first and overlay after picture lock. Do not ask Photon Flash to stamp “servo” onto the glass. Proof the overlay against the cut sheet. Overlay before you reframe to 9:16 or the line sits in the chrome.

Do not:

• Prompt a temperature, a humidity, or a “stable” flag onto the display.
• Generate a printout, a graph, or a completion chime as proof of a case.
• Grow a limb, a blanket bundle, or a parent at the porthole.
• Put a talking face next to the hood to explain servo versus air.

If a hospital needs a mode on screen, that mode is a string they already publish. Motion does not create a reading. A still model does not either.

A generated neonate is a YPP-shaped health file

Synthetic presenters on health and finance are a YPP kill. YouTube’s third inauthentic-content bucket is AI personas that present as human experts on health. A constructed neonatologist walking servo mode or “what you should do with this infant” is that shape. Labelling the face AI does not leave the bucket.

A silent empty hood is not that shape. Authored type is typesetting, not a mouth. A photoreal infant in the canopy is a private figure you invented. Empty mattress. Overlay the mode. If a protocol needs a voice, that voice is a real named credentialed person on a different day.

Do not I2V a body into the hood to “humanise” the listing. A 1-credit 4K plate of the closed hood is the OEM job. Keep the hood empty. Burn servo mode. Stop.

FAQ

Can we generate a generic infant so buyers see how the hood is used?

No. A photoreal body in the canopy is still a generated patient. Photograph the closed, empty isolette. Overlay servo mode. A real named clinician, on a different shoot, if you need a voice. Do not invent a neonate so the unit has a scene.

Does Luma Photon Flash make video of the isolette?

No. Image content type, empty durations, empty audio. 4K stills at 1 credit, text-to-image and edit-image. Approve the empty plate here. Do not I2V a body into the hood later.

Why overlay servo mode instead of prompting it onto the canopy?

Mode is a spec. Photon Flash is a 1-credit 4K stills row; it does not typeset a string you will stand behind. Burn the line after lock, as an overlay you wrote.

If we disclose that the neonatologist is AI, can they explain servo mode?

No. Disclosure does not move a constructed health expert out of YouTube’s third bucket. Keep the empty hood. Keep the overlay. If someone must speak, put a real named credentialed person on camera.