Hospital beds: first-last of the deck rise, empty mattress
Two stills of the same empty bed, low then high — Flux 3 first-last, never a generated patient on the deck.
Two stills of the same empty bed, low then high. Flux 3 first-last. Do not generate a patient on the deck.
The product is the deck rising. Frame A is the mattress at the low stop. Frame B is the same mattress at the high stop. Same rail, same headboard, same empty vinyl. Interpolation is the ad. A generated body on that vinyl is a patient you did not photograph, a likeness you do not have, and a health clip the moment a mouth starts advising.
The empty mattress is the SKU
Hospital-bed OEM video is not a ward tour. It is the mechanism on the catalog cut: Hi-Lo travel shown between two photographs of this frame. The buyer is a facilities planner or a biomedical tech. They already know a mattress. They need to see the deck move without the rails morphing into a cousin SKU.
Photograph the empty bed at the low position. Lock the tripod, lock the lens height, lock the room lights. Raise the deck. Photograph the same empty bed at the high position. Those two files are the brief. A paragraph that says "cinematic hospital room, patient resting, bed quietly elevating" is a different object. It invents a person, a room, and a bed.
First-last frame turns an open-ended generate into a bounded landing. Ordinary image-to-video knows where it begins and guesses the rest. First-last has to finish on the high still you approved. The rail geometry, the side-rail latch, the caster lock — those survive because they were already in both pictures.
Do not composite a mannequin. Do not prompt a sleeping figure as scale. Empty is the point. Scale is the headboard and the wall plate that were already in the shot. If the mattress has a wrinkle, keep the wrinkle. A model that "tidies" the sheet is already rewriting the SKU.
Flux 3 interpolates the rise
Create a first-and-last-frame transition video is the named job: two photos in, one transition out. The agent passes them as first and last frame and picks a first-last-capable model. Cost is priced per model, shown before you confirm. This is not the plain "turn one photo into a video" capability.
Flux 3 First Last Frame to Video is the catalog row: start frame and end frame, native audio, up to 1080p, durations from 5 seconds through 20. The listed rate is 9 credits per second. Audio is on. Write the mix you want — a quiet actuator, a padded room — or write silence. A silent brief still gets a soundtrack you did not choose.
Prompt the travel, not the furniture. The stills already contain the deck. Name the mechanism: locked camera, slow rise, no new rail, no one entering frame. If the endpoints share geometry, the morph reads as a Hi-Lo demonstration. If they do not — a low still of one serial into a high still of another, or a pendant close-up into a wide of a different bay — you get a warp you could have made in an editor without a video bill.
Inspect the midpoint. Ends are held by construction. The middle is where a generator cheats: an extra siderail, a pillow that appears, a crease that becomes a knee. If a body starts to form in the sheets, discard the take. Reroll the interpolation. The stills stay empty. Five seconds is enough for a deck rise. Twenty seconds invents a second headboard.
Do not generate a patient on the deck
A photoreal person on a hospital mattress is a patient. You did not cast them. You do not have a release. Synthetic presenters on health and finance are a YPP kill when that face starts advising. Even without a mouth, a generated body on a medical deck is the wrong object. Facility and SKU stills. Overlay if you need a model number. No generated patients, no fake clinicians.
Do not text-to-video a peaceful patient. Do not use a last frame that has a person if the first frame does not — that is a materialisation, not a rise. Do not glue a constructed nurse explaining Hi-Lo onto the interpolate. If a human has to speak, that human is on payroll and on camera on a different day. A talking generate about recovery is health advice from a persona. Labelling the face AI does not leave the bucket. The empty-deck rise is machinery.
OEM SKU, not the urgent-care page
/for/urgent-care-clinics is a facility page: waiting room, hours, the plate of the clinic. This post is the bed the OEM sells into that clinic. Do not rewrite the clinic page as a Hi-Lo demo. The nearby audience is the facility. The SKU is the empty deck.
Home-care pages are clinician and caregiver. They are not this listing. Hours go on overlay after picture lock. Speech is a real named person. The interpolate is only low mattress to high mattress. A loop (the same still in both slots) is ambient motion on one plate. This offer needs two states.
FAQ
Can I text-to-video the rise if I do not have two stills?
You can. You will get a bed the catalog did not sell, and often a body on it. Photograph the empty deck at both stops, then interpolate.
Why Flux 3 instead of a talking-head model?
Because the product is interpolation between two approved stills, not a mouth. Flux 3 first-last is 5–20 seconds, up to 1080p, native audio, 9 credits per second. A presenter row will talk over a ward it invented.
Do the two photos have to be the same serial number?
They have to share geometry: same rails, same headboard family, same standing spot, same lighting neighbourhood. A low still of SKU A into a high still of SKU B is a morph the buyer will clock as a different bed.
Is this the same as uploading one image of the low deck to image-to-video?
No. One image is open-ended motion. The ending is then the model's guess, and the guess is often a patient. Pin the empty high still as the last frame or you are not selling the rise.