C-arm OEM: the arc still, empty table, never a generated fluoro limb

Photograph the C-arm. I2V after identity holds. Do not invent anatomy on the detector or a talking rad-tech.

A C-arm listing that books is hardware: the C, the detector housing, the generator cart, the empty table in the suite you actually ship. It is not a constructed wrist on the receptor and not a photoreal rad-tech walking a protocol. Photograph the arc. Leave tissue off the plate.

The empty arc is the catalog plate

A mobile C-arm is a SKU with a serial, a detector size, and a cart a biomedical engineer will recognise. Text-to-video will invent a prettier cousin: a different C radius, a cleaner paint job, a logo that is not on your hardware. That is a different machine.

Photograph the unit you sell. Table empty. No one on the pad. No phantom in the beam. If the only photo includes a limb on the receptor, crop until only metal remains, or shoot again. Do not prompt the anatomy out and hope the model rebuilds the detector.

MRI ads are the gantry still, not a talking radiologist is the nearby imaging rule: the bore is the facility plate; a constructed clinician is a kill. This listing is OEM. The magnet post owns the suite a patient walks into. This post owns the C you manufacture. Do not rewrite the gantry page as a fluoro commercial.

Re-describing “a compact 9-inch C-arm in a cinematic OR” is how the model starts drawing a second C. The C, the wheels, the detector face survive because they were photographed.

I2V after the identity holds

Image-to-video starts from the frame you upload. Frame one is the file. Motion is allowed only after the still is true: a slow orbit of the C, a tilt to the detector, a hold on the empty table edge. Camera and physics. No new ports. No new collimator. No invented fluoro image on the glass.

Kling Video V3 Standard Image to Video is the catalog row for that job: image required, native audio, 3s through 15s, 63 credits listed. Aspects 16:9, 1:1, 9:16. No max output resolution is listed on this slug. Do not invent 1080p or 4K. Camera control and element composition are listed; they do not license a wrist on the receptor. The still you upload is the still you will watch move.

3s, 4s, and 5s are the honest tests. The spec will let you go to 15s. Identity drift likes the long end. Until the C holds, stay on 3–5. A 15-second miss is a long miss of a melted detector.

Native audio is on. Write “no speech, no invented clinician, room tone only,” or you will get a mouth you did not cast. A silent brief still mints a soundtrack. Do not ask the model to narrate kVp. A single hero still of the empty arc with a slow move is the usual OEM ad.

Never a generated fluoro limb

Prosthetic shops: the component still, never a generated residual limb is the same body rule applied to a stump. A fluoro limb is that rule on the detector. A generated wrist, a “generic ortho case” from a teaching file — if someone who knows them would say that is their anatomy, you have a replica problem. Calling it demo fluoro does not move the statute.

Do not prompt a photoreal hand under the C so the beam “looks in use.” Do not upload a case photo and ask image-to-video to anonymise the bone. Do not invent a residual limb or a reduction on the receptor. A child-sized forearm is a hard stop.

Ship the empty C, the empty table, the detector with no anatomy, and type you authored after lock. A calibration pattern you photographed on this receptor is hardware. A generated radius is a person.

A talking rad-tech is a different file

Synthetic presenters on health and finance are a YPP kill. Help’s third inauthentic-content bucket is AI-generated personas that present as human experts on health. A constructed rad-tech, “fluoro nurse,” or radiologist telling a viewer how to collimate, when to pulse, what dose is “safe,” is that shape. Labelling the face AI does not leave the bucket.

Do not glue a talking white coat onto the C inside one native-audio generate. If a protocol needs a voice, that voice is a real named credentialed person on a different day. Generation can still do the arc. It cannot be the clinician.

Hours, a SKU, “9-inch receptor” — overlay after picture lock. Add a text overlay burns one authored line, whole clip, top / center / bottom. It does not transcribe. Write camera only: “Slow orbit of the empty C, table static, no people, no anatomy on the detector.” Cover the logo. If a biomedical engineer cannot match the C to the quote, you made a health clip, not the OEM listing.

FAQ

If I prompt “no medical advice,” can Kling still put a rad-tech next to the C?

The face is still a constructed presenter. Fluoro talk is still health. Hold the empty arc. If someone must speak, put a real named credentialed person on camera, or do not prescribe.

Can I generate a limb on the detector if I never show a face?

No. Photoreal anatomy on the receptor is still a generated patient. Photograph the empty table and the empty detector. Do not invent tissue to prove the C is in use.

Do we start from a photo of our actual C-arm?

Yes, when the SKU has to be ours. That is image-to-video from the still, not a paragraph about a cinematic OR. A stock C is a different serial.

Why not a 15-second take on the first try?

Kling V3 Standard I2V lists 3s through 15s, native audio, 63 credits on the catalog figure. Identity drift likes the long end. Prove the C on 3–5 seconds. Pad only after the detector still matches the still you uploaded.