Purifiers: the unit still, captions for CADR, no medical VO

Captions not a clinician.

A purifier clip that holds is this tower — grille, door, control cluster — then motion on that picture, then CADR as a line you typed. It is not a talking clinician. It is not a model painting "300 CADR" onto the fascia. The rating already exists on the box. Burn it. Do not ask a sampler to typeset it, and do not ask a synthetic doctor to explain lungs.

Lock the tower before you buy seconds

The SKU is plastic, a filter door, a logo, a display that repeats a lab number. "Sleek HEPA in a sunlit nursery" is a casting call. It will not match the PDP.

If the still would not print on the carton, stop. Zoom until you can read the fascia. Text-to-video of "premium purifier, golden hour" invents a cousin tower. Retailers do not list cousin towers.

Animate the locked pack through AI image-to-video on the Seedance 2.5 family: 4s–30s, 720p, 47 credits, native audio. Name only motion — a slow turn, outlet vapour, a hand on a filter that is already in the still. Do not restate CADR in the paragraph. Repeating the number is how the display gets redrawn.

A closing still is useful when you already signed the end composition. Do not close on type the open frame did not carry. Another I2V take on the same unit is another take of the same SKU. Another cinematic paragraph is another tower.

CADR is a lab string. Overlay it.

CADR is not a vibe. Legal already signed it. Glyphs generated on the housing are a misspelled rating you cannot edit. Keep the pack clean. After picture lock, type the string. Check it against the carton.

Add a text overlay is one authored line for the whole file: CADR, room size, a CTA. Top, center, or bottom. Size, color, optional scrim. The tool does not listen. Nobody has to say the number out loud.

CADR, then filter stage, then a CTA, each on a clock, is timed text overlays. That is still your copy. It is not a persistent badge pretending to be subtitles, and it is not a transcriber.

Reframe to 9:16 before you burn. Placement enums move when the crop moves. If the lab number or the picture is still in review, wait. Offers change. CADR does not, until the lab says so.

Captions listen. They do not write CADR, and they do not translate.

Add captions to a video is VEED speech-to-subtitles: 21 BASIC presets, 9 DYNAMIC at double, 165 language codes. It transcribes. It does not translate. The language code has to match the language on the tape. Noise and overlapping talk degrade the transcript. Run it on a keeper that actually has speech.

A mute tower has nothing to transcribe. Captioning it spends a pass on empty air. Overlay the CADR instead, or record a human and then caption that human.

A product line from a real person — "this is the unit on the box" — can wear captions. A constructed clinician cannot. Motor noise is not a health claim.

Seedance 2.5 lists native audio. Fan and room are allowed. A generated expert on lungs is not. Native audio is a bed.

No medical VO

YouTube's third inauthentic-content bucket is AI-generated personas that present as human experts on health, legal issues, finances, or politics. CADR plus asthma plus "run this in a nursery" is health advice in a filter housing. A synthetic clinician on a spec sheet is the same shape as a synthetic doctor on a protocol. Disclosure that the face is AI does not leave the bucket.

A turning tower with a burned CADR is not that shape. The kill is persona plus prescription: a constructed face styled as the person who knows air quality.

Do not glue a medical voice onto the pack inside one generate. A lab report that needs a voice needs a real named person, on a different day. Generation can still turn the unit. It cannot be the clinician.

Hold the SKU. Overlay the number. Caption a human if a human spoke.

FAQ

Can the motion prompt paint CADR on the LED panel?

Do not. Restating the number invites a redraw. Overlay the approved string after picture lock.

Can I generate a clinician to walk through CADR?

No. The rating is overlay. A synthetic expert on air and wellness is a health-adjacent persona. Hold the unit. If you need a voice, caption a human.

Will captions turn a Chinese VO into English CADR copy?

No. VEED transcribes; it does not translate. 165 language codes select the spoken language. Overlay English CADR you wrote, or dub first and caption the dubbed file.

Do I need a last frame for the open filter door?

Only if that door is a second signed still. Both frames have to be approved. The filter is a product photo, not a sentence in the prompt.