Detailing ASMR is Vidu diegetic sound, not TTS on a closed mouth
Q3 I2V for the spray and towel. TTS is a different file.
Q3 I2V for the spray and towel. TTS is a different file. Detailing ASMR is the wet panel, the hiss, the towel drag. It is not a closed-mouth host with a voice laid on top.
A detailing clip that works is an object doing object things: foam hitting paint, a microfiber catching a bead, a nozzle clicking off. That sound has to be born with the motion. A line about "ceramic protection" on a silent mouth is a different file.
The spray is a still with a mix
Photograph the real panel, the real bottle, the real towel. If you would not post the still, do not buy motion.
Vidu Q3 Image to Video is the close: image required, audio on, durations only 5s / 10s / 15s, qualities 360p through 1080p, listed 16 credits and marked discounted. Features named: image-to-video, motion generation, high fidelity. No aspect-ratio menu on this slug — crop 9:16 or 16:9 before upload.
Write the mix into the prompt — spray hiss, foam crawl, towel drag, no music, no speech — because a silent brief still gets a score. Atmosphere lives here. A locked spoken line does not.
The shortest legal clip is 5s. If the close fails in five seconds, you do not have a 3s cheaper retry on Vidu Q3. Fix the still. The contract is the still, not the slider. Spray sound will not save a bottle the shop does not sell.
Image-to-video keeps the upload as frame one. Prompt motion. Do not re-describe the car.
TTS is a different file
Native audio versus TTS on the same brief is the split. Native audio is picture and soundtrack from the same pass — the hiss lands on the spray because both came out of one generate. TTS is a voice you control: a cloned founder, a style-locked narrator. They are different contracts. Pick one per mouth.
A detailing ASMR take has no mouth. Generating a silent Q3 clip and dropping TTS on a closed face is two products fighting one frame: the lips belong to nothing, the voice belongs to a person who is not in the shot, and the spray no longer owns the mix.
Do not generate Q3 with native audio, duck it, and lay a founder line on the same take. Do not prompt a talking detailer "so we have a hook" on the spray still — that is a presenter row. Captions are how the file works muted; burn them after the mix is decided, not instead of deciding.
If you need a spoken claim, shoot or generate a talking take on a mouth-capable row, or overlay the line. Do not ask Vidu Q3 to lipsync a script you never attached. Native audio on this slug is a mix you specify, not a locked VO.
Keep the still, write the hiss
Three stills beat one prompt: wet quarter-panel, spray in hand, towel on glass. Animate each on Q3 with a different mix note. Cut. The test is whether a customer who just paid would say that is the bay.
Unspecified audio is how you get a music bed the shop never played. Name two or three sounds. State "no music, no speech" when the deliverable is ASMR. Vidu Q3 native-audio product stories is the same rule: write the soundscape or the model invents one.
Keep TTS for the week you need the same voice across thirty episodes. Keep Q3 for the spray.
FAQ
Can I generate a talking detailer and call it ASMR?
No. A talking shot is a mouth job. ASMR here is diegetic spray and towel. Run Vidu Q3 I2V on the still. If you need a host, that is a second file on a talking row — not TTS glued to a closed mouth.
Does Vidu Q3 require an image?
Yes. Image-to-video, image required. "Transform static images into dynamic videos" is the catalog line. A text-only bay is a mislabel, and it will invent a car.
When is TTS the right detailing file?
When the voice is a contractual object — a named owner, a series narrator — and nobody in frame is speaking. Off-screen VO over a silent close is TTS's home turf. On a visible mouth, native audio is the default. Do not mix both on one face.
Can I stitch three 5s Q3 takes into a 15s detail?
You can cut. You cannot pretend it is a single-pass hold on another row. Q3 durations are only 5 / 10 / 15. Write the action to one of those buckets, or change rows.