Bowling is pin-fall diegetic sound
Vidu Q3 if the still is the lane. TTS on silent pins is a different file.
Vidu Q3 if the still is the lane. TTS on silent pins is a different file.
A bowling clip that works is the ball, the rack exploding, the sound landing on the impact. That noise is in the world of the shot. Animate a locked lane photo and drop a voiceover on mute pins, and you have two products glued together: a silent generate plus a speech file that does not know the crash happened.
Vidu Q3 Image to Video is the row when the still is already the lane. Image required. Audio on. Durations only 5s, 10s, and 15s. Max output 1080p. Sixteen credits listed; the row is marked discounted.
The crash has to be born with the picture
Native audio means soundtrack and frames from the same pass. Pin-fall is the textbook case: the hit is visible, so the hit has to be audible in the same denoise. Retrofitting a "strike" sample onto invented motion is how the crash lands on the backswing.
Write the mix into the 5s prompt. Name two or three sounds: ball on wood, pin explosion, room tone. State "no speech, no music" if you want clean foley. Leave audio unspecified and Q3 invents a mix — including a murmur that does not belong in a house.
Q3 will not find your logo in a paragraph. Crop the still to 9:16 or 16:9 before upload; this slug has no aspect-ratio list. If the still is the wrong lane, every second is waste.
Three lengths, one still
You pick 5, 10, or 15. No 3s retry, no 8s "almost a hook." A roll-and-crash is often five seconds. Do not pad to 15s because the slider has two more stops. A discounted 15-second miss is still a wrong lane with a soundtrack attached.
1080p is the ceiling, not 4K. Spend 1080p on the 5s take you will actually post.
Native audio versus TTS on the same brief is the tree. Mouth in frame: native. Off-screen VO that must be the same manager next week: TTS. Pin-fall with no face: native foley.
TTS on silent pins is a second contract
Text-to-speech is a voice file. AI text-to-speech is that launcher. It does not know the ball. Laying it on a silent generate is presenter energy over dead wood. Laying it on native audio is two voices fighting one world.
Use TTS when the shot has no mouth and the line has to match next week's reels. Keep the pins native. Duck the VO under the crash; do not replace the crash with the VO.
Do not generate Q3 silent-on-purpose. Audio is on. A "silent" brief still gets a band. Mute plates belong on a silent image-to-video row.
Would you hear it with the picture off?
Play muted. If the pins look like they should clang, and unmuted they do, you used Q3 correctly. If a cloned host is talking while the rack is frozen, you used TTS as a substitute for physics.
Caption after. Five seconds of native audio is not a subtitle track.
FAQ
Can I generate a 3-second pin-fall on Vidu Q3?
Not on this ladder. Durations are 5s, 10s, and 15s. The shortest legal clip is five seconds. If you need a three-second hit, pick a model that lists 3s, then cut. Stretching a 5s Q3 take in an editor is a later decision.
Does Q3 require a photo of our actual lanes?
Yes, as image-to-video. The still is the contract. A text-only "bowling alley at night" is a different family. If the house, the oil pattern, or the logo must survive, lock that frame first.
Should I TTS the "strike!" call over the clip?
Only as off-screen VO, and only if that voice has to match next week's reels. The pin-fall itself should stay native. Do not mute Q3 and replace the crash with a shout. That is two files pretending to be one event.
Is 16 credits the whole cost of a 15s 1080p take?
16 is the listed row figure; billing still depends on duration and resolution in the matrix. Treat 16 as the named cost, not a cap you can ignore at the long end. The still has to be right before any of those numbers matter.