The Master Guide / 10 Voice and lip sync · Speaking shots and lip sync
The camera during speech
During the words the camera breathes or pushes very slowly, the push written large enough to show; bigger moves come before the first word or after the last (decided).
During the words the camera breathes or pushes very slowly, the push written large enough to show; bigger moves come before the first word or after the last (decided).
Gentle movement is the default for every speaking shot, Arabic included (decided). A locked camera only where you choose it for that shot, written in the plan as a named exception with its reason: a deadpan comic beat, or a held confession the treatment asks for.
A breathing frame is a handheld that sways in place with the operator's breath.
A very slow push is written like any move: its start, its landing, its job, and its dose written twice, in centimetres and as the visible change of framing (see Camera and lens → The camera dose).
The working figure [best available, not yet validated]: over the line the face gets about a fifth bigger in frame, from a push of about a sixth of the camera's distance: 20 cm at 1.2 m, 35 cm at 2 m.
Bigger moves (an arc, a drift, a step to a new framing, a wide orbit, a reveal) sit before the first word or after the last.
The limits while the words run limit the move; they are never a reason for a still camera:
the face mostly toward the lens, within about 45° of it;
the mouth and eyes in frame, never cut by the edge, never behind a foreground object;
no whip, snap zoom, crash zoom, fast reframe or sudden turn;
the move never changes which side of the face the mouth is seen from.
The motion standard still holds: a static camera where the plan wants motion fails, and a lock you name for the shot is judged as a lock, with the talent and the place still alive (see Video with Seedance → Take selection and review).
Why 45°: Seedance's plate must keep a steady head position through the line. Sync-3 itself has no fixed angle limit and handles profiles.
Check the moving plate on every take: the draft check and the first-plates check, then the ±2-frame windows at the widest point of the move (see Speaking shots and lip sync → Checks on the 1080p result). If the move breaks the lips, slow the push and keep it moving. A lock is your choice, never an automatic fallback.
Measure it free on the sample batch, from takes you made anyway: how much of each written speaking move arrives (see Camera and lens → Measuring the dose on the sample batch).
Why: creators keep the camera alive through dialogue, but small. Vacancy and Davide Morando's The Weight of the Moon keep a near-still frame with a slight handheld breath through the words; Adham Sharkawy's Tommy pushes in subtly toward the speaker. None arcs or drifts through a spoken line. Sync's own tips ask for a near-frontal face. And Seedance can under-deliver a small written move, so the push is written large enough to read.
Sources
Vacancy by @narrowingthegap-ai (project); Davide Morando's The Weight of the Moon (project); Tommy by Adham Sharkawy (project); Sync, improving lip-sync quality.
