This entry covers the push-in — a camera movement where the camera physically moves forward toward the subject: when to push in, how to keep it steady, and how it differs from a zoom-in. It suits shooters just learning camera movement, and anyone who needs to direct the camera precisely when writing AI video prompts.
When to use
Good for: drawing the audience’s attention from the environment to the subject (pushing toward a face in a crowd); building emotion (slowly closing in on a person for tension or intimacy); revealing a key detail (pushing toward a key on a table).
Avoid when: the frame has no clear focal point (empty scenery); fast-paced dialogue cutting back and forth; documentary segments that need an objective, observational stance.
How to do it
- Decide the end point first: know exactly where the push stops (e.g., medium shot to close-up) → work backward to set the starting composition and path → both the opening and closing frames should be fully composed shots, not “just walk forward.”
- Steady speed, straight line: lower your center of gravity, take small slow steps (roll from heel to toe), ideally with a gimbal → the frame stays smooth with no bounce.
- Lock focus: as distance changes, focus drifts — use continuous autofocus locked on the subject, or stop down the aperture for deeper depth of field → the subject stays sharp throughout.
- Match speed to emotion: a slow push is a gaze or buildup; a fast push is surprise or impact → rhythm is emotion.
- Hold at the end: after reaching the final position, hold for 1–2 seconds before cutting → gives the editor room to work.
Parameters / key points
- Travel distance: commonly 0.5–2 m; 0.3–1 m for small indoor scenes
- Duration: slow push 4–8 s; standard 2–4 s; impact-style ≤1.5 s
- Focal length: 24–35 mm gives noticeable perspective change and strong presence; 50 mm and above gives a gentler, more composed shift
- Push-in vs. zoom-in: a push-in physically moves the camera forward, so the relative positions of foreground and background change (parallax) — it feels like “walking into” the scene. A zoom only magnifies; perspective stays the same — it feels like “being pulled closer.” On a phone, don’t use digital zoom — just walk closer. Combining the two in opposite directions is the advanced “Hitchcock zoom” (covered separately in the option pool).
AI prompt writing
- Core keywords:
push in/dolly in - Chinese template: 镜头缓慢向前推进,从〔起始景别〕推至〔结束景别〕,聚焦〔主体〕
- English template:
slow push in from a medium shot to a close-up of [subject], smooth dolly movement - Key points: specifying the starting and ending shot sizes is far more reliable than just writing “push in”; use slow / steady to control speed; if you want the “walking in” perspective effect, write dolly in — not zoom in
- Model response differences: unverified (to be filled in after real-world testing)
Common mistakes
- Starting the push without knowing the end point → the closing composition is incomplete and the whole take is unusable
- Using zoom to fake a push-in → you get magnification but no sense of presence
- Inconsistent speed or shaky footsteps → makes viewers dizzy; better to shorten the distance than lose smoothness
- Pushing in too far until you’re right in the face → focus loss and distortion; leave some breathing room
- Cutting immediately at the end without holding → the editor has no way to close the shot