video-reframe — Traditional AI Model
Video Reframe pushes a list of clips to 9:16 at 1080×1920 for Shorts and Reels, audio intact. Fit over a blurred backdrop keeps the whole frame, so nothing important is cropped at the edges.
Details
- CategoryTraditional
- Year-1
- License
Inputs & Outputs
- ListInputlist:video
Video list to reframe to 9:16
- ClipsOutputlist:video
Reframed 9:16 clips
Parameters
- modestringdefault Fit + Bluroptions: Fit + Blur, Fit + Black, Crop to Fill, Subject-Aware
Reframe strategy. 'Fit + Blur' centers the whole frame over a blurred background; 'Fit + Black' uses solid black bars; 'Crop to Fill' crops to fill the frame with no bars; 'Subject-Aware' crops to fill but keeps the tracked subject in frame — a person's face first, then (when no face is present) a pet/animal or the person's body, falling back to a centered crop. Fit modes never crop.
- blur_strengthintegerdefault 20range 5–50advanced
Background blur strength for 'Fit + Blur'.
- framingstringdefault Autooptions: Auto, Full, Face Close-up, Mouth (Money Shot), Food Close-up
Subject-Aware punch-in (virtual multicam). 'Auto' picks each clip's framing from the analysis shot_type (wide→full, face→face, mouth→mouth, food→food) for an automatic intercut; 'Full' keeps the subject at natural size; 'Face Close-up' punches in tight on the face; 'Mouth (Money Shot)' punches in tightest and anchors on the mouth (best for mukbang); 'Food Close-up' tracks the dish instead of a person and punches in on it. Explicit choices force the same framing on every clip. Only applies to 'Subject-Aware'.
- zoomfloatdefault 0range 0–4advanced
Subject-Aware punch-in override. 0 = auto (use the framing preset's zoom); 1.0 = no punch-in; up to 4.0 tighter. Only applies to 'Subject-Aware'.
- subjectstringdefault Autooptions: Auto, Person, Pet (cat/dog), Animal
What to track in Subject-Aware mode. Auto tracks people and animals together; naming one stops the others competing.
Which subject the Subject-Aware tracker follows. 'Auto' (default) considers people and common animals together and picks whichever occupies the most frame area — which is wrong when a bigger, nearer animal shares the shot (measured: a dog in a farm pen lost to a nearer alpaca on every frame, because alpacas/sheep are in the same candidate pool). Name the subject to remove that competition: 'Person', 'Pet (cat/dog)', or 'Animal' (animals only, no people). A non-person choice also skips face/active-speaker detection, so a human in shot cannot steal the track from the pet. Only affects mode = Subject-Aware.
- camerastringdefault Noneoptions: None, Slow Push-In, Slow Pull-Out, Cut Punch-In, +1 more
Subject-Aware camera move. 'Slow Push-In'/'Slow Pull-Out' ease across the WHOLE clip — a drift you are not meant to notice, which keeps a long hold from looking frozen. 'Cut Punch-In'/'Cut Punch-Out' instead confine the same move to the first fraction of a second and then hold, so the cut itself gets an accent you ARE meant to see. The move runs at the clip's own frame rate and never changes its length. Only applies to 'Subject-Aware'.
- camera_amountfloatdefault 0range 0–0.5advanced
How far the camera move travels, as a fraction. 0 = auto (0.08 for the slow moves, 0.18 for the cut punches — a slow move is meant to stay under the threshold of 'a zoom happened', a punch is not). Note the cost: on a 16:9 source at zoom 1 the frame has no spare resolution, so the tight end of the move is enlarged by this much. Only applies to 'Subject-Aware' with a camera move.
- camera_ramp_msintegerdefault 250range 80–800advanced
How long a 'Cut Punch' takes to land, in milliseconds; it then holds for the rest of the clip. Ignored by the slow moves, which always span the whole clip. Only applies to 'Subject-Aware' with a Cut Punch.
- bandfloatdefault 1range 0.5–1advanced
How much of the source height to keep, measured from the top, as a fraction. 1 = keep it all (default; the clip renders exactly as before). Lower values trim the BOTTOM of the frame before reframing, which is the only way to remove a caption burned into the bottom of a landscape source — the 9:16 window keeps full height, so no crop this node makes can drop it. Note the cost: the trim takes whatever else is down there with it (a dog filmed from behind has its tail in the bottom sixth), and less height to cover means a larger upscale, so the delivered frame holds less of the source width. Set it per clip, not once for the whole video. Only applies to 'Subject-Aware'.
- pan_fromfloatdefault 0range 0–1advanced
Where the 9:16 window sits horizontally, as a fraction of the source width (0 = hard left, 1 = hard right). Leave at 0 with 'pan_to' also 0 to let subject detection choose, which is the default. Set it when detection picks the wrong subject, or when the framing is a deliberate choice rather than a follow. Only applies to 'Subject-Aware'.
- pan_tofloatdefault 0range 0–1advanced
Where the window ENDS, as a fraction of the source width. Leave at 0 to hold at 'pan_from' for the whole clip; set it to travel from one position to the other over the clip's length — 'open on one subject, close on another' when the two are too far apart to share a vertical frame. Ignored unless 'pan_from' is set. Only applies to 'Subject-Aware'.
Tags
Alternatives in Video Editing
- Audio Edit
Trim a sound, fit it to a video or a length, dip or mute parts, and fade it.
- Audio Finish
Normalize a video's loudness to a broadcast target (2-pass loudnorm + peak limiter).
- Audio Mix
Mix narration, music and sound effects into one track, with ducking and loudness.
- Audio Replace / Mix
Replace or mix a clip's audio with a new sound, keeping the video as is.
- Chroma Compose
Cut a green-screen person out of one video and place them over another.
- Hook Teaser (Cold Open)
Prepend a cold-open teaser of each clip's hook moment to the front.
- Ken Burns (Still to Video)
Animate a still image into a video with a moving crop window.
- Narration Mux (Cold Open)
Prepend a voiceover cold-open (frozen first frame + narration) to a clip.