Use vertical video reframing to keep both partners visible

Name the required content before changing the canvas

For every shot, list what a viewer must see. In an approach, that may be both faces and the distance between them. During the embrace, it may be the hand-to-shoulder contact. In a memory, a shared object may explain the relationship. These requirements determine the crop more usefully than automatic centering on the nearest face.

Mark the requirements on a still and then watch their movement through the clip. A crop that fits both heads at the start can lose a hand or the farther partner during a turn. The whole usable range needs to fit the chosen composition, or the window needs a deliberate movement.

CapCut's desktop cropping guide documents crop and aspect-ratio operations. Use its current controls for the application you have. This article supplies original framing calculations and shot decisions; it does not claim the Zombie Couple preview has a video reframing editor or an automatic two-person tracking feature.

Calculate a 9:16 crop from a 16:9 source

Take a 1920 by 1080 source and keep its full height for a 9:16 crop. The required width is 1080 multiplied by 9 divided by 16, which equals 607.5 source pixels. That is about 31.64 percent of the original width. A center crop therefore discards roughly 68.36 percent of the horizontal image, split across both sides.

A practical editor may resample or align the crop to whole pixels; 607.5 is the geometric result, not an instruction to export a fractional pixel. To fill a 1080 by 1920 output, the source is scaled by 1920 divided by 1080, or about 1.7778. Its scaled width becomes about 3413.3, of which the output shows 1080.

These equivalent calculations explain why two people far apart in a wide shot disappear in a vertical fill. Increasing the export resolution does not restore the discarded composition. Check the required subjects against the narrow source window before spending time on fine positioning.

Test the people as a group rather than two centers

Measure the horizontal span needed to retain the meaningful content. In an original example, the leftmost required face edge is at x = 520 and the rightmost required hand edge is at x = 1130 in a 1920-wide source. The required span is 610 pixels. It already exceeds the 607.5-pixel full-height window before adding breathing room. That shot cannot preserve those exact requirements with a static full-height 9:16 crop. Moving the crop left or right only chooses which edge to lose. If the hand is optional, revise the requirement explicitly; if it explains contact, choose a different layout or shot. Do not call an incomplete crop acceptable simply because both eyes remain visible.

When the span does fit, center the crop around the required group rather than the frame's original midpoint. Include room for expected movement and inspect the entire range. A face on the edge may be technically inside the frame while feeling cramped or disappearing after a slight lean.

Move the crop only when the action calls for it

A moving crop can follow a shared action that travels across the wide source. Set its beginning and ending positions around the required group, then watch the movement at ordinary speed. The reframing should feel like a coherent camera choice rather than a series of corrections whenever a face approaches an edge.

Avoid tracking A independently when B's response remains part of the shot. A crop that follows the moving partner can abandon the stationary partner at the emotional decision. If the action requires seeing both and the group span becomes too wide, no tracking path solves the geometry. Use a wider fit or another angle.

Keep crop movement separate from source camera movement in your notes. A source pan plus a crop pan can accelerate the perceived motion or reverse its emphasis. Compare the reframed shot with the wide original to confirm that viewers still understand who approaches whom and where the contact occurs.

Fit the wide shot when the full relationship matters

Fitting a 1920 by 1080 source to the 1080-pixel width of a 1080 by 1920 canvas gives a scale factor of 0.5625. Its displayed height becomes 607.5 pixels, leaving 1312.5 pixels vertically outside the image area. With a centered fit, that is 656.25 pixels above and below before any layout adjustments.

Those numbers describe the layout tradeoff: the whole interaction remains visible, but the people are smaller on the phone screen. Use the surrounding space intentionally, with a restrained background or concise story text when appropriate. Do not fill it with decoration that competes with small faces. Review the result on an actual phone-sized view.

A blurred duplicate background is a possible editorial treatment, not recovered scene content. Keep it subdued and inspect whether a duplicated face creates a confusing second pair. A plain background may communicate the wide composition more cleanly. If the main image remains too small to read, select a closer shot or create a separate composition.

Reframe connected shots as a pair

At an embrace match cut, compare the crop on both sides together. The hand-to-shoulder anchor should remain visible and land in a comparable area. Centering each shot independently on its nearest face can create a new position jump even when the wide version matched well.

Check screen direction across the sequence too. A crop may hide the doorway or object that explained where a partner came from. The next shot can then feel spatially disconnected. Keep a little context when it carries information, or use an establishing shot that remains legible before moving into a tighter composition.

Place captions and other overlays only after the important picture content is protected. Review against the destination's actual preview because interface elements and text can cover the edges. This guide does not invent a universal safe-zone measurement for every platform. Use a conservative layout and inspect the rendered preview for the current publishing destination.

Inspect the file that will actually be published

Export the intended vertical version and open that file independently of the editing project. Check every shot's start, hardest movement and end, plus every match cut. Confirm the output dimensions and orientation. A correct canvas setting does not prove the export used the same settings or the same timeline version.

Watch at ordinary size before zooming in. Can you recognize both partners and see the gesture without searching? Then inspect any edge concern more closely. If the composition depends on a face or hand barely touching the boundary, add room or choose another layout before publication.

Keep the wide master and a short reframing log. Note which shots were cropped, moved, fitted or replaced. A future square version should be derived from the accepted wide composition rather than from a vertical copy that already discarded most of the horizontal information. The practical result is a delivery version that preserves the story, with explicit choices for each difficult shot.

• Both people remain visible wherever the story requires them.
• Hands and contact anchors survive the entire movement.
• Connected shots retain compatible positions.
• Overlays and platform preview controls do not obscure the action.
• Export dimensions match the intended delivery example.