How to Convert Horizontal Video to Vertical Without Losing Quality

This works in Premiere Pro, DaVinci Resolve, CapCut or any editor with a custom canvas size. The details for each step come later in the article.

A 16:9 frame at 1080p is 1920x1080 pixels. A vertical 9:16 slice of that frame at full height is only 608x1080. That is about 32% of the original pixels. To fill a 1080x1920 vertical video, the editor has to stretch that slice by roughly 1.78x on each side, which means about 3 out of every 4 pixels in the final frame are interpolated, not recorded. Hair, text and skin texture are the first things to go soft.

The second loss is generational. A typical path looks like this: camera file, edited export at a modest bitrate, that export cropped and exported again in a phone app, then the platform recompresses it a third time. Every lossy encode throws away detail, and compression artifacts get baked in and then enlarged by the crop.

The third loss is the starting file. A video downloaded from YouTube or a cloud preview is already compressed for streaming. Cropping it and scaling it up enlarges those artifacts too.

The cleanest fix happens before editing. If you shoot in 4K and edit on a 1080p horizontal timeline, export the vertical clips from the 4K original, not from the 1080p edit. A 9:16 crop from a 3840x2160 frame gives you 1215x2160 pixels, which is more than the 1080x1920 you need. You can even punch in a bit (up to about 112% of a full height crop) and still not upscale.

If you only have 1080p, that is fine for talking heads and podcasts, just be aware the crop will be slightly soft and keep sharpening light. Heavy sharpening on an upscaled crop creates halos around faces and makes the platform compressor work harder.

This is the reason Clipzi renders vertical exports from the original file you uploaded, not from the compressed preview you see in the browser. Each clip is cut and encoded once, at 1080x1920 and a delivery bitrate around 12 Mbps.

Crop to fill is the default for talking video. It puts the face big on a phone screen, which is what holds attention. The cost is resolution (on 1080p sources) and losing whatever sits at the edges of the frame.

Fit with a blurred background keeps the whole horizontal frame visible at native resolution and fills the top and bottom with a blurred, zoomed copy of the same video. Use it for wide shots, gameplay, slides, screen recordings or anything where cutting the sides would remove information. The downside: the actual content takes up only about a third of the screen height, so faces look small.

A practical middle ground for podcasts: crop for the single speaker moments and switch to fit for the wide shots. Leave the top and bottom of the blurred area clear for captions and the platform interface. Our guide on Shorts, TikTok and Reels safe zones shows where the buttons sit.

A static center crop works until the speaker leans, turns to a guest or walks. Then you get half a face. Two ways to fix it:

Manual keyframes. Set a position keyframe on the clip, move the playhead to where the person shifts, reposition, and the editor animates between the two. For interviews, use hold or quick moves on cuts between speakers rather than slow drifts, which look like a shaky camera.

Auto reframe. In Premiere Pro, right click the sequence in the Project panel (or go to Sequence) and choose Auto Reframe Sequence. Pick Vertical 9:16 as the target aspect ratio and set Motion Tracking to Slower Motion for talking heads, so it does not chase every small movement. You can then turn on Overwrite generated path in Effect Controls and fix Reframe Offset by hand where it guessed wrong. CapCut and DaVinci Resolve Studio (Smart Reframe) have similar tools.

In Clipzi's AI video editor the reframe step has the same idea built in: choose a layout (fullscreen, centered crop, split screen, three person or spotlight) and add keyframes so the crop follows the speaker across the clip.

When two people talk on a wide shot, a single 9:16 crop forces you to pick one. Split screen stacks them: one speaker in the top half, the other in the bottom half, each panel 1080x960.

Here is the part most guides skip: split screen is often the sharpest option on a 1080p source. Each panel needs a 1080x960 area, and you can take a crop that size straight from a 1920x1080 frame at 100% scale. No upscaling, as long as each person fits in their crop. Compare that with a full 9:16 crop, which needs a 1.78x upscale.

Keep the eyes of each speaker roughly in the upper third of their panel and leave a small gap or line between panels so the cut reads clearly. If the two people sit far apart in the original, crop each one separately; if they are close, a 50/50 split of one region works. For podcast workflows, see how to turn a podcast into clips.

Your export should be the only lossy encode before the platform's own. Use these as a starting point:

One clip a week from a single camera: do it in your editor. A new 1080x1920 sequence, a keyframed crop and the export settings above take ten minutes and cost nothing.

Several clips from every long episode is where manual reframing turns into hours. That is the job Clipzi was built for: upload the long video, the AI transcribes it and suggests the best moments as clips, and you adjust the layout and keyframes before exporting. The Free plan gives you 2 long videos in total with a watermark, no credit card needed. If you prefer to see the trade offs with other tools, compare it with Opus Clip.