Hook in under two seconds
Cut on the first strong word, not the breath before it. If the spoken hook is slow, add a 3-word title card over the opening. See short video hooks.
CapCut is the editor most clippers already have. This guide covers both ways to cut shorts in it: the manual split-and-trim workflow where you control every frame, and the AI route that turns a long video into a batch of captioned 9:16 clips. Plus the export settings and habits that get clips approved on campaigns.
Manual clipping in CapCut means you scrub the timeline, split at the moments you want, delete the rest, set the ratio to 9:16, add captions and export. You decide the hook, the cut point and the framing, which is why it still wins for high-stakes clips.
AI-assisted clipping uses CapCut's long-video-to-shorts clip maker: upload the source, choose clip lengths, and it detects highlights, reframes to 9:16 with subject tracking, adds subtitles and suggests titles. It's fast and good at finding candidates, less good at knowing which one will actually land.
Works the same on desktop and mobile; the desktop keyboard shortcuts make it much faster.
New project → Import the source → drag it to the timeline. Before you cut anything, set the ratio to 9:16 (the aspect button under the preview). Doing this first means every cut you make is judged in the vertical frame.
Scrub to the segment you want. Then find the exact frame where the strongest sentence starts: that's your in-point. Shorts live or die in the first two seconds, so the cut should land on the hook, not on a breath before it.
Put the playhead on the in-point and press Split (Ctrl+B / ⌘B, or the scissors icon). Do the same at the out-point. Select the parts before and after and delete them. Aim for 20–45 seconds; trim the end so it finishes on a complete thought.
Scale the clip up and drag it so the speaker's face sits in the top third of the vertical frame, leaving room for captions below. For two speakers, split the clip where the speaker changes and reposition each piece, or use a split-screen layout.
Text → Auto captions → pick the language and generate. Choose a bold preset, enlarge it, and check the transcript for names and product terms the recogniser missed. Add a 3-word title card over the first two seconds if the hook is visual rather than spoken.
Export at 1080p, 30 fps (60 for gaming), MP4/H.264, bitrate on "Recommended" or higher. Name the file with the source and timestamp so you can find the moment again if a campaign asks for a re-cut.
The long-video-to-shorts tool lives in CapCut Online and the desktop app's AI tools. It takes sources up to 3 hours or 10 GB.
Open the long-video-to-shorts tool, then upload a file or import from Google Drive or Dropbox. Keep the tab open while it uploads and analyses; a two-hour podcast takes a few minutes.
Pick a duration band (under 30s, 30–60s, 60–90s) and, optionally, limit analysis to a time range so it skips the intro and outro. For campaigns, match the band to the platform's rules; most reward 30–60s.
The clip maker detects highlights, cuts them, reframes to 9:16 with subject focus, adds subtitles and drafts a title for each. Review the batch in order; the top few are usually usable, the tail is where you'll find mid-sentence starts and dead air.
Open a candidate with "Edit more". Tighten the in-point to the hook, check the caption transcript, swap the caption preset to your brand style, and nudge the reframe if the face drifts. This is the part the AI gets roughly right and you get exactly right.
Export each approved clip at 1080p, or share straight to TikTok, Instagram or YouTube from the tool. For campaigns, export to a folder named by episode so you can log which clips went where and what they earned.
| Manual | AI clip maker | |
|---|---|---|
| Time per clip | 5–10 min | Seconds, in batch |
| Cut precision | Frame-exact | Roughly right; fix by hand |
| Finds the moments | You do | Highlight detection |
| Reframing | Manual scale and position | Auto 9:16, subject focus |
| Captions | Auto captions, any style | Auto, 20+ languages |
| Titles / descriptions | Write your own | Drafted for you |
| Source limit | Whatever your machine handles | 3 h / 10 GB |
| Cost | Free | Free tier; some AI features need Pro |
| Best for | Hero clips, sponsor reads | Campaign volume, first pass |
Cut on the first strong word, not the breath before it. If the spoken hook is slow, add a 3-word title card over the opening. See short video hooks.
Save one caption preset and reuse it. Bold, high-contrast, two lines max, sitting in the lower third but above the platform UI.
Leave the bottom third for captions and the platform's buttons. For two speakers, split at the speaker change rather than cropping both into a tiny frame.
Trim the end so the clip resolves. A sentence that trails off costs completion rate, and completion rate is what the algorithm pays for.
1080p MP4 at 30 fps is the safe default. Name files by episode and timestamp, and track views per clip so the earnings calculator has real averages to work from.
Some campaigns reject obviously auto-generated clips or require a manual touch. AI-find-then-hand-finish satisfies most of them. See clipping campaigns.
CapCut itself is free, and the long-video-to-shorts tool has a free tier. Some AI features, higher export limits and watermark-free output on certain effects need CapCut Pro, and Pro pricing varies by region and platform. Check the pricing page in your app.
The clip maker takes uploads and cloud imports rather than public video URLs in most regions. Download the source first (with permission), or use a YouTube video clipper that accepts links, then finish in CapCut.
No. It's an editor with AI features: it won't watch a channel, run on a brief or publish on a schedule by itself. For that, see best AI clipping agents. CapCut is where many clippers finish what an agent produces.
Desktop for manual clipping (keyboard shortcuts, bigger timeline). CapCut Online for the AI clip maker with large sources. Mobile for quick fixes and posting. Projects don't always sync across all three, so pick one per clip.
In the AI clip maker, reframing to 9:16 tracks the subject for you. In manual mode, scale and position by hand, or split the clip at each speaker change and reposition each part.