How to turn a long video into shorts.

Step 1, Pick the right moments

This is 80% of the job. A short doesn't fail because the export was wrong; it fails because the moment wasn't strong enough to begin with. A clip-worthy moment usually has three traits:

1. A hook. The first sentence makes a curious viewer stop scrolling. Most often this is a surprising claim, a sharp question, or a confident assertion.
2. A build. The middle delivers on the hook, explanation, story, or reasoning.
3. A payoff. The ending earns the watch with a resolution, a punchline, or a "huh, didn't know that" moment.

If a moment has only two of the three, leave it out. Don't try to manufacture the third, viewers will feel the gap.

Step 2, Trim to 30–60 seconds

YouTube Shorts cap at 3 minutes, but the algorithm rewards completion. Aim for 30–60 seconds for most clips. Start on the hook (no preamble), end on the payoff (no "and that's all for today"). If your clip needs context, write a one-line caption that gives it instead of using video time.

Step 3, Reframe to 9:16

Your source video is probably 16:9. Vertical platforms are 9:16. You can't just letterbox, you need to crop to the speaker, keeping their face in the upper-middle third of the frame (the lower third is covered by platform UI).

If you move while talking, you need speaker tracking, the crop should follow you. Doing this by hand means animating a position keyframe every few seconds. Tools that auto-track make this trivial. Avoid static crops if you ever leave the center of the frame.

Step 4, Burn captions in

Native platform captions get muted, mistranscribed, or simply skipped by viewers. Burned-in captions, meaning the text is part of the video pixels, survive every platform's compression and can't be turned off. They also let you control style: word-by-word reveals, keyword highlighting, brand fonts.

Two rules: never let captions cover the speaker's face, and never let them cover the platform UI overlay (bottom 25% on most apps).

Step 5, Export per platform

• YouTube Shorts: 1080×1920, H.264, < 3 minutes.
• TikTok: 1080×1920, H.264, 15–60s ideal.
• Instagram Reels: 1080×1920, H.264, < 90 seconds for max reach.

Same source, three exports. Yes, you could just upload the YouTube Shorts version to all three, but Instagram detects watermarks and deprioritizes anything that looks like a TikTok crosspost.

Manual workflow time cost

A 30-minute long video into 8 shorts, done manually:

• Watching back and marking moments: ~45 minutes
• Cutting clips and trimming hook/payoff: ~60 minutes
• Reframing with keyframed crops: ~90 minutes
• Captioning each clip: ~90 minutes (or pay for auto-captions and clean them up)
• Exporting all three platform versions: ~30 minutes

Total: 5 hours, roughly. That's why most creators give up after the first three videos.

AI workflow

The right AI shorts generator collapses all five steps into "upload, review, export." AutoCuts picks moments using a model trained on hook structure, trims to hook/payoff, tracks the speaker for 9:16 reframe, burns in captions in your chosen style, and exports per platform. The work that's left is review, accept or reject each clip, which usually takes about 30 minutes for a long video.

Other tools can help with the clipping piece after a video is already edited or published. The difference is scope: AutoCuts starts from the raw upload and gives you the polished long-form cut, the shorts, the captions, and the motion graphics in one workflow.

What about ranking clips by virality?

Some tools attach a "viral score" to each clip. Take these with salt, they're trained on past virality, which doesn't generalize cleanly to your channel and your niche. They're useful as a ranking signal ("look at the top 5 first") but not as a publish/don't-publish decision.