Long video to Shorts, with AI doing the watching
Upload a recording or paste a YouTube link. The AI reads the whole thing, picks the moments that stand on their own, and hands back vertical clips with captions and a speaker-following crop
Free plan: about 15 clips a month (300 credits) · no card needed
How it works
- 01
Paste a link or upload
A YouTube URL or your own file: podcasts, interviews, streams, talks, matches. Longer is fine, every minute gets scored
- 02
The clipper finds the moments
It transcribes, reads faces, voices, reactions and audio peaks, then ranks candidate clips for hook, payoff, pacing and whether they stand alone
- 03
Review, tweak, post
Each clip opens as an editable project in the Studio with captions and crop already in place. Export or schedule straight away
What you get
Reads the whole video
Word-level transcription plus faces, voices, reactions and audio peaks, so no minute is skipped
Scored, then double-checked
Each candidate is scored for hook, payoff, pacing and standalone sense, then a second critic pass tightens or drops it
9:16 that keeps people in frame
The crop follows the active speaker, switches to split tiles for two people and tracks the action in sports
Captions on by default
Word-level animated captions inside the platform safe area, in a theme you can swap in one click
Editable clips
Each Short is a project on a browser timeline: trim, re-crop, restyle, then export MP4
What "video to Shorts" actually involves
Turning a long video into Shorts is four jobs that usually get done by hand: finding the forty seconds worth posting, cutting them cleanly, reframing a 16:9 shot to 9:16, and adding captions people can read with the sound off. Done manually, one good Short from an hour-long recording can take most of an hour. The point of an AI tool is to do the watching and the mechanical work, and leave you with the decision of which clips to post
shitShorts does those four jobs in one run. You upload the file or paste a YouTube link, choose how many clips you want and roughly how long, and get back a ranked set of vertical clips with captions and a crop that follows the speaker. Each one opens as an editable project, so the AI's choices are a starting point rather than a final render
How the AI decides what is worth a clip
The audio is transcribed with Whisper at word level, so the system knows exactly when every word starts and ends. The transcript is split into overlapping windows, and a language model scores each window for four things: the hook (would the opening line stop a scroll), the payoff (does it deliver something by the end), pacing (is there dead air or rambling) and whether it stands alone without the conversation around it
Scores alone pick too many near-misses, so a second model acts as a critic. It reviews the top candidates, moves their start and end points to cleaner lines, and rejects clips that only make sense with context. Signals from the picture feed in too: laughter and audio peaks, visible reactions, scene changes and how many faces are on screen. Every clip comes with the reasons it was chosen
Keeping the subject in a vertical frame
A long video is almost always shot wide. Cropping the centre of a 16:9 frame to 9:16 throws away two thirds of the picture, and whoever was not in the middle disappears. The clipper tracks faces through the clip and matches voices to faces, so the vertical frame sits on whoever is talking and moves when the speaker changes. When two people share a moment it switches to stacked split tiles. On sports footage it tracks the ball and players instead of faces
The crop is a keyframed layer in the editor, so if the AI framed a shot differently from how you would, you can nudge it or hold it in place without re-running anything
Which long videos work best
Talking footage gives the best results: podcasts, interviews, panels, webinars, lectures, livestreams with a facecam, and vlogs. Sports and action footage work well too, thanks to object tracking. Screen recordings without a face, slide-only presentations and music videos give the clipper less to work with; it can still find moments from the transcript, but you may want to adjust the crop by hand
Length is not a problem. Every minute is scored, and you see the credit estimate before anything runs. If you own the original file, upload it rather than pasting a link: you get a cleaner picture and avoid the quality loss of a re-encoded stream
From clips to posts
Clips arrive as projects on a browser timeline with the captions, hook text and crop as separate layers. Fix a misheard name, change the caption theme, trim a second off the end or add your logo, then export an MP4. Captions sit inside the safe area of Shorts, Reels and TikTok so the platform buttons do not cover them
If you post regularly, the same clips can go straight onto the scheduling calendar for YouTube Shorts, TikTok, Instagram Reels, Facebook Reels, LinkedIn, X, Threads and Pinterest, each with its own caption and hashtags
What it costs
The free plan includes 300 credits a month. A clip costs about 20 credits, so that is roughly 15 clips, exported at 720p with a small watermark. Your first 3 clip exports are free, and after that each clip export on the free plan costs 10 credits. The Clipper plan is $10 a month for 1,000 credits, about 50 clips, at 1080p with no watermark and 300 clip exports a month. Creator is $19 and Pro is $49 if you also want the rest of the studio at higher volumes. Editing, previews and captions never cost credits
Only clip videos you own or have permission to use
Compare AI clippers
- shitShorts vs OpusClip
- shitShorts vs Submagic
- shitShorts vs Captions
- shitShorts vs ChopCast
- shitShorts vs Klap
- shitShorts vs Munch
- shitShorts vs 2short.ai
- shitShorts vs OpenShorts
- shitShorts vs Vizard
- shitShorts vs Riverside Magic Clips
- shitShorts vs Descript
- shitShorts vs CapCut
- shitShorts vs ClipPod
- shitShorts vs LunarClip
- shitShorts vs SendShort
- shitShorts vs Quso
- shitShorts vs StoryShort
Questions people ask
Updated October 2026