Record the video, describe the edit in plain English, or arrange it yourself on a timeline. Every operation runs on your own machine, so there is no upload, no queue, and no watermark.
Supercut records your screen, microphone, and camera and hands the result straight to the editor, already seated on a backdrop with the corners and shadow applied. It is built for demos, walkthroughs, and the update you would rather not hold a meeting for. Start in the browser with nothing to install, or use the desktop app for computer audio and zooms that follow your cursor.
Screen, microphone, and camera are captured separately, and the camera is never composited into the screen. That is what lets you move, resize, restyle, or delete the bubble after the take instead of committing to a corner before you press record.
On the macOS desktop app the recorder samples your cursor and records your clicks, so the push-ins land on the things you actually clicked. Each one arrives as an editable item you can retime or remove. In the browser there is no way to watch the cursor outside the page, so you place the zooms yourself.
Every scene is CSS and SVG, one hue family per scene, with a real shadow seating the window. There is no asset to fetch, which is why the preview, a browser export, and a packaged desktop export all paint the same frame.
A short count-in runs once you have picked a source, so you get a moment to settle. Pause and resume as often as you need: the paused time is folded out of the final length, and out of the cursor and click timestamps with it.
The take drops into Studio inset on its backdrop with the padding, rounded corners, and shadow already applied. Trim it, caption it, then export a 4K mp4 or a GIF.
Capture, edit, and export all run locally. There is no server copy, no generated share link, and no watermark, which is what makes it usable when the screen has an unreleased build or a customer account on it.
| Capability | macOS desktop app | Windows desktop app | Browser |
|---|---|---|---|
| Computer audio | Yes, the whole machine | Yes, through loopback | Audio from a tab you share |
| Cursor tracking | Sampled 60 times a second, plus clicks | Sampled 60 times a second, no clicks | None |
| Automatic zooms | Anchored to where you clicked | You place them | You place them |
| The pointer in the picture | Redrawn, when you record a display | Baked into the capture | Baked into the capture |
| What you need | The desktop app | The desktop app | Chrome, Edge, Brave, or Arc |
When a prompt is not the right shape for the job, open the Editor tab. Studio is a proper timeline: layered video, audio, text, and captions, with trimming, splitting, and an inspector for each item. It is where a recording lands, where a picture-in-picture gets positioned, and where several clips become one piece.
Video, audio, text, and captions each sit on their own track. Drag to move, drag an edge to trim, split at the playhead. Clips carry a filmstrip and an audio waveform so you can see the thing you are cutting.
Transcribe a clip on your device, then remove the words you do not want. The cut is applied to the timeline, which is the fastest way to strip filler out of a demo or an interview.
Select any item and set its position, scale, crop, timing, and text styling. Cropping reshapes the frame the item is laid out in, so a cropped clip sits correctly rather than being stretched back to its original aspect.
Describe an edit and Supercut runs it against a scratch copy first, then shows you exactly what it would do. Nothing touches your timeline until you accept it, and undo covers the rest.
Save a .unocut project file with the assets inside it, so a whole edit moves between machines as a single file rather than a folder you have to keep together.
Preview, browser export, and desktop export all render the same composition, so what you watched is what you get. Browser export needs a Chromium browser. The desktop app renders faster and is steadier at 4K.
Type what you want and Supercut works out the operation, then runs it locally. Only the text of your request is ever sent anywhere. Each card below is a real prompt you can paste into the editor.
Cut to a specific time range
“Trim from 0:00 to 0:30”
Vertical aspect for TikTok / Reels
“Crop to 9:16 for TikTok”
Square aspect for Instagram feed
“Crop to 1:1 square”
Widescreen aspect
“Crop to 16:9 widescreen”
0.25x to 4x speed control
“Speed up 2x”
Frame-interpolated slow-mo
“Smooth slow motion 0.5x”
Reduce camera shake
“Stabilize shaky footage”
360p to 4K
“Downscale to 720p”
Reverse playback
“Play the video backwards”
Rotate or flip
“Rotate 90 degrees”
Generate subtitles with Whisper
“Add captions”
Auto-detect and blur faces
“Blur faces for privacy”
Follow faces when cropping
“Smart crop to 9:16”
Strip audio from video
“Remove audio”
Get audio only
“Extract audio as MP3”
Make speech clearer
“Enhance voice clarity”
Balance volume
“Normalize audio levels”
Clean up audio
“Reduce background noise”
Adjust loudness
“Increase volume by 6dB”
Film-like color grade
“Add cinematic effect”
Retro tape look
“Apply VHS effect”
Golden hour tones
“Add warm filter”
Blue undertones
“Add cool filter”
Grayscale
“Make it black and white”
Boost saturation
“Make colors vibrant”
Smooth transitions
“Add 1s fade in and out”
Forward then reverse loop
“Make it a boomerang”
9:16, optimal settings
“Optimize for TikTok”
Reel-ready format
“Export for Instagram Reels”
Shorts-optimized
“Export for YouTube Shorts”
X video specs
“Export for Twitter”
Under 8 MB
“Compress for Discord”
Animated GIF output
“Convert to GIF”
Text overlay
“Add watermark 'Your Text'”
Reduce file size
“Compress the video”
Change format
“Convert to WebM”
Take the audio from one clip, put it on another (replace @clip with a clip name)
“Use audio from @clip”
Join two clips end-to-end (replace @clip with a clip name)
“Stitch @clip onto the end of this one”
Inset a second clip in a corner (replace @clip with a clip name)
“Overlay @clip as picture-in-picture”
Deep-dive guides for high-intent searches, each links straight into the editor.
Supercut decodes, edits, and encodes on your device with WebAssembly, and the desktop app does the same with a native build. Your footage is never uploaded, which is the whole reason recording an internal dashboard or an unreleased build is fine here.