Your evening back
One tap of Auto-Polish cuts the silence and the filler, captions it, cleans the voice, adds the zooms and scores it. You start from a finished edit and adjust, instead of starting from a raw take and grinding.
FEATURES
Record it. One tap finishes it. Nothing leaves your device.
The work you actually want back is the cutting, the captioning and the cleanup. Zella does that pass for you the moment a recording opens, on the chip already in your hands. No render queue, no account, no upload, no watermark, and no subscription waiting to lapse. Free is the complete editor and all the AI on every platform.
Zella for Mac is coming to the Mac App Store, and Zella for iPhone is coming to the App Store. One Pro purchase covers both. See the mobile app →
Quick answer
Zella is a screen recorder and AI video editor for Mac and iPhone. Zella records your screen or camera, then one tap of Auto-Polish cuts the silences and filler words, writes on-device captions, cleans the voice and adds emphasis zooms. Background removal works with no green screen, and reframe to 9:16, 1:1 or 16:9 is one click. Everything runs on your own device, with no cloud, no account and no watermark on any tier.
WHAT ACTUALLY CHANGES
Four things stop being your problem.
Nothing to upload, nothing to wait for
Recording, transcription, background removal, grading and export all run on the chip in front of you. No render queue, no account, no upload. Footage under NDA stays where it is legally supposed to stay.
One cut, every aspect ratio
Face-aware reframe puts you in the middle of 9:16 instead of centre-cropping your head off. You stop re-editing the same video three times to feed three platforms.
No watermark, no subscription, no trap
Free exports are clean, permanently. Free is the complete editor and all the AI on every platform, and Pro is one payment of $89 that covers Mac and iPhone together.
THE PARTS PEOPLE SWITCH FOR
Six reasons this is not another editor.
Each one leads with what it does for you, then shows its receipts. If a claim here has no specific under it, do not believe it.
One tap between a raw take and a finished edit.
This is the feature the whole app is built around. It is not a filter and it is not a preset: it is the full retention edit, run on-device, before you have made a single decision.
- Silence and filler cuts, from the on-device transcript
- A steady cut cadence with varied transitions and a matched sound effect under every cut
- Beast-style keyword captions with emoji
- Noise removal and voice polish
- A hook title with an opening punch-in
- Emphasis zooms on the lines that carry the point
- No colour grade - your footage keeps the look you shot it with, and grading stays a separate control you reach for on purpose
- Mood-matched music with an arc, ducking under your voice
- Pattern interrupts, so the middle does not sag
Both platformsRuns on Mac and iPhone, and so does Make it Viral: the whole ready-to-post pass in one tap, on top of the shape you already cut.
How Auto-Polish works →Your footage never leaves the room.
Not "encrypted in transit". Not "deleted after 30 days". There is no upload step at all, which is the one thing a cloud editor cannot copy, because the cloud is their product.
- Recording, transcription, background removal, grading and export all run locally
- No account to create and no sign-in to sit through
- Works in airplane mode
- Client footage under NDA, staging data, student work and pre-launch product never move
- The only two things that can touch the network are optional and opt-in: the Assistant sends your typed prompt on your own key, and stock B-Roll needs a free key you add yourself
Both platformsTrue on Mac and iPhone equally. On iPhone, Auto B-Roll and caption translation have no network path at all.
Mac and iPhone, side by side →A clean background with no green screen, live while you film.
You see the final look in the viewfinder, not an hour later in the editor. That is the difference between shooting one take and shooting four.
- On-device AI cutout, no green screen and no plugin
- 40 built-in scenes (20 still, 20 animated), or blur, a solid colour, or your own image or video
- Subject staging with a soft backlight and a contact shadow so you look placed, not pasted
- Defringe cleans the colour halo off the cutout edge
- Two on-device vision models must agree on every frame on iPhone, so gestures do not drag the room with them
Both platformsOn the Mac it runs on the webcam bubble and the main video. On iPhone it is baked in live on the Neural Engine while you record, and again in the editor.
Camera and backgrounds →Captions on-device, and nineteen languages without a network.
Captions are how muted feeds and non-native viewers get your video at all. Doing them on the device means no transcript service, no per-minute bill, and no student or client audio sitting on someone else’s server.
- 10 caption presets on Mac, 9 on iPhone, with active-word highlight
- Verbatim WhisperKit transcription on iPhone, which is why "um" and "uh" actually get cut there
- On-device caption translation into 19 languages on iPhone (iOS 18 and later), with no network path at all
- SRT import, a transcript word editor, and custom font import for any language
- Text Behind Subject, using the same on-device subject cutout
Both platformsSpeech recognition runs on-device on both. Translation and verbatim filler removal are genuinely iPhone features today.
Captions →Free on every tier. On both platforms. Permanently.
Type the edit instead of learning where it lives. We gate output, never intelligence, so neither of these has ever been a Pro feature and neither will be.
- 101 editing actions on Mac, 71 on iPhone
- Content-aware cuts: "cut the part where I fumble the intro" finds it in the transcript
- It can see the frame, so "is my face centred?" and "is it too dark?" are answerable
- Auto B-Roll on iPhone generates cutaway cards on the device with no API key and no network
- Optional Pexels or Pixabay stock footage if you add your own free key
- Bring your own LLM key: Groq, Gemini, Cerebras and Ollama are free, and keys stay in the Keychain
Both platformsFree on Mac and iPhone. Your typed prompt and a short project summary (including caption text) are sent, and with Ollama not even that leaves the machine.
AI Assistant →Reframe once, ship to every platform.
The version of this job that costs an hour is re-cropping and re-exporting per platform. The version that costs a click is one edit and a queue.
- One-click 9:16, 1:1 and 16:9 with Vision face-detection auto-track, or manual offset
- Make Vertical on iPhone reframes to 9:16 around your face in a single tap
- Platform presets: TikTok, Reels and Shorts at 1080x1920, YouTube at 1920x1080, Instagram at 1080x1350
- MP4, MOV, GIF and M4A, with captions and zooms baked into the GIF too
- Pro adds unlimited length, 4K (8K on Mac), transparent-background export, 60 fps, a custom bitrate and the multi-aspect queue
Both platformsExport renders on the Mac or the iPhone itself. Or hand the project across as a .zella file and finish on the bigger screen.
Reframe →CAPTURE ANYTHING
Never re-shoot because something interrupted you.
Press record and stop managing the recording. Takes are written in crash-safe segments, so a phone call, a closed lid or a flat battery costs you nothing. Your room is dealt with while you film rather than in a fix-it pass afterwards, and system audio and your mic arrive as separate things you can still balance later.
On Mac
Capture the whole screen, a region you drag, one single window, or camera-only, with system audio and mic together. Cursor and click tracking runs underneath, which is what lets the editor zoom on real clicks later. A menu-bar recorder and the full keyboard set mean starting a take never interrupts what you were doing.
On iPhone
Front or back camera with a live background you can blur, cut out or replace while you film, screen recording and whole-device broadcast via ReplayKit, and import from Photos or Files. Takes are segmented and crash-safe, so a phone call never costs you the recording.
THE PROOF
- Full screen / region / window / camera-only capture on Mac
- Camera, screen and whole-device broadcast capture on iPhone
- Webcam bubble overlay with blur, AI background removal (RVM), solid or live backgrounds
- System audio + mic, with pause / resume mid-take
- Crash recovery and sleep-safe long recordings
- Runs native on Apple Silicon + Intel Macs, and on the Neural Engine on iPhone
ONE CLICK, READY TO POST
The edit you were dreading is already done when the file opens.
A twenty minute take normally costs an evening: cutting the dead air, hunting the ums, writing captions, fixing the audio, adding the movement that keeps a static frame alive. Auto-Polish does that pass on-device before you touch anything, and you edit its result instead of starting from a raw wall of footage.
On Mac
Auto-Polish runs the moment a recording opens. Make it Viral is the whole ready-to-post pass in one click: tightened pacing, transitions, matched sound effects, word-pop captions and a ducked music bed, on top of the shape you already cut. Auto-zoom follows the clicks your cursor actually made, because the Mac tracked them at capture.
On iPhone
Auto-Polish runs the same one-tap retention edit on the phone - silence and filler cuts, extra cuts where you naturally pause, a cross-dissolve on every join, Beast keyword captions, noise removal and voice polish, a hook title, emphasis zooms and a ducked music bed. Make it Viral is here too, adding the colour pop and the matched sound effects on top. Clean Up, Make Vertical and Reel Pack are one tap each, and deleting a word in the transcript ripple-cuts the video.
THE PROOF
- Auto-Polish - captions, zoom-ins and voice polish, auto-run on every new recording (it leaves your colour alone)
- Make it Viral - the whole ready-to-post pass in one tap: dead-air trim, punch-in zooms, transitions, SFX and a ducked music bed (keeps your original shape)
- Viral templates - Reel Pack, Hook Captions, Beat Drop, Cinematic Reel, Photo Cards, Reaction Split
- Auto-zoom that follows real cursor clicks - the Screen-Studio look
- Remove silences and filler words (um / uh / like), reviewed before you apply
- On-device auto-captions + neural noise removal - no upload, no transcript service
EDIT BY TYPING
You do not have to know where the setting lives.
Every editor punishes you for not knowing its vocabulary. Type what you want instead and the Assistant runs the whole chain, finds the fumbled line in the transcript, or looks at the frame and tells you your face is off-centre. It is free on every tier on both platforms, because charging for the intelligence is the thing we refuse to do.
On Mac
101 editing actions driven by chat, from cutting silence to grading, reframing, adding callouts and setting up the export. Bring your own key and it stays in your Keychain; Groq, Gemini, Cerebras and Ollama are free, and Ollama never leaves the machine at all. Only your typed prompt and a short summary of the project (clip counts, durations, caption text, effect settings) are sent. Your video and audio never are.
On iPhone
The Assistant ships on iPhone with 71 editing actions - cut, caption, grade, reframe, set clip and overlay volume, add zooms, change platform. It runs on your own API key, and only your typed prompt and a short summary of the project - including caption text - ever leave the device; your video never does.
THE PROOF
- Plain-English editing - "remove the dead air, add captions, and make it vertical" runs the whole chain for you
- Content-aware cuts - "cut the part where I fumble the intro" finds it in the transcript and removes it
- It can see your video - ask "is my face centered?" or "is it too dark?" and it checks the frame on-device
- Auto B-Roll - "add b-roll" places matching stock clips from your captions automatically (free Pexels key)
- Adds music, built-in sound effects (43 on Mac, 30 on iPhone), titles, callouts and logos on command
- 101 editing actions on Mac, 71 on iPhone - the Assistant is free on both, it is never a Pro feature
- Bring your own key (Groq, Gemini, Cerebras and Ollama are free) - keys stay in your device’s Keychain, nothing is uploaded by Zella
PRO EDITOR, ZERO LEARNING CURVE
Simple enough on day one. Still there on the day you outgrow simple.
Most editors make you choose between a toy and a flight deck. Zella opens as a timeline you can use immediately, and an Advanced Tools switch reveals keyframes, bezier curves, luma scopes, Color Match and masks only when you go looking for them. Nothing you learn gets thrown away, and nothing you have not learned is in your face.
On Mac
Keyboard-driven: B to cut, the full shortcut set, ripple delete, drag-trim, drag-to-reorder and 50 levels of undo. The Library keeps every recording and project in one place instead of a Downloads folder. Autosave and versioned snapshots mean a crash costs you seconds.
On iPhone
A real multi-clip touch timeline: cut, trim, ripple-delete, drag to reorder, per-clip volume, and an Audio tool holding every level in one place. Picture-in-picture overlays resize free-form (independent width and height), zoom blocks and speed control are free, and keyframe curves, masks and speed ramps are the Pro depth toolset - the same law as the Mac.
THE PROOF
- Timeline with ripple delete, drag-trim, drag-to-reorder and 50-level undo - B-key cut and the full keyboard set on Mac
- Named for beginners on both apps - tabs say Text and Arrows, the toggle says "Music gets quieter when you talk", one-time tips teach the two gestures nobody finds, and every delete answers "tap Undo to bring it back"
- Advanced tools switch - one toggle reveals motion keyframes, the bezier curve editor, live luma scopes, Color Match and effect masks
- Speed control - slow / fast / freeze / reverse free; ramps, blade-speed and instant replay with Pro
- J / L cuts for natural-sounding edits
- Per-clip volume, autosave with crash recovery, versioned project snapshots
EFFECTS & COLOR
Footage that looks graded, without a colourist or a plugin bill.
Raw camera footage looks like raw camera footage, and that is the difference a viewer reads as amateur before you have said a word. Pick a look, pull the intensity back until it stops shouting, and the whole video sits in one world. Auto-Polish deliberately leaves your colour alone, so a grade is always your decision.
On Mac
38 built-in looks across Cinematic, Film, Modern, Special, Studio and Vintage, three of them Founder-exclusive, plus your own .cube LUTs. Live luma scopes and Color Match sit behind the Advanced Tools switch for the day two shots refuse to match. Blur callouts and feathered masks keep keys, faces and staging data out of frame.
On iPhone
All 42 colour looks with an intensity slider, the full colour board, .cube LUT import and the whole effects pack, on-device - including the 2026 creator pack (Soft Glow, Bright Airy, Peachy, Terra Brown, Retro Halation, Tokyo Soft, Punchy), every one tuned to lift rather than shade. The three Founder cinematic grades ship here too, taking the phone catalogue to 45, and Auto-Polish deliberately leaves your colour alone.
THE PROOF
- Colour grades across Studio, Cinematic, Film, Modern, Special and Vintage - 43 on Mac and 42 on iPhone, plus 3 Founder-exclusive grades (46 and 45 in total) + intensity slider
- Every bright look brightens - zero baked vignettes, and an automated corner test on a portrait frame keeps it that way on every build; only five deliberately dark looks shade at all
- The 2026 creator pack on both platforms - Soft Glow, Bright Airy, Peachy, Terra Brown, Retro Halation, Tokyo Soft, Punchy - the grades creators actually run, every one tuned to lift
- AI background removal with 40 built-in scenes (20 still, 20 animated) or your own image / video
- Subject staging - position, scale, a soft backlight glow and a contact shadow that grounds the subject on its new background
- Blur callouts and feathered rectangle / ellipse masks to censor faces, keys and secrets
- Vignette, glow, grain, pixelate, chroma, letterbox + exposure, contrast, saturation, temperature and tint
BACKGROUND REMOVE & REPLACE
Your room stops being part of the video.
No green screen, no lighting rig, no corner of the house you keep tidy for filming. The cutout runs on the device while you record, so you see the finished look in the viewfinder rather than discovering it failed an hour later. Two on-device vision models have to agree on every frame, which is why gestures do not drag the wall with them.
On Mac
Background removal runs on the webcam bubble and on the main video, with 40 built-in scenes (20 still, 20 animated) or your own image or video behind you. Subject staging adds position, scale, a soft backlight and a contact shadow so you look placed rather than pasted, and defringe strips the colour halo off the edge. With Pro, the same cutout exports as ProRes 4444 with a real alpha channel.
On iPhone
Background removal runs live on the Neural Engine - while you film and again in the editor, with no green screen. Two on-device vision models must agree on every frame and the cutout edge tightens where you actually move, so gestures don't drag the room with them. Removal lands on a soft bokeh backdrop by default, switchable to cutout, transparent, blur or any of the 40 backdrops.
THE PROOF
- AI background removal (RVM) runs fully on-device - no green screen, no cloud, nothing uploaded
- 40 built-in scenes (20 still, 20 animated), plus blur, solid colours, or your own image / video
- Subject staging - position, scale, a soft backlight glow and a contact shadow that grounds you on the new scene
- Works on the webcam bubble and the main video, with a live preview while you record
- Defringe cleans colour halos off the cutout so edges stay crisp on solid backgrounds
CAPTIONS, TITLES & CALLOUTS THAT SELL
The people watching on mute still get the whole thing.
Most feed video is watched with the sound off, and a video without captions simply does not land there. Zella transcribes on the device, styles the captions so they read as designed rather than auto-generated, and gives you twenty ways (nineteen on iPhone) to point at the exact thing you are talking about instead of saying "up here in the corner".
On Mac
10 caption presets with active-word highlight, 20 animated title styles, Text Behind Subject using the on-device subject cutout, 20 callout types and 16 blend modes. Speech recognition runs on the Mac itself, so nothing is uploaded to a transcript service and nothing is queued behind anyone else.
On iPhone
9 viral caption presets with active-word highlight, SRT import and a transcript word editor. Transcribes 27 spoken languages, 18 of them verified end to end, or captions any of them in English with one labelled switch. Separately, a finished caption track can be rewritten into 19 languages via Apple Translation. Plus Text Behind Subject, 12 title templates with 16 kinetic entrances, full-screen intro and outro cards, 19 arrow & label types (callouts) and 16 blend modes. Title fonts are a 51-face grid you judge by eye, five viral faces bundled, with the Lyric look chipping one line.
THE PROOF
- Caption presets - Word Pop, Hormozi, Karaoke, Pop Box, Neon Glow, Clean Minimal, Beast, Highlighter, Bounce, plus Hero on the Mac (10 on Mac, 9 on iPhone)
- Active-word highlight; custom font import for any language
- 20 animated title styles with gradient fills, stroke, shadow and box
- Title fonts you judge by eye - a 51-face grid on iPhone with five bundled viral faces (Fjalla One, Archivo Black, Bangers, Luckiest Guy, Marker Pro), favourites a long-press away
- Per-line and per-word styling on Mac - select the words and restyle just them; the Lyric look on iPhone chips one line and keeps your line breaks
- Text Behind Subject - titles that sit between your background and you, using the on-device subject cutout
- 16 overlay blend modes for titles and callouts - screen, multiply, overlay, soft light and more
- Anchored, responsive titles; canvas drag-to-edit for every callout
- 20 callout types on Mac and 19 on iPhone - arrow, box, ellipse, label, keystroke, spotlight, number, blur, leader line, popup, magnifier, highlighter, underline, pixelate-censor, emoji, sticky note, scribble, timer, glitch region, plus the Mac-only corner-bracket Target frame (the Arrows tab on Mac, Arrows & Labels on iPhone)
SOUND THAT SELLS
It sounds like a studio, recorded at your desk.
Thin, roomy, uneven audio is what makes a good video feel cheap, and it is the part most people have no idea how to fix. Polish Voice denoises, de-esses, compresses and normalises in one pass, music ducks under your voice on its own, and there is a sound effect ready for every cut instead of a folder you have to go and find.
On Mac
43 built-in sound effects, 16 voice changers, music with Auto-Duck, and Polish Voice with five studio tones on one broadcast-clean foundation. Beat Sync puts cuts and punches on the kick, snare or hats, per band. J and L cuts, detach and reattach audio, and per-clip volume are all free.
On iPhone
43 sound effects, 16 voice changers, music with Auto-Duck and five studio voice tones - all behind one Audio button in the timeline, together with clip and overlay levels. Per-band Beat Sync is there when you want it, and never applied for you.
THE PROOF
- 43 built-in sound effects on the Mac, 30 on iPhone - whooshes, pops, risers, impacts and UI clicks, no files to hunt down
- 16 voice changers (chipmunk, robot, echo, reverb, megaphone and more)
- Music with Auto-Duck - the track dips under your voice, then swells back automatically
- Beat Sync pulses zooms and effects to the kick, snare or hats, tuned per band
- Polish Voice with five studio tones - Natural, Warm, Bright, Podcast and Cinematic voicings on one broadcast-clean foundation
- Detach / reattach audio and J/L-cut timing
SHIP TO EVERY PLATFORM
Cut it once. Every platform gets the version it wants.
Re-cropping the same video for TikTok, then the feed, then YouTube is the single most wasteful hour in a creator week. Reframe follows your face with Vision instead of centre-cropping your head off, platform presets are already the right pixels, and the export renders on hardware you own with nothing in front of you in a queue.
On Mac
Export MP4, MOV, GIF or M4A, 1080p free and 4K with Pro, plus 8K on the Mac alone. Pro adds ProRes 4444 with alpha, 60 fps, a custom bitrate and the multi-aspect queue, which renders 9:16, 1:1 and 16:9 from one edit in a single action instead of three exports.
On iPhone
Make Vertical reframes to 9:16 around your face in one tap, and export runs on the phone itself with TikTok / Reels / Shorts / YouTube / Instagram presets. Pro adds 4K, 60 fps, a custom bitrate and transparent-background export (HEVC with alpha), plus the multi-aspect queue. Or hand the project to the Mac as a .zella file and finish on the big screen.
THE PROOF
- One-click reframe to 9:16 / 1:1 / 16:9 (center-crop, manual offset, or Vision face-detection auto-track)
- Export MP4 / MOV / GIF / M4A at 1080p free - 4K with Pro, plus 8K on the Mac
- Platform presets - TikTok/Reels/Shorts 1080×1920, YouTube 1920×1080, Instagram 1080×1350
- GIF with captions + zoom baked in
- Versioned project snapshots + autosave recovery
THE SPECIFICS, SINCE YOU ASKED
The numbers behind the promises.
25 transitions Mac and iPhone
46 / 45 colour looks Mac / iPhone; 43 and 42 base, 3 Founder-exclusive on each
10 / 9 caption presets Mac / iPhone
20 / 19 callout types Mac / iPhone
16 blend modes both platforms
8 / 12 title templates Mac bundles / iPhone one-tap templates; 16 kinetic entrances on iPhone, 20 animated title styles and 3 creator looks on Mac
16 voice changers both platforms
43 / 43 sound effects Mac / iPhone, synthesized on-device
40 backdrops 20 still, 20 animated
101 / 71 AI Assistant actions Mac / iPhone, free on both
19 caption languages on-device, iPhone
0 watermarks, accounts, uploads on every tier, forever
Plus your own .cube LUTs, SRT import, custom font import for any language, and 50 levels of undo. Counts are stated per platform wherever the two genuinely differ.
WHERE THE TWO APPS GENUINELY DIFFER
Two real apps, not an app and a viewer.
Nearly all of the page above is true on both. These are the parts that are not, listed plainly so you can decide before you install rather than after.
Only on the Mac
- 8K export
- ProRes 4444 with alpha, for a cutout that drops into any compositor
- Window capture and region capture
- The full keyboard set, including B to cut
- The menu-bar recorder
- Cursor and click tracking, which is what drives click-following auto-zoom
- The Library, which keeps every recording and project in one place
Only on iPhone
- Caption translation into 19 languages, on-device via Apple Translation (iOS 18 and later)
- Verbatim WhisperKit transcription, which is why "um" and "uh" actually get cut there
- Auto B-Roll with no key and no network: cutaway cards generated on the device
- Whole-device broadcast capture, the only honest way to show a mobile app working
Everything else on this page runs on both, with the counts noted where they differ. Move a project between them as a .zella file and the clips, cuts, captions, titles, callouts and effects come with it. Zella for iPhone →
WHAT COSTS MONEY, AND WHAT NEVER WILL
Free is the whole editor and all the AI.
Not a trial, not a demo, not a watermark you pay to remove. Free exports are clean and always will be. Recording is unlimited. There is no account. The only two ceilings anywhere are export length up to 2 minutes and resolution up to 1080p, and 1080p is free forever.
FREE FOREVER
$0
The complete editor and every AI feature, on Mac and iPhone. Auto-Polish, Make it Viral, the AI Assistant, Auto B-Roll, background removal, on-device captions, all 25 transitions, all the callouts and all the sound. Unlimited recording, with free exports up to 2 minutes. No watermark, no account, no upload.
PRO WEEK
7 days
Every new install runs the full Pro feature set for its first seven days. No card, nothing to cancel, and no countdown nagging you at the end. When it lapses you are on Free, which is the complete editor, not a crippled one.
PRO, ONE TIME
$89
Unlimited export length, 4K (8K on the Mac), transparent-background export, 60 fps, a custom bitrate, the multi-aspect export queue, saved templates and the depth toolset. One Universal Purchase covers Mac and iPhone together.
The AI Assistant and Auto B-Roll are free on both platforms and always will be: we gate output, never intelligence. Zella Founder is $299, a supporter tier led by every future major version free. Full pricing →
Try it on the next thing you record.
Nothing to sign up for, nothing to upload, and nothing stamped on the export. macOS 14 Sonoma+ · iPhone on iOS 18+ · Local-only · No cloud