Guide

Everything, step by step.

From the first launch to a finished multicam edit, subtitles, stories and clips. Every button, in the same words the app uses.

Getting started

What you need

  • A Mac with Apple Silicon (M1 or newer) running macOS 14 or later.
  • DaVinci Resolve Studio 19.1 or newer. The free edition of Resolve does not allow other apps to control it.
  • A project with the cameras synced and stacked on separate video tracks, and a microphone recording for each person — separate mic files, stereo files with one person per channel, and multitrack recorder files all work. If you only have the raw files, start with New Episode instead: it syncs everything and builds the project for you.
  • Keep the media drive connected while you work.

Everything SOLD CUT needs is already inside the app: no Terminal, no Homebrew, no Python install. The speech model used for subtitles (Whisper large‑v3, about 2.9 GB) downloads once, the first time you use Subtitles, and works offline after that.

Install

Open the DMG and drag SOLD CUT into Applications. If an older copy is running, quit it first.

Turn on scripting in Resolve

SOLD CUT drives Resolve through its scripting interface, so Resolve has to allow it once:

  1. In Resolve, open Preferences → System → General.
  2. Set External scripting using to Local.
  3. Restart Resolve.

Without this setting SOLD CUT cannot see or control Resolve at all, and the Resolve screen will explain that it can't be reached. See Resolve not found if that happens.

Your first episode in four steps

  1. Resolve. Choose the original synced timeline, or build it from the raw files with New Episode — Setup is then filled in for you.
  2. Setup. Name each person (2–10), choose their microphone and match their camera images: close‑ups, group shots, one wide shot, and an optional split screen. Save a show preset once you're happy with it.
  3. Analyze. Choose a pace and inspect the shot breakdown. Changing a setting always calls for a new analysis.
  4. Apply. SOLD CUT creates a copy of the timeline. Check the result in Resolve, then add subtitles, jump cuts, or social clips from that working copy.

The toolbar and the copy rule

The toolbar at the top of the app always shows the current project, the source timeline, and the working copy you're on. This is worth remembering throughout the app:

Every change SOLD CUT makes goes onto a new timeline copy. Earlier versions stay in Resolve, and the original timeline is never modified — you can always go back and start again from it.

New Episode (from raw files)

Use New Episode when you have a folder of raw camera and recorder files instead of an already‑synced timeline. It walks through six steps and hands you a finished Resolve project at the end.

1. Choose the folder

Point SOLD CUT at the episode's folder (cameras and recorder together). It groups the files into cameras and microphones automatically and shows a photo pulled from each camera so you can recognise them.

2. Say who is who

Check every line — names are guessed from file and folder names.

  • Give each microphone a person. For a stereo or multitrack recorder file, choose the channel that carries them, use Mix to combine every channel, or One person per channel when two people share one file on different channels.
  • Give each camera a role: a person's close‑up, the wide shot, a group shot, a split screen, or Other angle.
  • Cameras set to Other angle (illustrations, inserts) all share one Illustrations track; only where two of them overlap in time does a second one appear, on Illustrations 2. Their own sound (if they have any) goes on an Illustrations audio track, right below the recorded microphones, linked to the picture. It is never a person's microphone and never replaces the recorded screen audio the cut uses.
  • Split screens live here too, under the cameras: Split of A + B, Each pair, Everyone, or tick the people yourself. It's the same editor as in Setup — layout, lines dragged with the mouse, an optional divider PNG on top, and a camera per piece — just working on the scanned photos instead of a live timeline. Auto‑frame faces in new splits is on by default: for every new split, SOLD CUT reads about 24 frames from the camera files on this Mac (no Resolve involved) and centres each face, showing Face found, Moves a lot: check, or No face: frame by hand under each piece.

Blackmagic RAW and RED files cannot be read: export proxies with audio and scan the folder again.

3. Check sync

SOLD CUT matches every file to a reference microphone by sound, or by timecode when your cameras and recorder shared one jam‑synced clock. Files it isn't sure about are flagged, and you can type in where a flagged file actually starts, or skip this step and let sync run when the episode is created.

Files that couldn't be synced with certainty are never mixed into the cut. They land on disabled tracks named CHECK SYNC, placed after the main tracks — align them by eye in Resolve, then move them to the right track.

4. Frame rate, resolution, and where Resolve keeps its files

Choose the timeline's frame rate and resolution. Yellow chips warn you when a camera's frame rate doesn't match what you've chosen:

  • Exact multiples (50 on a 25 timeline, 60 on 30) look perfect; slower cameras simply repeat frames.
  • Almost‑equal rates (23.976 on 24, 29.97 on 30) repeat or drop one frame roughly every 42 seconds — usually invisible.
  • Other combinations (29.97 on 25, 30 on 24) judder on movement, and phones that record at a variable frame rate can drift. SOLD CUT warns you before creating the episode. Sound is never affected either way.

Resolve project location is the folder where Resolve keeps this project's cache, stills, and proxies. It defaults to whatever you used last, or to <episode folder>/SOLD CUT/Resolve otherwise. Choosing it here means Resolve won't ask you again with its own location dialog; the folder is created if it's missing and must be writable.

Resolve preset (optional) is a project preset already saved in Resolve. It's applied first, then the frame rate and resolution you chose here are applied on top.

5. Create episode in Resolve

This step creates a brand‑new Resolve project with a fully synced timeline. Whatever project was open before is saved automatically first (you can opt out, at the risk of losing its unsaved changes). Resolve is left on the Edit page with the new project and timeline open, and the first clip starts exactly on the timeline's first frame — no black frame at the head.

Files that couldn't be trusted land on disabled CHECK SYNC tracks, as above. The video tracks are ordered bottom to top: wide shot, group/2‑shot, the individual people's cameras, Split A + B (any split screens from step 2, built directly into the new timeline — no separate "SPLITS" copy is needed here), Illustrations, then CHECK SYNC.

If a split screen can't be built, the episode is still created in full: the timeline simply has no split screens yet, the report explains why, and the split appears in Setup so you can build it from there. A split that would start earlier than everything else is placed at the very start of the timeline instead, with the note "clamped to the timeline start — align by eye".

Playback frame rate. Resolve does not let other applications change it. If the project's playback frame rate differs from the new timeline's (for example playback at 24 while the timeline is 25), a yellow card appears: in Resolve, go to File → Project Settings (Shift+9) → Master Settings → Playback frame rate, set it to match the timeline, Save, then press Check again in SOLD CUT — the card turns green. Left alone, playback will judder and the sound will crackle. To stop this appearing on every episode, set the values once in Resolve and save them as your default: Project Settings → Presets → Save As, then Set as default, or use Save current settings as User Default Config. The same warning can appear on Resolve, Setup, and the Social Clips export screen for as long as the open project's playback rate differs from the timeline's.

6. Setup fills itself in

Setup is populated automatically from what you told SOLD CUT in step 2. From here, continue straight to Multicam Cut.

If you already have a synced timeline and don't need New Episode at all, use the classic path instead: open Resolve, choose the synced timeline, then go to Setup and Multicam Cut as usual.

Setup

Setup is where you tell SOLD CUT who's on screen, which camera shows whom, and how each shot should be framed. Everything here is saved with a show preset, so a returning podcast only needs a quick check.

People & microphones

Name each person (2 to 10) and choose their microphone. When a microphone track holds more than one channel — a stereo or multitrack recorder file — choose the channel that carries this person, or Mix to use every channel together. Two people can share a single file when each one uses a different channel.

Cameras

Give every camera a role:

  • Close‑up of one person.
  • Wide shot, showing everyone.
  • Group shot: a camera that frames two or three people but not everyone. Tick which people it shows — when exactly those people talk over each other, the cut reaches for this camera instead of the wide shot.
  • Split screen, described below.
  • Other angle, for illustrations and inserts: these are grouped onto their own Illustrations track, kept untouched by the cut, and their own sound never replaces the recorded screen audio.

Blackmagic RAW and RED files cannot be read: export proxies with audio and scan again.

Presets

  • Load preset brings back the saved people, camera assignments, vertical framing, Style, and tuning for a show. Check the camera images before continuing — track numbers can change between episodes.
  • Re‑detect replaces the proposed microphone and camera assignments with a fresh suggestion. Review the result before analyzing.
  • Save preset saves this setup, including your Style choice, on your Mac for the next episode. It never saves or changes anything in your Resolve project.

Vertical framing

Vertical framing sets where each camera is cropped for 9:16 output — vertical social clips and vertical stories. Move the frame over the person and zoom in if needed; it's saved with the preset, and it's what Social Clips and vertical Stories use to follow whoever is on camera.

Auto‑frame faces

Auto‑frame all cameras reads about 24 frames from each camera belonging to a person, finds the face using Apple's on‑device Vision framework — locally, nothing sent online — and sets the vertical framing: face centred horizontally, a little above the middle (eyes around the top third), zooming in only when the face is small. Under any single camera, the same button — just Auto‑frame face — does it for that camera alone, and can also be used on a wide or group shot if you want.

Wide shots, group shots, and split screens are not touched by "all cameras" — they hold more than one person. You'll see one of these messages once it's done:

  • Face found in N of M frames — done.
  • Moves a lot: check — the person moves a great deal; double‑check the framing.
  • N people in this shot — more than one face was found, and the largest was chosen.
  • No clear face found: frame it by hand — move the framing rectangle yourself.

SOLD CUT reads the original media files; if they're offline, it falls back to the Resolve proxy. Nothing in Resolve is changed by this step, and the result is saved into the preset exactly like a manual framing. Frames it has already read are cached on your Mac (in faces_cache inside the app's folder), so a second run is much faster.

Split screens

Split screens appear in Setup whenever no single camera frames two or more people together. Choose Each pair, Everyone, or tick your own group of people, then set up each split:

  • Layout: side by side, top and bottom, columns, one large piece plus two small ones, a grid, or Stacked (story) — two people at half size, stacked top and bottom, built for 9:16 stories.
  • Who sits where: a new split follows the studio convention, the same as React: person 1 on the right, person 2 on the left (and on top in a stacked layout). Drag one picture onto another to swap them.
  • Lines dragged with the mouse to resize each piece.
  • Who sits where — drag one person's photo over another's to swap them — plus each piece's camera, position, and zoom. Cameras keep their own size (100% zoom); they're only shifted and cropped into their piece, the same way a hand‑made double frame is built.

The dividing line can be generated — the studio's own look, black, 37 px wide at 4K (16 px on a story) — or a divider PNG placed on top: the classic 16:9 or 9:16 dividers, or your own, added through My overlay PNG…, which copies it into My overlays. Set as standard makes your choice the default for new splits and saves it in the preset.

Split screens can also be designed directly inside New Episode's step 2; those are built straight into the brand‑new timeline, and Setup then shows them as Built.

Build split screens copies the timeline to a new one named "… SPLITS". Each split is built in its own helper timeline ("… SPLIT A + B"), turned into a single compound clip named SPLIT A + B, and placed on a new video track immediately above the cameras — illustration and graphics tracks move up one to make room. If Resolve refuses the insert, the track is placed on top of everything instead and the report says "Split track placed above all tracks". If Resolve refuses to make the compound clip, the nested helper timeline is used in its place and SOLD CUT tells you so. The original timeline is never changed, and building again — even on a timeline that already has splits, such as one from New Episode — replaces the old split tracks in the new copy.

Once built, the copy becomes the source for Multicam Cut: every split screen is used exactly like a group shot, reached for whenever exactly those people are talking over each other.

Auto‑frame faces works per split too: it finds each person's face on the camera used for their piece and frames it (a little above centre, zooming only when the face is small — it does not touch the camera's own vertical framing), showing the same Face found / Moves a lot: check / No face: frame by hand status under each piece. If you've already run "Auto‑frame all cameras", new splits start out already framed.

After any change to a split, press Build split screens again to update the copy.

Multicam Cut

This is where SOLD CUT proposes the edit itself: which camera is on screen, and when.

Analyze

Analyze reads the microphone recordings and proposes camera changes. It writes nothing to Resolve — it's entirely safe to run as many times as you like. Run Analyze again after changing anything in Setup, Style, or Settings.

Style

Choose a pace — Slow, Balanced, or Fast — which sets how the cut treats a handful of related choices at once: how it handles interruptions (people talking over each other), how often it reaches for the wide shot, and how often it reaches for a split screen. Fine‑tuning any of these individually, in Settings, turns the Style into Custom. Always analyze again after a change here.

Result breakdown

The result shows the proposed shot list with its timecodes. Camera gaps — a stretch where the chosen camera has no footage — are automatically covered by another angle or by the wide shot, and group shots are used whenever exactly the people they frame are speaking together. If the report says CHECK or mentions holes, look at those spots before delivering. Every angle, whether it was used or not, is kept on the copy as a disabled clip, so you can always swap the visible shot yourself afterwards in Resolve.

Apply to a Copy

Apply (or Apply again) creates a new timeline copy — named "… AUTO" — and builds the proposed cut there; the original timeline is never touched. Any zoom, punch‑in, or grading already sitting on the individual cameras is preserved on the copy. Keep Resolve open while this runs, then check the result and watch the cut before you move on or deliver it.

Open in Resolve and Cancel

Open in Resolve selects the generated timeline so you can watch it and adjust the edit by hand. Cancel stops whatever is running and restarts the analysis engine; if you cancel while a timeline is being built, an incomplete copy may be left behind in Resolve for you to delete — the original is never affected.

React & illustrations

React appears in the sidebar whenever the timeline has an Illustrations track — New Episode creates one from any camera set to Other angle — or when the show is set up as a react show. While each illustration plays, it builds the studio's phone‑in‑the‑middle look on a copy named "… REACT" of the input timeline, normally the multicam "… AUTO" copy.

What it builds

For the length of each illustration:

  • The two close‑ups sit side by side — left and right of the phone — each framed by its own Fusion composition matching the studio's: a 1.1x transform, a mask covering half the frame (0.25 on the left side, 0.75 on the right), and a Center X value keyframed to follow the face.
  • The illustration itself is shrunk into the phone's screen: a 9:16 video at 0.8 scale, a 9:16 photo at 0.62 scale; other shapes fit inside the same screen.
  • A new track, React background (the phone body and its shadow), sits immediately below the illustration, and a new track, React frame (the outline with its Dynamic Island cut‑out), sits immediately above it.
  • Every other camera angle, split screen, group shot, and any track that would sit between the close‑ups and the illustration is turned off for that stretch.
  • The illustration's own sound plays, while the screen audio recorded in the room (the Bluetooth feed) is cut and muted exactly for the illustration's duration. The wide shot underneath and the microphones are left untouched.

Layout and sides

Who sits left and who sits right of the phone is chosen once per episode — the studio default is speaker 1 on the right, speaker 2 on the left — and Swap sides exchanges them. Each side's default camera is that person's highest camera track (their close‑up).

Phone layout offers Phone classic, the studio's own phone, or Add layout…, which asks for two PNGs sized to the timeline — a background with transparent sides and a frame — copied into My react layouts. The zoom values (close‑up 1.1, video 0.8, photo 0.62) can all be adjusted.

Follow faces

On by default: SOLD CUT reads one frame per second from each close‑up (locally, on this Mac), finds the face, and writes one Transform keyframe per second into the Fusion composition, easing between them the way an editor's own keyframes would. Small movements are ignored, so the picture doesn't jitter and a stray wrong detection is smoothed out. With no face found, or with Follow faces switched off, the framing stays fixed. You can inspect the resulting keyframes on the Fusion page of any close‑up.

Mute the screen audio

This ticks, by default, the recorder track set to Other in New Episode, the Setup screen‑audio track, and any track named like Bluetooth. Each one is cut at the start and end of every built illustration, and that piece is disabled for the illustration's duration. Microphones and the illustrations' own sound can never be ticked here.

Preview and Build

Preview changes nothing: it lists every illustration with its length, which tracks will turn off, and — with Follow faces on — the face keyframes for each side. Illustrations shorter than the minimum length (3 seconds, adjustable) are never built; the result lists them with their timecodes instead, marked "Place by hand" since they're too short to sync automatically.

Build creates the "… REACT" copy and verifies it — each close‑up should end up as one enabled clip with its composition, with no other camera or screen audio active during an illustration. If anything fails, the copy is deleted and the input timeline is never touched. Once built, it becomes your working timeline.

Find illustrations by sound

For older timelines that don't have an Illustrations track yet: choose the folder containing the episode's illustrations, then Listen — SOLD CUT matches each illustration's own sound against the recorded screen‑audio track to find where it played. Tick the ones you want built; Build first places them on a new Illustrations track (with their sound) in the copy, then builds the phone look around them as usual.

Subtitles

Transcribe

Transcribe reads the configured microphone audio and saves subtitles locally, running entirely on your Mac with the Whisper large‑v3 speech model. The model (about 2.9 GB) downloads once, the first time you use Subtitles, and transcription works offline after that. Redo replaces the saved transcript — export an SRT first if you want to keep your own edits.

Review with AI

Review with AI is an optional, online step: it reads the whole episode's transcript, deduces from context — what was said just before and after each piece — and corrects misheard words, diacritics, digits (numbers are written out as digits), and punctuation. Reels get an extra pass: the reel's own sound is transcribed in its original language and translated onto a separate translation track, and any change it makes can be undone.

This step requires your own Anthropic API key and a consent checkbox; the key is stored in the macOS Keychain, and only the transcript's text is ever sent — never audio or video. A slower, more careful model option is also available for higher accuracy, at a higher cost on your own Anthropic account.

Undo and edit

Undo restores one AI change at a time. Click any transcript line to edit its text, then Save. Re‑run Add to Resolve afterwards to bring your changes into Resolve.

Subtitle style

Cues are kept short — about a second, roughly four words — with a maximum of two lines of 26 characters each, and timed to the individual words. On import, a cue that starts within a few frames of a camera cut is snapped to start exactly on the cut.

Subtitle look

When you run Add to Resolve, choose how the subtitles should look:

  • Studio style — the studio's own subtitle look (Montserrat ExtraBold, sitting below the divider line on a vertical story). This requires the Montserrat font to be installed on your Mac.
  • Resolve default.
  • Copy the look of … any other timeline already in the project.

Add to Resolve and export

Add to Resolve creates a copy of the working timeline and imports the subtitles onto its first subtitle track. If subtitles are already there, you'll be asked before they're replaced in the copy. Export SRT saves a copy wherever you choose, and Show SRT file opens its location in the Finder — neither one touches the timeline itself.

Stories

The studio order

Stories are meant to come at a specific point in the workflow:

  1. Run Multicam Cut, then do your own review of the result — trim, delete, and correct shots in the "… AUTO" copy, or in your own copy of it.
  2. Run Subtitles on that reviewed timeline: Transcribe, then Review with AI. This is the transcript Stories will pick from, and the eventual STORIES copy reuses it rather than transcribing again.
  3. Only then open Stories, and choose that reviewed timeline as the Reviewed timeline.

Quick vs. With AI

  • Quick (reels) proposes a candidate at each reel, ending right before the next reel begins or at a natural pause. It needs no transcript — just tick the ones you want.
  • With AI starts with Transcribe this timeline (run once, after your review), then Pick stories with AI: it reads the conversation and chooses self‑contained stories, each with its own title.

Markers

Put N stories as markers on the timeline drops a green Story N IN marker on each story's first frame and a red Story N OUT marker on its last. Move either marker in Resolve to adjust a story's boundaries, or delete both to drop that story entirely — any markers of your own are left alone. If SOLD CUT's own markers aren't present, your own markers are read instead, in IN/OUT pairs. Once you've adjusted anything in Resolve, use Read the markers again to pick up the changes.

Build in a copy

Build N stories in a copy creates a copy named "… STORIES" that keeps only the stories, at their own positions on the timeline (with gaps between them), with every camera angle still present on its own track. Choose a vertical 9:16 format (using the framing from Setup) or the episode's own format.

Studio story look

The Studio story look is on by default for vertical stories:

  • While someone is talking, close‑ups stay as individual cameras. Wherever the wide shot or a split screen would have appeared instead, Double frame (on by default) builds the story's own double frame: the two people stacked, the person on the left sitting on top, each framed within their own half, separated by a 24 px black line placed on a Story double frame track. With Double frame switched off, the camera of whoever is speaking is shown instead (worked out from the microphones).
  • During a reel, the reel itself plays across the top 60% of the frame, over a blurred copy of itself; below it, the two of you sit side by side — left and right, exactly as in React — each framed so their face sits centred in their box. The React phone is turned off for this, and a card carrying the dividing lines sits above everything on a Story card track.

Left person / Right person lets you choose the person and their exact camera used both under reels and in the double frame — Automatic uses the person's own camera, without React's Fusion composition. Run Setup → Auto‑frame all cameras beforehand so every face starts out centred.

Subtitles for these stories

Subtitles for these stories works on the STORIES copy specifically. If the reviewed timeline already carried a transcript, it's carried over to the copy rather than transcribed again; otherwise only the stories themselves are transcribed, which takes a few seconds per story. At Add to Resolve, any line that falls on the double frame moves up to sit on the line between the two people, and a line never crosses the double frame's edge — it's cut there instead.

Jump Cuts & Social Clips

Input timeline

Both Jump Cuts and Social Clips start from the timeline chosen at the top of their screen, labelled Input timeline. By default this is your latest working timeline — the multicam cut — so short‑form edits inherit the finished camera switching automatically.

Jump Cuts

Choose the microphones to listen to, preview the quiet gaps found between them, then either:

  • Build a tightened copy with the gaps removed, or
  • Mark gaps only — put a red marker on every gap in a copy without cutting anything, so you can decide each cut yourself in Resolve.

The input timeline itself is never changed.

Social Clips

Suggested moments come from the local transcript and still need your editorial judgement before publishing; manual time ranges work even without a transcript at all. Follow speaker builds each vertical clip from the multicam cut and follows the active camera, using the vertical framing you set in Setup.

Export videos

Once Social Clips has created its timelines, choose an export folder and start the export jobs in Resolve. Refresh shows the real job status reported by Resolve — it never claims a clip is finished until Resolve itself says so.

If the open project's playback frame rate differs from the timeline being exported, the same warning described under Create episode in Resolve can appear here too. Fix it once in Resolve and save it as your project default.

What runs locally

SOLD CUT is built so that your footage never has to leave your Mac unless you explicitly ask for something that needs it.

TaskWhere it happens
Camera and face analysis, sync by sound, Auto‑frame faces, Follow facesOn your Mac only. Nothing is sent online.
Multicam Cut, split screens, React, Jump Cuts, Social ClipsBuilt entirely on your Mac; your audio and video never leave this computer.
Transcribe (speech recognition)Runs on your Mac with the Whisper large‑v3 model, after a one‑time download.
Review with AI (subtitle correction, Stories picks with AI)Optional, and only when you start it. Sends transcript text, the show description, and your vocabulary to Anthropic, using your own API key stored in the macOS Keychain. Your audio and video are never sent.
App updates, the one‑time speech‑model downloadFetched online.

Social suggestions are local transcript heuristics, not automatic editorial decisions — review every generated timeline in Resolve before you publish it.

Keyboard shortcuts

ShortcutWhat it does
⌘1–⌘9Switch screens: 1 New Episode, 2 Resolve, 3 Setup, 4 Multicam Cut, 5 React, 6 Subtitles, 7 Stories, 8 Jump Cuts, 9 Social Clips.
⌘,Open Settings.
⌘ReturnRuns the main action of the current screen.
⌘.Stops the operation that's currently running.
⌘⇧RRefresh from Resolve.
⌘?Opens this Guide.

Troubleshooting

Resolve not found

If the Resolve screen says it can't reach Resolve:

  • Check Preferences → System → General → External scripting using → Local in Resolve, then restart Resolve.
  • Make sure you're running DaVinci Resolve Studio, not the free edition — the free edition cannot be controlled by other applications.
  • Make sure Resolve is open with a project loaded, then use ⌘⇧R to refresh.

Playback frame‑rate warning

Resolve doesn't let other applications change a project's playback frame rate. When it differs from the timeline you're working on, a yellow warning appears on the Resolve, Setup, or Social Clips export screens. Fix it once in Resolve: File → Project Settings (Shift+9) → Master Settings → Playback frame rate → set it to match the timeline → Save, then press Check again. To stop it coming back on every new episode, save those settings as your Resolve default: Project Settings → Presets → Save As, then Set as default, or Save current settings as User Default Config.

Cancel restarts the engine

⌘. (or the Cancel button) stops whatever is currently running and restarts SOLD CUT's analysis engine. If you cancel while a timeline copy is being built, an incomplete copy may be left behind in Resolve — safe to delete — but the original timeline is never touched.

Export diagnostics

If something isn't working, go to Settings → Diagnostics → Export diagnostics… and send the resulting zip along with the steps to reproduce the problem.

The diagnostics zip never contains your API key or your transcripts.

Media drive and unsupported formats

Keep the media drive connected while you work — several tools, including Auto‑frame faces, read the original camera files directly (falling back to a Resolve proxy only when the originals are offline).

Blackmagic RAW and RED files cannot be read by SOLD CUT. Export proxies with audio from those cameras, then scan the folder again.

What does this button do?

Every button in SOLD CUT, in the same words the app uses. Type in the search box to filter.

New Episode

Start from the folder of raw files instead of a synced timeline. The app groups the files into cameras and microphones (photos, sidecar and project files are ignored), you confirm who is who, it syncs everything by sound or timecode and builds a new Resolve project. Setup is then filled in for you.

Who is who

Filled in from file and folder names, so check every line. Name each person and pick their microphone; for a stereo or multitrack recorder file choose the channel, or use One person per channel. Give each camera a person, the wide shot, a group shot or split screen. Cameras set to Other angle (illustrations, inserts) share one Illustrations track, with a second one only where two of them overlap in time, and are left untouched by the cut. Their own sound goes on an Illustrations audio track right below the recorded microphones, linked to the picture; it is never the microphone of a person or the screen audio of the cut. Blackmagic RAW and RED files cannot be read: export proxies with audio and scan again.

Check sync

Matches every file to the reference microphone (Automatic uses the longest one; if nothing matches it, the app tries another microphone and tells you) by sound, or by timecode for jam-synced cameras. Each file shows how it was placed or what to check. Tick "clocks are jam-synced" only when every camera and recorder shared one timecode clock. Drift shows how far the end of a long file moves (+ means its end plays early); over one frame is highlighted. You can type where a flagged file starts, or skip this step and let sync run when the episode is created.

CHECK SYNC tracks

Files that could not be synced with certainty are never mixed into the cut: they go on disabled tracks named CHECK SYNC, after the main tracks. Align them by eye in Resolve, then move them to the right track.

Timeline frame rate

Defaults to the rate most cameras use; choose the rate you deliver. Exact multiples (50 on 25, 60 on 30) look perfect and slower cameras repeat frames. Almost equal rates (23.976 on 24, 29.97 on 30) repeat or drop one frame about every 42 seconds, usually invisible. Other combinations (29.97 on 25, 30 on 24) judder on movement and phones with a variable frame rate can drift: the app warns before creating. Sound is never affected.

Create episode in Resolve

Saves the project open now (you can opt out, at the risk of losing its unsaved changes), creates a new project with the chosen frame rate and resolution, imports the files into bins per device, places every synced file at its exact frame and names the tracks after the people. If something fails halfway, the report names the project that was left in Resolve so you can check it; the app never reports an incomplete episode as done.

Resolve / Refresh

Checks the open Resolve project and lists its timelines. Refresh reads the current timeline again and resets the episode preview. It does not edit clips. If Resolve cannot be reached, the Resolve screen explains why and how to fix it.

Load preset

Loads the saved people, camera assignments, vertical framing, style and tuning for a show. Check the camera images before continuing: track numbers can change between episodes.

Re-detect

Replaces the proposed microphone and camera assignments with a fresh suggestion. Review the result before analyzing.

Save preset

Saves this setup, including the Style choices, on your Mac for the next episode. It does not save or change your Resolve project.

Microphone channel

Shown when a microphone track holds more than one channel (a stereo or multitrack recorder file). Choose the channel that carries this person, or Mix to use all channels together. Two people can share one file when each uses a different channel.

Group shot

A camera that frames two or more people but not everyone. Tick the people it shows. When exactly those people talk over each other, the cut uses this camera instead of the wide shot.

Split screens

Shown in Setup when no camera frames two or more people together. Choose Each pair, Everyone or your own group, pick a layout, drag the lines, choose which person goes in which piece, their camera, framing, line thickness and colour. Cameras keep their own size and are shifted and cropped into their piece, like a hand-made split. Lines: generated (the studio look: 37 px black at 4K) or a divider PNG on top (the classic dividers, or your own from My overlays; Set as standard makes it the default for new splits and is saved in the preset). Stacked (story) puts two people at half size, top and bottom, for 9:16 stories. Auto-frame faces finds the face of each person on the camera of their piece and frames it (face a little above the centre, zoom only when the face is small; the vertical framing of the cameras is not changed); new splits start framed when faces were already found. Build split screens copies the timeline to "... SPLITS", builds each split in its own helper timeline, turns it into one compound clip "SPLIT A + B" and places it on a new video track right above the cameras (illustration and graphics tracks move up one; if Resolve refuses the insert, the track goes on top and the result says so). The copy becomes the source for the cut and every split is used like a group shot. The original timeline is never changed. Building again (also on a timeline that already holds splits) replaces the old split tracks in the new copy.

Split screens in New episode

In step 2 (Who is who), under the cameras: the same split editor as Setup, on the camera images of the scan. Add a split of two people, each pair or everyone; with Auto-frame faces in new splits (on by default) the app reads about 24 frames of each camera file on this Mac and centres each face in its piece, with the status under every piece. Create episode builds the splits straight into the new timeline (no SPLITS copy): tracks named Split A + B right above the cameras, under Illustrations and CHECK SYNC. Setup then shows them as built group shots. If a split cannot be built, the episode is still created without it; the report says why and Setup keeps the design so you can build it there.

React / Illustrations

Shown in the sidebar when the timeline has an Illustrations track (New episode puts the Other-angle cameras there, in sync, with their own sound) or the show is a react show. While each illustration plays it builds the studio look on a copy "... REACT" of the input timeline (normally the multicam copy): both close-ups side by side, each framed by a Fusion composition that follows the face; the illustration zoomed into a phone in the middle (a 9:16 video at 0.8, a photo at 0.62; other shapes fit the same screen); the phone background on a new track right under the illustration and the phone frame right above it; every other camera angle, split and group shot off, and any track between the close-ups and the illustration (a story track there would cover the close-ups); the illustration's own sound on and the screen audio (the room's Bluetooth feed) muted exactly for that time. The wide shot underneath and the microphones are untouched. Illustrations shorter than the minimum (3 s) are not built: the result lists them with their timecodes to place by hand. If anything fails the copy is deleted; the input timeline is never changed.

Left / Right close-up

Who sits left and who sits right of the phone is chosen per episode (the studio: speaker 1 right, speaker 2 left). Each side picks the person and their close-up camera (the highest camera track of the person by default); Swap sides exchanges them.

Follow faces

On (default): the app reads a frame of each close-up every second (on this Mac), finds the face and writes one Transform keyframe per second into the Fusion composition, held values that ease into each other like the editor's own keys; small moves are ignored so the picture does not jitter, a wrong detection is smoothed out. No face or Follow off: one fixed framing. Check the keys in the Fusion page of a close-up.

Phone layout / My layouts

Phone classic is the studio phone (3840x2160 background with the phone body and its shadow, and the outline with the Dynamic Island). Add layout asks for two PNGs the size of the timeline: the background (sides transparent) and the frame; they are copied into My react layouts. The close-ups split at the centre of your phone and the face targets follow its width. Built timelines use a private copy of the PNGs, so deleting a layout never takes them offline.

Mute the screen audio

Ticks the tracks that recorded the screen in the room (New episode: the recorders set to Other; also the Setup screen audio and tracks named like Bluetooth). Each is cut at the start and end of every built illustration and that piece is disabled; microphones and the illustrations' own sound cannot be ticked.

Preview / Build (React)

Preview reads the timeline without changing it: every illustration with its length and zoom, the short ones to place by hand, which tracks go off and, with Follow faces, the face keys per side. Build makes the copy and verifies it (each close-up one enabled clip with its composition, no other camera or screen audio during an illustration). The result becomes the working timeline.

Find illustrations by sound

For older timelines without an Illustrations track: choose the folder with the episode illustrations and Listen; the app finds where each video played by matching its sound against the screen audio track. Tick the ones to build; Build places them on a new Illustrations track (with their sound) in the copy and builds the phone around them.

Vertical framing

Sets where each camera is cropped for vertical social clips. Move the frame over the person and zoom in if needed. Saved with the preset.

Auto-frame Faces

Finds the face in about 24 frames of a camera (on this Mac, nothing is sent online) and sets its vertical framing: face centred, a little above the middle, zoomed in only when it is small. Auto-frame all cameras does this for every camera of one person; wide, group and split-screen shots keep their manual framing. Check each result.

Analyze

Reads the microphone recordings and proposes camera changes. Nothing is written to Resolve. Analyze again after changing Setup, Style or Settings.

Apply / Apply again

Creates a new timeline copy and builds the proposed cut there. Keep Resolve open. Check the verification results and watch the cut before delivering it.

Open in Resolve

Selects the generated timeline so you can watch it and adjust the edit.

Cancel

Stops the current operation and restarts the analysis engine. If you stop while a timeline is being built, an incomplete copy may remain in Resolve; the original is never touched.

Subtitles

Transcription runs on your Mac with the Whisper large-v3 speech model. The model (about 2.9 GB) downloads once, the first time you use Subtitles; after that, transcription works offline.

Transcribe / Redo

Reads the configured microphone audio and saves subtitles locally. Redo replaces the saved transcript; export an SRT first if you want to keep your edits.

Review with AI

Optional online correction of transcript text. Requires your Anthropic API key and the consent checkbox. The key is stored in the macOS Keychain. Provider usage costs are separate; your audio is not sent.

Undo / Edit transcript

Undo restores one AI change. Click a transcript line to edit its text, then Save. Re-import to reflect changes in Resolve.

Add to Resolve

Creates a copy of the working timeline and imports subtitles to track 1. Existing subtitles require your choice before replacement in the copy.

Export SRT / Show SRT file

Export saves a copy where you choose. Show opens the saved subtitle file location. Neither edits the timeline.

Stories

After the multicam and your review: suggest stories (Quick: on the reels; With AI: from the transcript of the reviewed timeline), put them as green IN / red OUT markers on the timeline, move them in Resolve, then build a copy “… STORIES” with only the stories at their places. Subtitles then work on that copy only.

Subtitle style

Short cues (about a second, at most 2 lines of 26 characters) timed by the words; on import a cue that starts a few frames from a camera cut starts on the cut. The AI review deduces misheard words from the context, writes numbers as digits and translates the reels’ own sound.

Input timeline

Jump Cuts and Social Clips start from the timeline you choose at the top of the screen. By default this is your latest working timeline (the multicam cut), so short-form edits use the finished camera switching.

Jump Cuts

Choose the microphones, preview the quiet gaps, then either build a tightened copy or mark every gap with a red marker on a copy so you can decide each cut yourself. The input timeline is not changed.

Social Clips

Suggested moments come from the local transcript and need your editorial review; manual time ranges work without a transcript. Follow speaker builds each clip from the multicam cut and uses the vertical framing set in Setup.

Export videos

After Social Clips creates timelines, choose an export folder and start Resolve export jobs. Refresh shows the real job status returned by Resolve; it does not claim completion until Resolve reports it.

Settings / Reset tuning

Advanced editing preferences. Values that Style also sets are marked; changing them turns the Style into Custom. Reset returns tuning to the defaults. Analyze again before applying a new cut.

Export diagnostics

Settings > Diagnostics saves a zip with app logs and versions for support. It never includes your API key or transcripts.

Nothing matches. Try a shorter word, or ask us.