How to Use AI Lossless Video Cutter for macOS

Start here

AI Lossless Video Cutter is a native Mac app for marking keep regions in a video and exporting them so picture and sound stay as close as possible to the original file. You are not choosing "delete this frame"; you are saying "keep this part of the timeline."

A simple project looks like this: open a clip, mark one or more segments with in and out points, preview, then export as one combined file or separate clips. Optional scans can suggest segments for you (similar shots, a specific face, objects, spoken words, and more).

Opening a video and playing it

Use File → Open or the keyboard shortcut ⌘O. Supported video formats include MP4, MOV, M4V, MTS/M2TS/TS (MPEG-2 transport stream), MXF, and QT. You can also open still images (PNG, JPEG, HEIC, TIFF, GIF, WebP) or audio-only files. Images are turned into a short clip at a duration you choose, and audio-only files can be trimmed the same way as video. If a file is protected or uses an unusual codec, it might not open or export the way you expect. Try a test export early when working with new sources.

Press Space to play and pause. Arrow keys jump backward and forward by a small amount (you can change the jump distance in settings). Optional looping sends you back to the start when the clip ends.

Once you have marked some keep segments, Play Segments Only (in the trim toolbar next to Loop, or under the Playback menu) plays back just those segments and jumps past the removed portions so you can audition the final cut before exporting. The setting persists between launches.

Basic trimming: In, Out, and segments

Think in three steps:

Scrub or play to where a keep region should start, then set In (I).
Move to where that region should end, then set Out (O).
The app commits that range as a segment and clears the working in/out so you can mark the next part.

On export, anything outside your segments is left out (unless you add more segments to cover it). If two segments overlap or touch, the app merges them into one continuous keep range so you do not have to micromanage edges.

Most automatic scans add suggested segments the same way. Two special tools (remove silence and remove black screens) work differently: they analyze the whole file and replace your selection with "everything except silence" or "everything except black frames" in one step.

The Invert button flips your selection: everything you marked as keep becomes cut, and everything outside becomes keep. This is useful when it is easier to mark the parts you do not want, then invert. You can also jump directly to the start or end of the active segment with ⌘[ / ⌘].

Timeline and preview

The timeline shows your keep segments as colored regions. You can click or drag to move the playhead, drag the edges of a segment to fine-tune in and out, and delete a segment when needed. When a long scan is running, you may see progress feedback on the timeline (for example, a pulse while the app searches).

The preview can zoom in so you can check fine detail, which is useful for faces, small text, or checking exact framing. When zoomed, a small navigator shows which part of the frame you are looking at. After some face scans, a brief highlight may appear on the preview to show where a match was found.

Exporting

When you are happy with your segments, start export with ⌘E to produce one merged file with all segments concatenated in order. When you have multiple segments, the Export button expands into a menu so you can instead choose Export Segments as Separate Files to write each segment as its own clip, which is handy for batching social cuts or archival splits. The app remembers whichever mode you used last as the default.

The app checks that you have enough free disk space before writing. If you use multiple segments or custom audio tracks, it may build a temporary edit internally and still export without re-encoding video when the format allows.

For workflows that need a small, easy-to-share file, there is also a proxy export path that re-encodes for size. That is not lossless, but it can be useful for previews or rough cuts.

To grab a still from the video without exporting the whole clip, use the camera button in the toolbar. It lets you save the current frame as an image file or copy it to the clipboard.

What "lossless" means (and why edges may shift slightly)

Lossless here means the app copies the original compressed video and audio instead of decoding to pixels and re-encoding. You avoid generation loss and keep the same quality as your source.

Compressed files like MP4 are not cuttable at every single frame. Safe cut points usually line up with keyframes (full refresh frames). The app may snap your in and out times slightly inward to the nearest valid points so players stay happy. Picture quality stays the same; only the exact edge timing may move a little compared to frame-by-frame editors that re-encode.

You do not need FFmpeg installed for normal use; the app relies on macOS media support. Some optional metadata tweaks after export may use FFmpeg when it is available on your system; the core trim is still designed around passthrough export.

Free version vs Pro

The free app includes the full core workflow: open files (video, image, or audio), set segments, use the timeline, detect the shot at the playhead or across the whole file, build a playlist, and export losslessly with keyframe-safe alignment.

Pro unlocks discovery features such as finding similar shots, finding a specific face or object, removing silence and black screens, speech-based search, and richer audio-track options. See pricing on the home page for the current list.

After purchase, use the Access Purchase button in your email to open the customer portal. Your license key is listed in the Benefit Grants section on that same page, with no extra navigation beyond opening the portal. Copy it, then paste it in the app's Settings to activate this Mac.

Smart scans: how to think about them

Scans are helpers. They propose segments; you can still drag edges, delete mistakes, or mix manual marks with automatic ones. Most scans work by sampling the video over time, so tighter sampling tends to find shorter appearances but takes longer. Looser sampling is faster but might miss very brief moments.

Many scans also run shot detection around each hit so each match becomes a full scene-sized segment, not just a single frame. You can switch between "real shot boundaries" and "fixed-length windows" in settings when you want a consistent clip length around each find instead of expanding to the full shot.

Face recognition and object detection use on-device models. Your video is not uploaded for those features. Speech search uses the system's speech APIs and needs microphone-related permissions where macOS requires them.

Detect shot (at the playhead)

When you are somewhere inside a shot, use Detect current shot (shortcut S). The app looks for where the picture changes significantly to the left and right of the playhead and adds a segment for that whole shot.

Use Detect All (⇧S) to scan the entire file at once and add a segment for every detected shot boundary. This is useful for quickly marking up a long clip.

This is classic editing assistance, not a labeled "AI scene classifier." Sensitivity and search width live in settings: higher sensitivity finds more cuts (including some you might not want); a wider search window helps on long takes. Tune scan interval under AI Scanner when Find tools feel too slow or too sparse.

Find visually similar shots (Pro)

Pick a reference look (usually the current frame or a still image), and the app scans the timeline for moments that resemble it. That is ideal for recurring B-roll, matching locations, or the same setup repeated across hours of footage.

Each match is turned into a segment using the same shot logic as above. If the match threshold is too loose, you get extra segments; if it is too strict, you may miss variations in lighting or crop. Adjust scan interval and sensitivity in settings until results feel right for your material.

Find a face (Pro)

Pause on a clear view of the face, or import a photo, and start a face scan. If several faces are visible, the app can ask you to pick the right one. It then searches for other moments where that face appears and builds segments around each appearance.

Face search is tuned for identity, not just "this exact framing," so it behaves differently from visual similarity on the whole frame. Progress often has two phases: first a fast pass across the video, then extra confirmation work grouped by time and likeness. When you want stricter matching, tighten face tolerance in settings; loosen it when lighting and angles vary a lot.

Find All Faces sweeps the entire video and shows one thumbnail per unique person in the Faces window. Click a face to seek (or open its Occurrences view when there is more than one appearance), click the name to rename, or drag one face onto another to merge identities by hand. If the automatic grouping merges too much or splits a person across multiple cards, open the Dedup sheet: a Match Sensitivity slider previews the resulting face grid live, and Apply commits the merge without rescanning the video. The Face Sheet export uses the same dedup pass, so it shows each person once and follows your AI Scanner settings rather than a per-export interval.

Find objects (Pro)

Object mode looks for categories you choose (people, cars, animals, and more depending on the built-in label list). You typically select a detection from the current frame or a still, then run the scan. The app samples the video and turns hits into shot-level segments.

Object finding prioritizes accurate framing over raw speed, so it may work harder per step than the visual-similarity scan. Confidence threshold in settings helps control how sure the detector must be before counting a hit.

Speech search (Pro, macOS 26+)

On macOS 26 or later, you can type a phrase and let the app transcribe the audio, match your wording (including close typos and similar meaning), and add segments around strong matches. macOS may need speech recognition enabled and language models downloaded. Follow any prompts the system shows.

A similarity slider in settings balances "must match words closely" versus "accept paraphrases." Raise it when you want precision; lower it when people say the idea in different words.

Remove silence and remove black screens (Pro)

Remove silence listens for stretches where the volume stays very low, merges them into gaps, and rebuilds your selection from the audible parts. This is great for cleaning interviews or screen recordings. Padding and minimum-length options in settings avoid choppy cuts and tiny slivers of audio.

Remove black screens looks for frames that are both very dark and low in texture, so ordinary night scenes with detail are less likely to be mistaken for blank gaps. Like silence removal, it replaces the whole selection in one go.

Settings worth knowing

Open Settings from the app menu to tune detection and export behavior. Besides scan interval and shot sensitivity, you will find thresholds for face matching, object confidence, silence and black-frame detection, and speech similarity.

Scan interval (visual / object) in Settings → General → AI Scanner controls how often the app grabs a frame during Find Similar Shot, Find Object, and speech search. Default is 0.25 s; you can raise it up to 3.0 s for faster scans on long files. A shorter interval catches briefer appearances at the cost of more scanning time; a longer interval finishes faster but may skip very short moments between sample times.

Export naming lets you choose filename patterns, for example a suffix like _trimmed and tokens that insert segment counts or aspect labels when relevant. You can also prefer saving next to the original file so exports land in the same folder as your source.

Display aspect is an advanced option: you can tag the file so players letterbox or scale using metadata instead of changing pixels. Not every player honors those tags the same way, but it is useful when your canvas ratio and stored pixels do not match how you want the video to appear.

Extra windows, keyboard shortcuts, and playlist

Beyond the main trimmer, the app offers a segments list for reviewing ranges as text, file info for technical details about codecs and resolution, logs for troubleshooting, and a help window. Support contact details appear in Help when you need to reach the developer.

Default keyboard shortcuts:

⌘O: open file
Space: play / pause
← / →: jump back / forward
I / O: set in / out point
⌘[ / ⌘]: go to in / out point
S: detect shot at playhead
⇧S: detect all shots in the file
F: find visually similar (from current frame)
G: find person by face (from current frame)
L: toggle loop
M: toggle mute
⌘E: export merged
⇧⌘E: export contact sheet
⌘⌫: clear all segments
⇧⌘P: open playlist
⌥⌘I: file info
⌥⌘L: show logs

All shortcuts are customizable in Settings → Keyboard Shortcuts. The Menus also list the full set of available actions.

The playlist (⇧⌘P) holds multiple files in a queue. You can align the active item with the trimmer, run certain scans across many clips with overall progress, export each item's segments in sequence, or create contact-sheet images (grids of thumbnails) when you need a visual index of footage (or faces across a batch, when that mode is available).

Practical tips

Save work incrementally. Export or duplicate projects before aggressive "replace whole timeline" actions.
Tune scans on a short clip first. Nudge interval and thresholds until hits feel right, then run on the full file.
Combine manual and automatic marks. Use detect shot for quick boundaries, then face or visual search to pull in distant matches.
Expect tiny edge shifts on export. That is normal for lossless cutting and keyframe alignment; it is not a quality problem.
Need help? Use the support form, or open Help in the app and choose Contact support.

← Back to home