SyncCut / Sync guide
From shared sound to an editing timeline
Synchronization estimates a time shift. It does not pick camera angles or create a finished edit. This guide explains what the numbers mean and how to check them before using your originals.
1. Prepare recordings that actually overlap
Each camera needs audible material from the same event. Keep the original audio and playback speed. Choose the clearest recording with the widest shared coverage as the first file: the current engine compares the other recordings against it. A clip that overlaps only a different camera may not align reliably. File extensions alone do not guarantee that the browser or decoder supports the codecs inside.
2. Upload and inspect the result
Selecting files only creates local previews. Starting synchronization uploads the selected videos for server analysis. Keep the tab open until completion. In multiview, inspect a distinct clap, spoken consonant or other shared transient. Compare more than one moment and listen to one audio track at a time. The quality label summarizes audio similarity; it is not a guarantee of frame-perfect synchronization.
3. Read timeline time and source time separately
The report sets the earliest estimated recording start to zero. A timeline start of 00:00:07.200 means that the beginning of that source belongs 7.2 seconds after the reference begins. It does not mean you should trim 7.2 seconds from that file. Source timestamps count from the beginning of each individual file. The worked example below shows why the two clocks differ.
Worked example: one clap, two recording clocks
Illustrative numbers, not a customer recording or an accuracy benchmark. Camera A runs for 60 seconds; Camera B runs for 40 seconds. The same clap appears at A’s 12.0 seconds and B’s 4.8 seconds.
- 12.0 − 4.8 = 7.2 seconds. Place the start of Camera B at timeline 00:00:07.200, with Camera A starting at zero.
- Their overlap is timeline 00:00:07.200–00:00:47.200. This is A source 00:00:07.200–00:00:47.200 and B source 00:00:00.000–00:00:40.000.
- HH:MM:SS.mmm is elapsed time in milliseconds, not frame timecode. At 30 fps, 7.2 seconds is 216 frames; other frame rates may require rounding in your editor.
4. Place the originals in your editing software
Create a sequence in your editor and import the same, untrimmed originals. Put the reference at the sequence origin, then place each other source start at its TXT timeline start on a separate video track. If the sequence starts at 01:00:00:00, add the reported offsets relative to that origin. TXT is a readable reference, not an XML/EDL project that imports the arrangement automatically. Use your editor’s snapping or audio controls to check the closest representable frame.
5. Decide whether you need TXT or ZIP
TXT uses the original synchronization result and does not include splits or deleted ranges from the SyncCut editor. It needs no second video upload or video render. ZIP renders one aligned MP4 per source; unrecorded and deleted ranges are black and silent, not removed from the shared timeline. For normal multicamera editing, retaining the originals and placing them yourself usually avoids that extra render.
6. Know when a single offset is not enough
Silence, wind, clipping, repeated music and little shared audio can produce ambiguous matches. Do not force an uncertain result: check different shared events or prepare a shorter common section while tracking any trims. If the beginning aligns but the end drifts, recording clocks or changed playback speed may be involved. SyncCut currently estimates fixed start offsets; it does not automatically correct drift or time-stretch recordings.