Batch Video Clipping: Turning a Backlog of Recordings Into a Month of Shorts
Batch video clipping works when the sources belong together. How to group recordings, write one brief, pick a clip count and review dozens of results without losing a week.
Say a training team has twelve recorded sessions from last quarter sitting in a shared folder, ten of them about an hour long and two about half an hour. Everyone agrees they hold good material for short clips. Nobody has time to open each file, scrub through it and cut three or four moments by hand, so the folder stays untouched. That is the situation batch video clipping is meant for: the hard part is not cutting one clip, it is getting through the pile.
When batch video clipping makes sense
Batching pays off when the recordings are alike. Twelve sessions in the same series, a season of podcast episodes or a month of live streams share a format, an audience and a reason to be clipped, so one description of what you want fits all of them. A mixed folder of a webinar, a product demo and a wedding toast does not. If the instructions you would give for one file are different from the next, those files belong in separate jobs.
It also pays off only when the clips can come from what people say. ToClips transcribes each upload with word-level timing and chooses moments from the transcript. Sessions where someone explains, argues or answers questions clip well. A silent screen recording or footage where the point is purely visual will not produce useful candidates, however many of them you add.
Setting up one job for many sources
In the ToClips workspace a single job takes up to 20 source files, with per-file size subject to the upload service’s current limit. Video formats include MP4, MOV, WebM and MKV. Audio files are accepted for transcription, but rendered clips need a video track, so a podcast that exists only as audio will not come back as video. Files have to be uploaded; there is no importing from a link.
Then you write the brief once, in plain language and up to 4,000 characters. For the training sessions it might be "one complete, practical tip per clip that makes sense without the rest of the session, 30 to 60 seconds, no inside jokes". That sentence does more work in a batch than in a single upload, because it is applied to every source at the same time. Choose the transcription language, Chinese or English, or let it be detected. Pick a platform preset (generic, Xiaohongshu, Douyin or YouTube Shorts) and one or more of the five aspect ratios.
Choosing how many clips to ask for
The target count runs from 1 to 200 per job, and with several sources ToClips allocates candidate quotas roughly in proportion to each file's length, then removes heavily overlapping candidates and ranks the combined results. The final number from each source can differ, and a job can finish below its target. In the example, the twelve sessions add up to 660 minutes. A target of 66 works out at a candidate quota of about six per hour-long session and three per half-hour session, before selection and deduplication.
Higher is not automatically better. Each clip is rendered in every aspect ratio you select, so 66 clips in both 9:16 and 1:1 means 132 video files to look through. If you would rather post a few strong clips than many average ones, a lower target is the simpler filter. It is easier to run a second job later than to discard half of a large one.
Reviewing dozens of results without watching everything
Every clip arrives with a title, an opening hook line, a caption with hashtags, cover text, a JPG cover and a subtitle file. Download the zip and open video-result.json: it records each clip’s source time range, selection reason and ranking score. The on-screen gallery does not display all those fields. The score helps order candidates; it is not a calibrated confidence probability. Skim the reasons first; a reason that quotes a weak line usually means a weak clip, and you can skip it without pressing play.
Then watch the top of the list properly. Look for cuts that start a beat late, end mid-sentence or lean on something said ten minutes earlier. In batch mode the transcript cannot be edited, so a misheard name is fixed clip by clip: load the original recording in a single-clip editor project, set the source start and end times, add the corrected subtitle cues and render again. Those edits are saved with the project.
When you are done, download clips one by one or the whole job as a zip with the clips, subtitle files, covers and a JSON manifest that records the source, timecode and reason for each clip. That manifest is handy if someone else will schedule the posts, because it tells them which session each clip came from.
Keep each job small enough to check
A batch that produces more than you can review is not saving time. One job per series or per month is usually easier to manage than one enormous job for everything, and a clip count close to what you will actually post keeps the review short. If you want to see what finished output looks like first, the public sample has rendered clips and their files to download, without spending credits.
Related tools and examples: Long video to shorts · AI video clipper.
What this can't do yet
Batch clipping needs an EndTo account and uses credits. Sources must be uploaded as files, there is no link import, and one brief covers the whole job. The transcript cannot be edited in batch mode, captions are not translated, speakers are not labelled and nothing is published automatically. Reframing is a centre crop with no speaker tracking. ToClips has only five real tools: AI video clipper, auto captions, reframe video, podcast clip generator and video-to-shorts — no full non-linear video editor. Batch mode caps at 200 clips per task and a fixed set of five aspect ratios (9:16, 3:4, 1:1, 16:9, 4:5), not arbitrary custom ratios. To refine a batch result, open a separate single-clip editor project and load the original recording, then set the source time range from the downloaded manifest and enter or correct the subtitle cues. The batch gallery does not automatically import a clip and its captions into that editor.
Frequently asked questions
How many videos can I clip in one batch?
One job takes up to 20 source files and a target of up to 200 clips. Per-file size is subject to the current upload service limit. The final clip count can be lower if there are too few usable, non-overlapping candidates; split larger backlogs into separate jobs.
How are clips shared out between the sources?
Candidate quotas are initially based on each source’s share of the total duration. The candidates are then deduplicated and ranked together, so the final number of clips from a source is not guaranteed to match that proportion.
Can I use different instructions for each video in a batch?
No. One brief applies to every source in the job. If two recordings need different instructions, such as a different clip length or focus, put them in separate jobs.
Can I fix transcription mistakes before the clips are rendered?
Not within the batch workflow. For a corrected version, load the original recording in a single-clip editor project, set the desired source time range and add the corrected subtitle cues before rendering. The gallery does not import these into the editor automatically.
Does batch clipping post the clips for me?
No. ToClips gives you the clips, captions, covers, titles and a manifest to download, singly or as a zip. Publishing and scheduling happen on each platform or in your own scheduling tool.