Clipping review · updated 2026-08-16
OpusClip
Use it only when the recording already exists. It finds the moment. It does not invent the video.
8/10 editorial fit for this job
Find the 30 seconds that already exist in a long take.
Best for
A 20–40 minute lecture or podcast that needs a week of Shorts without a hand recut.
- Operators sitting on long recordings who need daily Shorts output.
- Channels that already filmed the take and only need the 30-second sentence.
- People who will read the captions once before they publish.
Skip if
People who have no footage. That is a generator job, not a clipper job.
- You do not have a long recording. Open InVideo instead.
- You need word-level cleanup more than moment-finding. That is Descript.
- The footage is slides-only with no speaker. The 9:16 crop will fail.
Review
OpusClip is the clipper on this desk because the lecture channels already have long takes. The expensive part of that workflow is not rendering. It is finding the sentence that already works.
The product is a hook finder plus a 9:16 crop plus captions. That is enough when the source is a talking head. It is not enough when the source is a slide deck, a screen recording with no face, or nothing at all.
We upload one lecture, keep the clips where the sentence is complete, and throw the rest away. Volume from a clipper is only useful if the clip can stand alone. A mid-sentence crop is not a Short.
If you do not have footage, stop. Do not buy a clipper to solve a generation problem. That is the InVideo page.
Where we use it. Local AI clipping Shorts from long lecture takes. Clip Lab is the dedicated surface when that channel is bound.
What the channels prove. The 40-minute lecture cut is the proof case: hook finder, speaker crop, captions, then a human pass to drop incomplete sentences.
What works
- Hook finder ranks moments so you are not scrubbing a 40-minute file.
- Speaker crop gets most talking-head clips into 9:16.
- Captions come with the export. Still check them.
- Free / trial is enough to prove it on one recording.
What does not
- It cannot invent a video. No source, no clips.
- Captions will be wrong on jargon. Wrong words kill trust.
- Slides-only footage breaks the speaker crop.
- Pro is only worth it after the first 8 clips actually convert.
How we actually run it
- Upload the take. YouTube URL or MP4 with clear speech. One 20–40 minute lecture is the right test.
- Keep complete sentences. Throw away clips that start mid-thought. Five to eight keepers beat twenty fragments.
- Read every caption. Fix product names, model names, and numbers. Then export.
- Pin the clipping page. Description goes to /tools/opusclip, not to a voice tool.
Models and engines
- Hook finder. Ranks moments by visual change + speech energy. Not a full editor.
- Reframe / speaker crop. Keeps the talking head in 9:16. Fails on slides-only footage.
- Caption pack. Auto captions you still have to read once. Wrong words kill trust.
- Opus agent extras. Batch more cuts. Only useful after the first 8 clips convert.
What you actually pay for
- Free / trial. Prove it on one 20–40 min recording.
- Starter. A few videos per week. Enough for one channel.
- Pro. Daily clip factory. Only if you already have source footage.
Trial proves the job on one recording. Starter is one channel. Pro is a daily clip factory. Do not buy Pro to generate footage you do not have.
Against the other software in this job
| Check | OpusClip | CapCut | Descript | InVideo |
|---|---|---|---|---|
| Finds moments for you | Yes, core job | Manual | Limited | Manual / templates |
| 9:16 reframe | Automatic speaker crop | Manual | Good | Templates |
| Captions | Included, check them | Excellent | Good | Included |
| Needs source footage | Yes | Yes | Yes | No — generates |
Questions we get
OpusClip or Descript?
OpusClip finds the moment. Descript edits the words. Use both if you already recorded. Do not use either to generate a new video.
Can I skip CapCut?
Sometimes. If the crop and captions are clean, export. CapCut is the finish pass when the hook needs a title card.
Why not InVideo?
InVideo starts from an idea. OpusClip starts from a file. Mixing those jobs is how people waste a month.