A gamer’s streaming setup displays video-editing software arranging gameplay clips and reaction shots.
NeonClipping turns long streams and local MP4s into captioned vertical clips on a Windows PC, without sending the source footage to a clipping service. The free, MIT-licensed project from Ajay K Meena accepts YouTube, Twitch and Kick links, but its most useful feature is what it avoids downloading: in Fast mode, it analyzes the audio first and retrieves video only for the moments it picks. That’s a practical difference when the source is a three-hour stream and the output is a handful of Shorts.

A gamer’s streaming setup displays video-editing software arranging gameplay clips and reaction shots. What the 97-second benchmark actually measures​

The project’s README reports that Fast mode produced three clips from an 85-minute YouTube stream in 97 seconds on a machine with an RTX 2060 and 16 CPU threads. The source already had YouTube captions, so the app didn’t need to transcribe it. The same machine took 136 seconds to produce three clips from a three-hour Twitch recording using local speech recognition. These are maintainer measurements, not independent benchmarks, and the two runs used different sources. They show what the workflow can do, not what your PC will do.

Fast mode is the default for videos longer than 10 minutes. It looks for sudden changes in loudness and other audio cues, then uses the transcript to help identify promising cuts. If you enable Ollama, a local model can score those candidates and write titles and opening hooks. Ollama is optional: the README says the app can still select clips using its audio and motion scores without it. The default vertical render is 1080 × 1920, with captions burned into the MP4.

Deep mode is the better comparison to run on your own footage. It downloads the full video and adds motion and scene-cut analysis. That costs more bandwidth and storage for a remote source, but it gives the app a chance to surface something Fast mode might miss: a quiet visual gag, for instance, or a decisive play the streamer barely reacts to. Neither mode knows whether the clip makes sense out of context.

The Windows installer is unsigned​

The project documents support for Windows 10 and 11, Python 3.10–3.13 and FFmpeg. An NVIDIA GPU is optional, though it can accelerate encoding and transcription. According to the project site, NeonClipping-Setup.bat checks for missing dependencies, installs the app under %LOCALAPPDATA%\NeonClipping, creates a desktop shortcut and opens the interface in your browser. Setup needs an internet connection, and remote videos still have to be fetched from their platforms. Local processing is not the same as offline operation. The project also provides a Mac/Linux installer, but the maintainer says it has not been tried on a physical Mac.

The project explicitly says its batch file is not code-signed. Microsoft explains that Defender SmartScreen checks the reputation of downloaded files; on Windows 11, Smart App Control can separately block unsigned files that lack positive reputation. If you want to evaluate NeonClipping, get the batch file from the official project site or repository and open it in Notepad before executing it. If you can’t verify what it downloads and runs, don’t proceed. On a managed PC, use your organization’s software approval process rather than weakening Windows protections for a video editor.

Try it on a stream you know​

Start with a recording you own and can edit. Run Fast mode and note whether it reused captions or had to transcribe the source; otherwise, the stopwatch comparison tells you little. Then run Deep mode on the same footage and compare which moments made the cut. Watch the exports on a phone: check that the opening line makes sense, the captions match what was said, and the facecam crop hasn’t cut off the action. NeonClipping offers 11 layouts, including stacked facecam and gameplay. Keep the original source files in data/ if you want to change the framing or caption style later without fetching the footage again.

The README also documents chat-spike clipping for Twitch and YouTube Live, plus an optional PIN-protected remote on the same Wi-Fi. The phone controls the session; the PC still processes the video. The normal interface listens only on the PC itself, so leave phone access off if you don’t need it. And before publishing any clip, confirm you have permission to use the footage and audio. Keeping the edit local doesn’t grant reuse rights.

The useful test isn’t whether NeonClipping can beat 97 seconds. It’s whether the clips it finds save you more editing time than you spend correcting captions, reframing shots and throwing out false starts. For long recordings with usable captions, Fast mode gives Windows creators a concrete way to find out without uploading an entire stream to a clipping service.

 

References

  1. NeonClipping: Free Local YouTube Shorts Clipper (Nov 2026) - explainx.ai explainx.ai 2026-11-22T00:00:00+00:00
  2. GitHub - iamajmeena/neonclipping: Paste a link, get captioned 9:16 clips in ~2 min. Free, local, open-source clipping engine (audio-first, YouTube captions, local LLM). · GitHub github.com
  3. neonclipping by iamajmeena - GitHub Repository Analysis | GitDiscover gitdiscover.org