Create short-form content packages from long video links with a desktop tool
Paste a link to an interview or podcast source, and the tool prepares platform-specific vertical videos, subtitles, thumbnails, and posting copy in one go. It handles the editing process of cutting long videos. The output is ready for immediate download or upload to various platforms.
A workflow completed with a single link
Users input the source video link and select the target platform. The system automatically generates up to 10 highlight clips according to the platform's aspect ratio and length specifications. Each clip comes with a ZIP distribution package containing subtitles, titles, descriptions, and hashtags. If the results are not satisfactory, users can select specific scenes from the candidate list to generate additional clips.
Automatic optimization by platform
The tool automatically adjusts the video style based on the target platform. Douyin and Xiaohongshu default to an interview format, while TikTok, Reels, and Shorts apply a full-screen podcast format. The camera follows the speaker's face, and scenes without people retain the original screen. A brand ending is included by default but can be disabled in the settings.
Local processing and model choice
Video editing and rendering are performed directly on the user's computer. The AI model used for analysis can be selected from a cloud API or local execution via Ollama or LM Studio. Videos without subtitles can be converted to text using local Whisper or SenseVoice, or by utilizing cloud services. If all processing occurs locally, no data is sent to external servers.
Actual processing time and cost
According to real-world cases measured during development, a 1 hour 43 minute video was converted into 10 clips in approximately 7.5 minutes. A 2 hour 52 minute video took 29.5 minutes. The estimated text model costs for these processes were 0.09 yuan and 0.20 yuan, respectively. In previous versions, processing a 2 hour 6 minute video took 65 minutes, but the new algorithm has significantly reduced processing time.
Things to check before use
This tool works best with podcasts, interviews, lectures, and narration-centric videos. For gameplay or videos with little dialogue, the screen recognition feature must be enabled, and the results must be verified. The software itself is free, but using cloud models requires an API key and incurs costs based on usage. The desktop version supports macOS Apple Silicon and Windows x64, while Linux environments can install it via Docker. As a project maintained by an individual, immediate customer support is not provided.
