# Chopify

> Chopify is a free, MIT-licensed local video clipper with a browser dashboard and Python CLI. It turns recordings or supported URLs into reviewed, captioned MP4 clips and SRT subtitles for Shorts, Reels, and TikTok. Processing runs on the user's computer; no hosted inference, subscription, API key, or watermark is required.

## Documentation

- [Installation and overview](https://github.com/mehbul/chopify/blob/main/README.md): requirements, quickstart, examples, features, and limitations.
- [Workflow guide](https://github.com/mehbul/chopify/blob/main/docs/WORKFLOWS.md): local input, batch, resume, captions, branding, semantic search, split-screen, and project files.
- [Agent and API reference](https://github.com/mehbul/chopify/blob/main/docs/AGENT-API.md): CLI, Python interfaces, JSON formats, and authenticated loopback HTTP endpoints.
- [Troubleshooting](https://github.com/mehbul/chopify/blob/main/docs/TROUBLESHOOTING.md): ffmpeg, models, captions, CUDA, and failed jobs.
- [Product facts](https://github.com/mehbul/chopify/blob/main/docs/product.json): machine-readable capabilities, requirements, licensing, and exclusions.
- [Changelog](https://github.com/mehbul/chopify/blob/main/CHANGELOG.md): changes and release history.

## Product facts

- `python dashboard.py` opens Studio at `http://127.0.0.1:8765`; `python chopify.py SOURCE` runs the CLI.
- Inputs: local MP4, MOV, MKV, WebM, M4V, AVI, or yt-dlp-supported HTTP(S) URLs.
- Pipeline: source -> faster-whisper timestamps -> heuristic/local LLM/keyword/semantic selection -> review -> ffmpeg.
- Features: resumable projects, serial batch queue, editable captions and clip ranges, SRT, JSON brand presets with logo, two-panel podcast framing, filler/silence removal, loudness normalization, single-frame face tracking.
- Outputs: 16:9, 9:16, or 1:1 H.264 MP4; AAC if source audio exists; SRT for drafts and final clips; poster and metadata JSON for final clips.
- Semantic search uses optional local Ollama embeddings at `127.0.0.1:11434`, default `nomic-embed-text`. LLM selection uses a separate optional local chat model.
- Requirements: Python 3.11+, ffmpeg/ffprobe with libx264/AAC/libass, requirements.txt. Windows is primary; CI also tests Linux.
- License/price: MIT; no subscription or paid inference API. Hardware, storage and setup downloads are supplied by the user.

## Limitations

- Models may need initial downloads; URL input needs internet. Local-file processing can run offline after setup.
- Scores are editing suggestions, not virality predictions. Review boundaries and captions.
- Split-screen uses fixed adjustable crop centers, not speaker diarization or two-face tracking.
- Heuristic cue lists are English-oriented; transcription and punctuation vary by language.
- No social publishing/scheduling, cloud collaboration, hosted processing, translation, bilingual captions, or dubbing.
- Studio binds to loopback for one trusted local user, not public/multi-tenant hosting.

## Optional

- [Repository](https://github.com/mehbul/chopify)
- [License](https://github.com/mehbul/chopify/blob/main/LICENSE)
- [Documentation website](https://mehbul.github.io/chopify/)

Updated: 2026-10-23. This file is a factual navigation aid, not a search-engine requirement or ranking signal.