A free, open-source Premiere Pro extension that brings AI-powered video editing automation, caption generation, audio processing, and visual effects -- all running locally on your machine. This project aims to replace all of the leading paid Premiere extensions and products on the market.
- Adobe Premiere Pro 2023 or later
- Python 3.9+ (python.org)
- FFmpeg (ffmpeg.org) -- must be on PATH
- Windows 10/11 (macOS/Linux: server works, CEP panel is Windows-only)
git clone https://github.com/opencut/opencut.git
cd opencut
Option A -- Double-click installer (recommended):
Run Install.bat as Administrator. It handles everything: FFmpeg check, Python deps, CEP extension deployment, registry keys, and launcher creation.
Option B -- Manual setup:
pip install -e ".[all]"
python -m opencut.serverThen copy extension/com.opencut.panel to your Premiere Pro CEP extensions folder and enable unsigned extensions via the registry key PlayerDebugMode = 1.
Option C -- Cross-platform Python installer:
python install.py- Start the backend: run
Start-OpenCut.batorpython -m opencut.server - In Premiere Pro: Window > Extensions > OpenCut
- Select a clip in your timeline and use the panel
OpenCut v1.0.0 includes 34 processing modules, 116 API routes, and 6 panel tabs covering every major video editing automation task.
| Feature | Description | Engine |
|---|---|---|
| Transcription | Speech-to-text with word-level timestamps | WhisperX / faster-whisper |
| 19 Caption Styles | YouTube Bold, Neon Pop, Cinematic, Netflix, Sports, etc. | Pillow renderer |
| Animated Captions | CapCut-style word-by-word pop, fade, bounce, glow, highlight | Pillow + OpenCV |
| Caption Burn-in | Hard-burn styled captions directly into video | FFmpeg drawtext |
| Speaker Diarization | Identify who's speaking for podcasts | pyannote.audio |
| Filler Word Detection | Auto-detect and remove um, uh, like, you know | WhisperX |
| Translation | Translate captions to 50+ languages | deep-translator |
| Feature | Description | Engine |
|---|---|---|
| Stem Separation | Isolate vocals, drums, bass, other | Demucs (htdemucs) |
| Noise Reduction | AI noise removal + spectral gating | noisereduce / RNNoise |
| Normalization | Loudness targeting (LUFS) | FFmpeg loudnorm |
| Beat Detection | BPM analysis and beat markers | librosa |
| Audio Ducking | Auto-lower music under dialogue | FFmpeg sidechaincompress |
| Pro FX Chain | Compressor, EQ, de-esser, limiter, reverb, stereo width | Pedalboard (Spotify) |
| TTS Voice Generation | Text-to-speech with 100+ voices | Edge-TTS / F5-TTS |
| SFX Generator | Procedural tones, sweeps, impacts | NumPy synthesis |
| AI Music Generation | Text-to-music from prompts | MusicGen (AudioCraft) |
| Feature | Description | Engine |
|---|---|---|
| Stabilization | Deshake / vidstab | FFmpeg |
| Chromakey | Green/blue screen removal + compositing | OpenCV HSV |
| Picture-in-Picture | Overlay PiP with position/scale controls | FFmpeg overlay |
| Blend Modes | 14 blend modes (multiply, screen, overlay, etc.) | FFmpeg blend |
| 34 Transitions | Crossfade, wipe, slide, circle, pixelize, radial, zoom | FFmpeg xfade |
| Particle Effects | Confetti, sparkles, snow, rain, fire, smoke, bubbles | Pillow renderer |
| Title Cards | Animated titles with 6 presets (fade, slide, typewriter, lower third) | FFmpeg drawtext |
| Speed Ramping | Time remapping with ease-in/out curves | FFmpeg setpts |
| Scene Detection | Auto-detect cuts and scene changes | PySceneDetect |
| Film Grain | Adjustable noise overlay | FFmpeg noise |
| Letterbox | Cinema aspect ratio bars | FFmpeg pad |
| Vignette | Adjustable corner darkening | FFmpeg vignette |
| LUT Library | 15 built-in cinematic LUTs + external .cube/.3dl support | FFmpeg lut3d |
| Feature | Description | Engine |
|---|---|---|
| AI Upscaling | 3 tiers: Lanczos, Real-ESRGAN, Video2x | FFmpeg / Real-ESRGAN |
| Background Removal | Remove video backgrounds | rembg (U2-Net) |
| Face Enhancement | Restore/upscale faces | GFPGAN |
| Face Swap | Replace faces with reference image | InsightFace |
| Style Transfer | Neural artistic style transfer | PyTorch models |
| Object/Watermark Removal | Remove logos via delogo or LaMA AI inpainting | FFmpeg / LaMA |
| Color Correction | Exposure, contrast, saturation, temperature, lift/gamma/gain | FFmpeg eq/colorbalance |
| Color Space Conversion | sRGB, Rec.709, Rec.2020, DCI-P3 | FFmpeg colorspace |
| Auto Thumbnails | AI-scored frame extraction for thumbnails | Florence-2 / OpenCV |
| Feature | Description | Engine |
|---|---|---|
| Platform Presets | YouTube, TikTok, Instagram, Twitter, LinkedIn, Podcast | FFmpeg encode |
| Batch Processing | Process multiple clips in parallel | ThreadPool |
| Transcript Export | SRT, VTT, plain text, timestamped | Built-in |
+-----------------------+ HTTP/JSON +-----------------------+
| Premiere Pro CEP | <================> | OpenCut Server |
| Panel (HTML/JS) | localhost:5679 | (Python/Flask) |
| | | |
| 6 tabs, 24 sub-tabs | | 34 core modules |
| Dark theme, 6 color | | 116 API routes |
| schemes | | Async job queue |
+-----------------------+ +-----------+-----------+
|
+-----------+-----------+
| | |
+--+--+ +----+---+ +---+----+
|FFmpeg| |Whisper | |PyTorch |
|OpenCV| |Demucs | |Models |
+-----+ +--------+ +--------+
The server runs entirely on your local machine. No data leaves your computer. No API keys needed for core features.
OpenCut uses a tiered dependency model. Only the core tier is required -- everything else is optional and auto-detected at runtime.
flask, flask-cors, click, rich
pip install opencut[standard]
Adds: faster-whisper, opencv-python-headless, Pillow, numpy, librosa, pydub, noisereduce, deep-translator
pip install opencut[all]
Adds all standard deps plus: whisperx, demucs, pyannote.audio, pedalboard, edge-tts, realesrgan, rembg, gfpgan, insightface, audiocraft, simple-lama-inpainting
pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu121| Variable | Default | Description |
|---|---|---|
OPENCUT_PORT |
5679 |
Server port |
OPENCUT_HOST |
127.0.0.1 |
Bind address |
OPENCUT_OUTPUT_DIR |
Source file dir | Default output directory |
WHISPER_MODELS_DIR |
~/.cache |
Whisper model cache |
6 dark themes switchable from the gear icon: Cyberpunk, Midnight OLED, Catppuccin Mocha, GitHub Dark, Stealth, Ember.
Q: The panel says "Server offline" A: Make sure the backend is running. Check that port 5679 is not blocked.
Q: Transcription is slow
A: Install CUDA-enabled PyTorch and use faster-whisper with a GPU. The tiny model is fastest.
Q: I get "module not found" errors for AI features
A: Most AI features are optional. Install them individually or use pip install opencut[all].
Q: Can I use this without Premiere Pro? A: Yes. The server runs standalone with a REST API. Call any route with curl or build your own frontend.
Q: Does this send data to the cloud? A: No. Everything runs locally. No telemetry, no API keys for core features. Edge-TTS requires internet.
Issues and PRs welcome. The codebase:
opencut/
core/ # 34 processing modules (one per feature area)
server.py # Flask server with 116 routes
cli.py # CLI entry point
extension/
com.opencut.panel/
client/ # CEP panel (index.html, main.js, style.css)
CSXS/ # Extension manifest
MIT License. See LICENSE for details.
Built with FFmpeg, Whisper, Demucs, PyTorch, and many other open-source projects.