AMVerge

Fast desktop scene-splitting software for editors.
AMVerge helps editors turn long videos into usable clips quickly. Import a video, split it into scenes, preview results instantly, merge false cuts if needed, and export only what you want.

---

Features

• Fast keyframe-based scene splitting
• Instant clip previewing
• Batch clip export
• Merge selected clips into seamless scenes
• Folder + episode organization system
• HEVC / H.265 support detection
• Proxy preview workflows for unsupported codecs
• Resizable desktop-style interface
• Customizable themes / UI settings

---

How It Works

Frontend (React + TypeScript)
          ↓
Desktop Layer (Tauri + Rust)
          ↓
Backend (Python)
          ↓
FFmpeg / FFprobe / PyAV

Frontend

Handles:

• importing videos
• previewing clips
• selection workflows
• exporting clips
• sidebar organization
• UI settings

Tauri Layer

Handles:

• desktop packaging
• secure command bridge
• filesystem access
• progress events

Backend

Handles:

• keyframe extraction
• clip splitting
• thumbnail generation
• export / merge workflows
• codec support helpers

---

Why Keyframes?

Older scene detection experiments used frame-by-frame analysis.

The current version uses keyframes because it is:

• much faster
• simpler
• practical for real editors
• no full re-encode on import
• easy to correct with merge tools afterward

---

Repository Structure

AMVerge/
│
├── backend/
│   ├── app.py
│   ├── requirements.txt
│   ├── backend_script.spec
│   ├── bin/
│   │   ├── ffmpeg.exe
│   │   └── ffprobe.exe
│   ├── utils/
│   │   ├── binaries.py
│   │   ├── progress.py
│   │   ├── keyframes.py
│   │   ├── video_utils.py
│   │   └── hevc_script.py
│   ├── deprecated/
│   ├── test_scripts/
│   ├── build/
│   └── dist/
│
├── frontend/
│   ├── src/
│   │   ├── App.tsx
│   │   ├── components/
│   │   │   ├── clipsGrid/
│   │   │   ├── previewPanel/
│   │   │   └── sidebar/
│   │   ├── hooks/
│   │   ├── pages/
│   │   ├── styles/
│   │   ├── types/
│   │   └── utils/
│   ├── src-tauri/
│   ├── public/
│   ├── package.json
│   └── ...
│
├── README.md
└── documentation.md

---

Getting Started

Requirements

Install:

• Python 3.10+
• Node.js + npm
• Rust
• Windows (current main target)

---

Backend Setup

cd backend
python -m venv venv
venv\Scripts\pip install -r requirements.txt

---

Frontend Setup

cd frontend
npm install
npm run tauri dev

---

Build Desktop App

npm run tauri build

---

Build Python Sidecar Manually (Optional)

cd backend
pyinstaller backend_script.spec

---

Current Focus

• More export formats
• Export using original codec/settings by default
• Dual-stream export support
• Quality slider for export bitrate
• Hover audio playback (toggleable)
• Clip timestamps from original episode
• Optional timestamps shown under grid clips
• Original aspect ratio clip cells
• Better merge-export stability
• Fix occasional merged clip stutter
• Performance optimization for freezes during heavy exports
• Ability to combine clips from multiple episodes into one compilation
• Setting to move the episode list location inside Menu > Settings

License

AMVerge is licensed under the GNU GPL v3.0.

Any derivative work must also be open-source under the same license.