Installation & Setup

System Requirements

Component	Minimum	Recommended
OS	Windows 10 64-bit	Windows 11
RAM	8 GB	16 GB+
Storage	5 GB for the application + bundled AI models	NVMe SSD recommended

Note

No dedicated GPU required. Onset Engine runs on your CPU - including 5-year-old laptops. An NVIDIA GPU with CUDA (GTX 16-series / RTX 20-series or newer, including RTX 50-series) will accelerate AI analysis, and any NVENC-capable NVIDIA GPU enables hardware-encoded exports. See System Requirements for detailed performance expectations.

Windows only

Onset Engine is currently available for Windows only. There is no macOS or Linux version yet. On a Mac? Get notified when the Mac version is ready.

Installation

1. Download

Download Onset Engine from onsetengine.com/pricing. You’ll receive a download link and a license key via email.

2. Install

Run the installer you downloaded, OnsetEngine-Setup.exe. If Windows warns that it protected your PC, choose More info, then Run anyway. Click Download: it downloads the latest version of Onset Engine (about 2 GB), checks it, and starts the setup. Keep the window open until the setup appears.

If you downloaded the full installer instead, it is named for its version, for example OnsetEngine-v1.1.2-Setup.exe, and starts the setup straight away.

Is the installer safe?

Every Onset Engine installer is signed by Onset SAS: right-click it, choose Properties and check the Digital Signatures tab. While setup runs, Task Manager and antivirus software may show it as OnsetEngine-Setup.tmp or OnsetEngine-Setup-Full.exe. See Is the installer safe? and The installer seems stuck.

The application installs to your chosen location:

C:\Program Files\Onset Engine\
├── OnsetEngine.exe    # Main application
├── ffmpeg/            # Bundled FFmpeg (ffmpeg, ffprobe, ffplay)
├── mpv/               # Bundled MPV (mpv.exe, libmpv-2.dll)
├── drivers/           # JSON driver files + examples
├── music/             # Starter music tracks
└── _internal/         # Bundled dependencies (do not modify)

FFmpeg, MPV, and all AI models are bundled - no separate downloads or PATH configuration needed.

3. License Activation

On first launch, Onset Engine will prompt for your license key:

1. Click the tier badge (e.g., “Demo”) in the upper-right corner of the application.
2. Paste your license key into the input field and click Activate.

This makes a single HTTPS call to validate your purchase. After that, Onset Engine works offline. When you’re online, it re-checks the license in the background at most once a day.

The first time you add footage, Onset Engine also downloads its visual analysis model once (about 300 MB, or about 860 MB for High Quality). The download starts as soon as you choose your footage, so it is often done by the time the ingest begins, and its progress is shown in the status bar. After that, it loads the model from your computer.

Tip

The free Demo works fully without a license key - just launch and start editing. Activate a key when you’re ready to unlock Core or Studio features.

First Ingest

Once installed, build your clip library:

1. Launch OnsetEngine.exe
2. Click 📥 INGEST in the asset bin toolbar (left panel)
3. Add your footage, in any combination:
   • + Add Folder - every video in a folder, including its subfolders
   • + Subfolders… - pick a folder, then tick which of its subfolders to add
   • + Add Files… - individual videos, from any folders. They stay where they are.
4. Optionally choose Add to collection to group these videos (see Collections)
5. Under Analysis, choose Standard (faster) or High Quality (better clip matching). Keep one model for your whole library: once your library has clips, the dialog starts on the model they use and asks before you mix the two.
6. On Core and Studio, Don’t make copies of clips (under More options) is on by default, so your footage is read in place instead of copied - see the tip below
7. Click 📥 Add to library. The bar above the timeline shows each video as it is analyzed and about how long is left, and the ingest ends by saying how many videos were added
8. Your library is now searchable and ready for rendering

Tip

Virtual / Pointer-Only is the default on Core and Studio, and is the mode you want: it indexes footage without copying files, ingests around 3x faster, and lets the editor hold a shot longer when the music calls for it - something a physically cut clip cannot do, because it cannot be read past its own end. The Demo tier cuts physical clips instead. See Physical vs Virtual Ingest.

Configuration

Global settings can be adjusted via ⚙️ Preferences in the application header. Key settings include:

• Default quality - Render quality tier: high (master) or draft (preview)
• FPS - Output frame rate (30 or 60)
• Output directory - Where rendered videos are saved
• Jamendo API key - For free music discovery (optional)

See the full configuration reference in the Configuration guide.

Related Articles

• FAQ & Common Fixes
• Autopilot