Deep Frame is a high-performance, real-time frame interpolation engine and overlay for Windows. It leverages hardware-accelerated capture and AI-driven motion estimation to enhance visual fluidity in any windowed application.
- Low-Latency Capture: Built on DXGI Desktop Duplication for minimal overhead capture.
- AI Frame Generation: Integrated ONNX Runtime support for motion-compensated frame interpolation.
- Direct3D 11 Overlay: Seamless, high-performance presentation using a dedicated D3D11 overlay layer.
- Native NAPI Bindings: A robust C++ core bridged to a modern Electron/React interface for a seamless developer and user experience.
- Flexible Modes: Customizable interpolation profiles (Fast, Balanced, Quality) to match hardware capabilities.
core/: The high-performance C++ engine.capture/: DXGI-based screen and window capture logic.inference/: AI model execution and tensor processing via ONNX.pipeline/: Asynchronous processing pipeline and ring buffering.present/: D3D11/D2D1 overlay and presentation layer.napi/: Node.js native addon bindings.
ui/: Modern React-based dashboard for control and monitoring.
- Node.js: v18 or later.
- Build Tools: Visual Studio 2022 with C++ desktop development workload.
- CMake: v3.20 or later.
- ONNX Runtime: (Optional) Required for AI-based interpolation.
- Clone the repository.
- Install UI dependencies:
cd ui npm install - Build the native engine:
cd core/napi npm install npm run build
Run the UI and the Electron host in development mode:
cd ui
npm run devDeep Frame operates by capturing the target window's backbuffer, processing it through an interpolation pipeline (either simple blending or AI-driven motion estimation), and presenting the generated frames via a transparent overlay window. The architecture is designed for maximum throughput, utilizing asynchronous processing stages and thread-safe ring buffers to minimize impact on the target application's performance.
MIT