Removes backgrounds from images in real-time, just feel like zoom. Works with webcams, video files, or any texture. Uses a neural network running on your GPU.Pre-requirments:Unity SentisReal-time human segmentation and background removal powered by neural network. This plugin automatically separates people from backgrounds in real-time, producing high-quality alpha mattes suitable for compositing, virtual backgrounds, AR applications, and interactive experiences.What's included:Complete source code (C# and compute shaders)Pre-trained model (26MB)Three working demo scenes (webcam, video, image)Ready-to-use UI components and prefabsComprehensive documentation and API referenceCustomization options:Adjustable inference resolution for quality/performance balanceConfigurable threshold for alpha matte refinementCustom shader support for background replacementFull access to intermediate processing stagesExtensible architecture for custom modelsSuitable for:Video chat applications - Virtual backgrounds, portrait modeLive streaming - Real-time compositing and effectsAR/VR projects - Person isolation and mixed realityGame development - Cutscenes with live actors, motion capture visualizationContent creation tools - Video editors, photo booths, filtersInteractive installations - Museums, exhibitions, marketing displaysEducation - Virtual classrooms, green screen alternativesFitness apps - Pose tracking with background removalCore Functionality:✓ Real-time human segmentation using MODNet architecture✓ GPU-accelerated inference via Unity Sentis✓ High-quality edge preservation (hair, fine details)✓ Single-pass processing with minimal latency✓ Automatic resolution scaling and alignment✓ Alpha matte output for flexible compositingInput Support:✓ WebCamTexture (real-time camera)✓ VideoPlayer (video files, streaming)✓ Texture2D (static images)✓ RenderTexture (camera renders, custom sources)✓ Any Texture-compatible formatEase of Use:✓ Two integration methods: component-based or direct API✓ Drag-and-drop prefabs for quick setup✓ Inspector-friendly configuration✓ Extensive code examples and documentation✓ Debug mode for troubleshootingPlatform Support:✓ Windows (DirectX 11/12)✓ macOS (Metal)✓ Linux (Vulkan)✓ Android (Vulkan, OpenGL ES 3.2+)✓ iOS (Metal)LimitationsDesigned for human subjects (not general object segmentation)Requires GPU with compute shader support (Shader Model 5.0+)Performance scales with resolution (higher = slower)Single-threaded (must run on main Unity thread)Works best with clear human subjects in frameThe cover and card image is created using ai.
