VisionCamera V5

VisionCamera V5 is the next major milestone for react-native-vision-camera. It is rewritten from the ground up using Nitro Modules, and offers both a declarative API as well as an imperative API for maximum control.

Some new features;

• ⚡️ Faster performance (thanks to Nitro!)
  • An in-memory Photo type for even faster capture (convertable entirely in-memory to a react-native-nitro-image type)
• 📐 Native support for Coordinate System conversions (e.g. Frame -> Preview coordinates, or Barcode -> Preview - including Skia Preview!)
• 🏗️ Improved Frame Processor architecture
  • Native Frame Processor Plugins are now just Nitro Modules, as those work on every Worklet Runtime out-of-the-box
  • Native Frame Processor Plugins are now more powerful, allowing extra typed parameters, return values, and can even return a Frame again
• 🎨 New Skia Frame Processors (with react-native-worklets and Reanimated SharedValue support)
  • Focus support thanks to new coordinate system foundation
• ⚙️ Full manual Focus/Exposure/White-Balance controls
• 🎞️ Multi-Camera sessions (front and back Camera at the same time)
• 🌊 Depth Data Streaming (and Frame + Depth Output Synchronization!)
• 🧪 Barcode Scanner is now unified - Android uses MLKit, iOS uses either native Object Output or MLKit too (new!)
  • It can scan using a declarative output, or the imperative Frame Processor Plugin

V5 is currently sponsor-only, meaning if you 💖 sponsor me on GitHub ($20+) 💖 you get access to the private repo. You'll be able to test it early, and provide feedback or feature requests via GitHub issues in the new private repository.