// the find
margelo/react-native-vision-camera
📸 A powerful, high-performance React Native Camera library.
VisionCamera is the go-to native camera library for React Native when you need more than snap-a-photo: frame processors let you run JS worklets against every camera frame at native speed via JSI, so it's built for real-time barcode scanning, face/object detection, or AI pipelines rather than simple photo capture. It's for teams already comfortable with native builds (pod install, Gradle) who need control over FPS, resolution, multi-camera zoom, and HDR that higher-level wrappers don't expose.
Frame processors run as JS worklets directly against camera frames over JSI, avoiding the bridge serialization overhead that kills real-time use cases in older RN camera libs. The native control surface is unusually deep for this category — fish-eye multi-camera zoom, FPS up to 240, HDR/night mode, and a custom C++/GPU (Metal/Vulkan) frame resizer are all exposed, not just photo/video basics. Testing infrastructure is well beyond typical RN library norms: a dedicated harness app with per-feature test suites actually runs on real hardware via AWS Device Farm in CI, not just emulator checks.
No Expo Go path — it's native-linked only, so you're building a custom dev client and eating pod install / Gradle rebuild cycles from day one. Smooth zoom and some other features pull in react-native-reanimated as a peer dependency, which means you're now tracking two native libraries' version compatibility instead of one. The project already broke compatibility once (V4 is archived with its own frozen docs site now that V5 is current), so expect real migration work on future majors, not just a changelog skim. Writing anything beyond the bundled frame processors (a custom detector, say) means dropping into C++/Swift/Kotlin — the JS worklet story has a hard ceiling the moment stock functionality doesn't cover your use case.