The Flutter Journey of Building a Live Streaming App
summary
Built a Flutter live‑streaming app that uses AWS WebRTC for sub‑second to several‑second latency and native SDKs for capture/playback, solves heavy CPU and heat by merging game and camera frames into one native composite instead of two Flutter surfaces. iOS uses Metal MSL shader via AI to composite two CVPixelBuffers, Android uses a single ConstraintLayout platform view. Measured CPU drop 4.74 → 0.91 (relative), device surface temp reduced >40°C to mid‑30s, but energy impact remains high and maintenance cost of native GPU code is nontrivial.
AI-written summary · about
chapters (22)
- 0:00Live streaming app project
- 1:25Introduction
- 2:47Sports live streaming app
- 4:11AWS IVS and WebRTC streaming
- 5:36Viewing and low-latency
- 7:00Native SDKs with Flutter UI
- 9:50Platform view texture performance
- 10:54Performance metrics tools
- 14:05iOS profiling tools
- 15:32Android profiling challenges
- 16:14Viewing screen optimizations
- 17:59Streaming screen optimizations
- 18:46Breaking performance bottlenecks
- 19:50iOS video compositing
- 21:56Metal video compositing
- 24:11iOS Metal results
- 25:41Android platform view issues
- 27:26Android rendering bugs fix
- 28:45Android widget ghosting fix
- 30:14Performance results and feedback
- 31:20Further optimizations and plans
- 33:07Conclusion
related talks
28mAMA with Brian Egan and Filip Hracek
49mWhat's Awesome about Flutter
41mWhy is Flutter the mobile SDK developers deserve...
49mFirst Flutter, then Fly: Build Reactive & Performant Cross-Platform Apps
40mMastering Flutter Web
35mBuilding Modern Cross-Platform Apps with Flutter
43mIntroduction to Flutter
24mKeynote: Flutter