Building One of the World's Largest Interactive Data Visualisation Experiences
summary
Built VizInteract on Android-x86 by modifying the kernel to add driver support for a PQ Labs IR overlay and enable 12 simultaneous touch points on an 80‑inch 4K panel. Uses Canvas with custom gesture detector, per-shape draw routines, region quick-reject and combined Path batching to hit interactive frame rates; avoids GPU bitmap scaling and heavy string/decimal parsing. Implements undo as an operation log, tunes driver fuzz (≈10) for noise, adjusts velocity thresholds, and reports 3 ms canvas lock on x86.
AI-written summary · about
chapters (16)
- 0:00Introduction and background
- 3:25Demo of Vizentra tool
- 5:11Challenges with charting libraries
- 6:54Canvas architecture and learnings
- 8:59Implementing undo operations
- 11:27Event handling and transformation matrix
- 12:52Canvas scaling and bitmap tips
- 14:17Performance optimizations
- 17:49Data loading with SurfaceView
- 20:21Running on large touch displays
- 23:55Kernel module hacking for touch
- 28:11Touch input delays and fixes
- 31:02Velocity tracking on large screens
- 33:49UX adjustments for large displays
- 35:13Final demo and collaboration use case
- 36:38Q&A and closing remarks
related talks
37mAndroid Canvas: Building one of the worlds largest Interactive DataViz exp.
13mFrom Dashboard to Driver Display
37mAndroid Industrial: Real-time data visualization
46mExplorations in Interactive Visual Analytics
40mAndroid Large Screen:How your app can power a workstation
19mInvestigating the Use of Spatial Interaction for 3D Data Visualization on Mobile Devices
01h 01mAn Android Retrospective
42mInteractive Data Analysis: Visualization and Beyond