Camera permission needed

Allow camera access to try Airpoint.

Scroll Demo

Use the "Scroll Gesture" (Index & Middle fingers together) inside the virtual trackpad.

Section 1: The Beginning

Lorem ipsum dolor sit amet, consectetur adipiscing elit. Sed do eiusmod tempor incididunt ut labore et dolore magna aliqua.

Section 2: Deep Dive

Sed ut perspiciatis unde omnis iste natus error sit voluptatem accusantium doloremque laudantium.

Horizontal Scrolling

Card 1

Card 2

Card 3

Card 4

Card 5

Section 3: Advanced Mechanics

  • Point to move cursor
  • Pinch to click
  • Two fingers together to scroll

Section 4: Motion Control

Hand tracking enables intuitive control without touching any device. The camera captures your hand movements at 30-60 FPS and translates them into precise cursor movements on screen.

Section 5: Gesture Recognition

Machine learning models analyze hand landmarks to detect gestures like pointing, clicking, scrolling, and grabbing in real-time.

Section 6: Smoothing Filters

One-Euro filters reduce jitter while maintaining responsiveness. The filter adapts based on movement speed - more smoothing when still, less lag when moving fast.

Section 7: Sensitivity & Acceleration

Pointer sensitivity multiplies your hand movement. Acceleration adds extra boost when moving quickly, making it easier to cross large distances while maintaining precision for small movements.

Section 8: Distance Compensation

Palm size normalization ensures consistent cursor behavior regardless of your distance from the camera. Move closer or farther away and the sensitivity adjusts automatically.

Section 9: Scroll Mechanics

Scroll velocity is computed from cursor delta with momentum and inertia. Release the scroll gesture and watch the page continue scrolling with natural deceleration.

Section 10: Axis Locking

When scrolling, the system can lock to vertical or horizontal axis based on initial movement direction. This prevents diagonal drift during intentional single-axis scrolling.

Section 11: Click Detection

LSTM neural networks analyze sequences of hand poses to detect click gestures with high accuracy. Hysteresis prevents accidental double-clicks and false positives.

Section 12: Performance

The pipeline runs at camera frame rate with sub-frame latency. GPU acceleration via WebGL keeps inference fast, leaving headroom for smooth rendering at 60 FPS.

Section 13: Conclusion

Keep scrolling to test momentum and edge bounce.

Section 14: The End

You've reached the bottom of the scroll demo! 🎉