Skip to content

roziki-dev/Fingr-proj

Repository files navigation

🦾 FINGR // Hand-AR Interface

FINGR is an interactive, browser-based Augmented Reality (AR) web application powered by Next.js and Google MediaPipe. Step into a TRON-inspired cybernetic environment, where your real hand movements manipulate beautiful, responsive neon digital entities directly on your screen.

🌐 Experience the Live Interface Here: https://fingr-proj.vercel.app/

Fingr AR Preview


⚡ Interactive Mechanics

The system translates your physical gestures into visual AR manipulation using real-time machine learning via your webcam.

1. Neon Fish Spawning (Gesture: Pinch)

Pinch your thumb and index finger together to birth artificial "Neon Fishes" into the canvas.

  • Flocking AI: Fishes use Newtonian physics and specialized boids algorithms to chase the nearest detected hand while systematically avoiding colliding with each other (Separation & Attraction).
  • Smooth Interpolation: Movement and rotational turning is rendered with continuous linear interpolation (Lerp) for buttery 60fps frame animations.
  • Maximum Limit: The simulation organically caps at 20 entities to prevent engine performance degradation.

2. Singularity/Black Hole (Gesture: Pointing Index Finger)

Fully extend your index finger while completely clenching the rest of your fingers toward your palm to command a Black Hole. (Requires maximum of 20 fishes on screen).

  • Hold to Charge: The mechanic features an intuitive "Hold" buffer. Aim your finger rigidly for exactly 2 seconds to watch the destructive aura collect energy in a glowing loading ring.
  • Safe Cancel: Flinch or lower your finger during the charge phase to safely dissipate the anomaly.
  • Devour: Once the neon loading ring connects at 100%, the singularity opens, activating massive spatial gravity that violently sucks in and annihilates all Neon Fishes from the screen.

💻 Technical Architecture

Built aggressively to optimize performance and browser-level stability:

  • Core Framework: Modern Next.js App Router implementation with deeply separated Server & Client Component orchestration.
  • AI Hand Tracking: Utilizing @mediapipe/tasks-vision's WebAssembly graph for accurate, rotation-independent real-world landmark estimation.
  • Performance Optimized: The heavy AI WASM payload is injected fully asynchronously using strictly next/dynamic without Server Side Rendering (ssr: false). It lazy-loads entirely in the background while a futuristic TRON loading screen distracts the user.
  • Security & Privacy First: Implements rigorous React unmount boundaries to kill the device MediaStream tracks instantly upon exit, guaranteeing the webcam hardware shuts off automatically.
  • TRON UI: Visual layers built with standard Tailwind CSS. A responsive parallax 3D floor constructed via deep multi-tiered mouse position tracking matrices and DOM transforms on the home page.

⚙️ Running Locally

  1. Clone this repository to your local machine.
  2. Install the necessary packages:
    npm install
  3. Initialize the development environment:
    npm run dev
  4. Access the portal at http://localhost:3000.

Created by Muhammad Sokhib Roziki.

About

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors