Forem: Sushan Shetty

🌍 Deep-Time Mirror: An AI Lens into Our Ecological Soul For Earth Day Edition.

Sushan Shetty — Fri, 17 Apr 2026 16:59:49 +0000

This is a submission for Weekend Challenge: Earth Day Edition (April 22)

What I Built

I built the Deep-Time Mirror, an immersive, AI-powered "temporal lens" that allows users to witness the environmental evolution of any location on Earth.

The goal was to move beyond cold, flat data and create an emotional connection with the land. The Mirror takes you through three distinct epochs:

The Deep Past (~10,000 BCE): A cinematic look at the primeval ecosystem before modern intervention.
The Present (The Anthropocene): A gritty, real-time look at current urban and climate impacts, including a "Recent Pulse" with local environmental events from the last 7 days.
The Restored Future (~2100 CE): A hopeful, Solar-punk vision where humanity has successfully restored biodiversity and integrated technology with nature.

🌱 The Earth Auditor (Gaia Corp)

Beyond observation, I built a personal accountability system:

Users “confess” their daily habits to a brutally honest AI Earth.
The system generates a Planetary Stability Score (0–100).
It breaks down impact across:
Energy ⚡
Transport 🚗
Waste ♻️
Water 💧
Diet 🍽️

All of this is visualized through an interactive radar chart, turning abstract habits into tangible environmental impact.

Demo

Experience the Mirror live here:

👉 Deep-Time Mirror Live Demo

Code

The full source code and documentation can be found here:

sushanshetty2601-pixel / deep-time-mirror

🌍 An AI-powered 3D cinematic portal into Earth's ecological evolution. Visualize the primeval past, the modern Anthropocene, and a solar-punk future for any location. Built with Gemini AI, React, and Three.js for the Earth Day 2026 Challenge.

🌍 Deep-Time Mirror

Piercing the veil of time for Earth Day 2026.

Deep-Time Mirror is an immersive, AI-powered ecological portal that visualizes the environmental history and restored future of any location on Earth. Built for the Earth Day Challenge, it combines high-performance 3D graphics with real-time generative storytelling to bridge the gap between our primeval past and a sustainable solar-punk tomorrow.

⚡ The Experience

🔭 The Deep-Time Mirror

Temporal Synthesis: The AI stabilization engine kicks in, generating a cinematic, full-screen 3D visual of your location.
Deep Past (~10,000 BCE): Witness the land in its untouched, primeval glory. Meet ancient species and see the world before the Anthropocene.
The Present: A gritty, realistic look at current ecological challenges, complete with a "Recent Pulse" that grabs local environmental data from 1 day and 1 week ago.
Restored Future (~2100 CE): A solar-punk vision of restoration. Witness how architecture and nature can coexist…

View on GitHub

How I Built It

My technical approach focused on Temporal Synthesis—using AI to ground creative visuals in scientific fact.

Google Gemini 1.5 Flash: I used Gemini to act as a "Temporal Oracle." It performs real-time searches to gather ecological data about the user's location, then generates structured JSON data for the narratives and climate statistics.

3D Cinematic Visuals: I utilized a generative visual pipeline that creates a full-screen, immersive background for every location. This "Art-First" approach ensures that users see the impact of climate change before they read about it.

React 19 & Three.js: The core interface uses React 19 for high-speed state management and Three.js (via React-Three-Fiber) to create the "Temporal Sphere" at the heart of the experience.

Design Philosophy: I used a high-end editorial/luxury aesthetic with Tailwind CSS v4, utilizing Glassmorphism and specialized scanline filters to give the app a "Retro-Futuristic Science" feel.

Data Visualization: I integrated Recharts to create custom-themed radar charts and built SVG Progress Dials to visualize the Planetary Score. These are fully responsive, data-driven components that transform AI output into actionable insights.

Advanced State Management: Leveraging React 19 hooks and Framer Motion, I created seamless “Temporal Transitions” that make switching between the Mirror and the Auditor feel fluid and cinematic.

Prize Categories

Best Use of Google Gemini

🌍 Grounded Real-Time Search: To find hyper-local environmental changes from the last 24 hours and 7 days.
🧠 Complex Reasoning: To generate scientifically plausible “Restored Future” scenarios based on geography and climate constraints.
🎭 Brutally Honest Personality Logic: Gemini is engineered as a strict planetary auditor, not a passive chatbot. It detects harmful habits (like plastic use or waste) and dynamically assigns an Earth Grade (A+ to F), creating a strong moral feedback loop.
📈 Multi-Dimensional Data Synthesis: It outputs structured data that powers the Planetary Stability Score and a 5-axis environmental radar chart, ensuring every user interaction produces unique, data-driven insights.

✨ Special Features

🌐 High-Fidelity Details

Custom Temporal Cursor: A glowing Earth orb with an orbit ring and subtle leaf trail that reacts dynamically to user interactions.
Live Pulse Integration: The landing experience includes simulated global activity (12k+ locations, 3.2M trees restored), making users feel part of a global movement.
Glassmorphism Design System: Every UI layer uses blur, transparency, and subtle borders to mimic a futuristic “Gaia Corp” terminal.
Cinematic Transitions: Moving between the Mirror and Auditor feels like navigating a high-end sci-fi interface.

🍃 A Message for Earth Day 2026 (April 22)

Earth Day is a reminder that we are not just residents of this planet—we are its custodians.

The Facts We Face:

Carbon Consciousness: CO₂ levels are currently at their highest in over 4 million years. Our “Deep Past” view reminds us of the baseline we are trying to return to.
Restoration is Possible: Scientists estimate that restoring just 15% of converted lands in priority areas could avoid 60% of expected extinctions. My “Future” view is a digital blueprint of this possibility.
Every Day Matters: The “Recent Pulse” in the app monitors changes from just 24 hours ago. Climate action isn't a long-term goal—it’s a daily rhythm.

Built with 💚 for the planet by: Sushan Shetty (sushanshetty2601-pixel)
Thank you 😇

✨ What I Built for the DEV April Fools Challenge: 🎵 Shake-to-Volume Music Player + 🐍 Shake-Controlled Snake Game

Sushan Shetty — Wed, 08 Apr 2026 17:20:03 +0000

🎵🐍 Shake-to-Volume Music Player + Tilt-Controlled Snake: Double Chaos Edition

A hilariously useless double feature where EVERYTHING is controlled by shaking/tilting your device!

What I Built

I created the world's most frustrating double feature:

A music player where volume ONLY works by shaking your device
A Snake game controlled entirely by tilting your phone or violently moving your mouse

Both features are completely useless, intentionally difficult, and will make you question reality.

Demo

🔗 Live Demo: https://silly-dev.preview.emergentagent.com/
🔗 GitHub: https://github.com/sushanshetty2601-pixel/shake-chaos-player

Screenshots

Features (All Intentionally Useless)

🎵 Shake-to-Volume Music Player

Shake Detection: Uses accelerometer on mobile, requires violent mouse movements on desktop
Volume Control: NO buttons, NO sliders - ONLY shaking works
Volume Decay: Stop shaking? Volume drops to zero. Keep shaking forever!
Shake Meter: Real-time visualization of your pathetic shaking intensity
Stats Tracking: Counts every single shake and scores your desperation
🤖 AI Shake Coach: Google AI roasts your shaking technique in real-time
Calibration Mode: 5-second test that judges if your shaking is "worthy" (spoiler: it's not)
Shake History: Database permanently records your shameful shaking attempts

🐍 Shake-Controlled Snake Game

Tilt Controls: On mobile, tilt device to change direction. On desktop, move mouse away from screen center.
No Arrow Keys: Traditional controls are for cowards
Game Over Messages: Sarcastic AI roasts like "😴 Pathetic tilting skills"
Score Tracking: Watch your failure in real-time
Classic Snake: Eat food, grow longer, inevitably crash into yourself while trying to tilt properly

Why This Is Hilariously Useless

Music Player Problems:

Want to turn up the volume? Shake violently for 30 seconds straight.
Listening on a bus? Everyone thinks you're having a seizure.
Desktop users must thrash their mouse around like a possessed etch-a-sketch.
Stop shaking for 2 seconds? Silent. Forever shaking or forever silent.

Snake Game Problems:

Playing Snake is already hard. Playing it by TILTING is impossible.
Desktop users must remember which quadrant their mouse is in.
Every game ends in 10 seconds with a sarcastic AI insult.
You'll look ridiculous tilting your phone back and forth.

Combined Chaos:

Try playing Snake while the music volume keeps resetting because you stopped shaking.
Ultimate multitasking nightmare.

How I Used Google AI (Gemini 3 Flash)

I integrated Google Gemini 3 Flash API to create an "AI Shake Coach"

Implementation:

Backend: FastAPI endpoint /api/shake-coach
Model: gemini-3-flash-preview via Google Generative AI API
Integration: Used Emergent's LLM integration library for seamless Gemini access

What the AI Does:

Analyzes Shake Data: Takes intensity, count, score, and device type
Generates Sarcastic Feedback: Creates custom roasts based on performance
Assigns Roast Levels: Pathetic → Mediocre → Acceptable → Legendary
Provides "Useless Tips": Like "Shake like you're trying to wake up your phone's ancestors"

Sample AI Responses:

"Congratulations on vibrating your desk with the raw power of a drowsy hamster..."
"At this rate, you'll reach 'audible' volume by the next solar eclipse"
"I've seen more intensity from a leaf of lettuce in a mild breeze"

Why This AI Usage Is Perfect for the Challenge:

The AI takes something already useless (shaking for volume) and makes it MORE useless by roasting you
Gemini's creative language generation makes every roast unique
It's meta-useless: AI judging your performance in a completely pointless task

Tech Stack

Frontend: React, Tailwind CSS, Web Audio API, DeviceMotion API
Backend: FastAPI, Python
Database: MongoDB (to forever store your shame)
AI: Google Gemini 3 Flash via Generative AI API
Deployment: Emergent platform
Motion Detection: DeviceOrientationEvent (mobile), MouseEvent (desktop)

How to Try It

Mobile:

Visit the live demo on your phone
Click "Play Music"
Shake your phone aggressively
Watch volume go up (and immediately drop when you stop)
Try the Snake game by tilting
Get roasted by AI

Desktop:

Visit on your computer
Move mouse violently in all directions
Or spam keyboard keys
Try controlling Snake by moving mouse to screen edges
Fail spectacularly

Code Highlights

Shake Detection (Mobile):

window.addEventListener('devicemotion', (event) => {
  const { x, y, z } = event.accelerationIncludingGravity;
  const acceleration = Math.sqrt(x * x + y * y + z * z);
  if (acceleration > 20) {
    const intensity = Math.min(acceleration / 50, 1);
    setVolume(Math.floor(intensity * 100));
  }
});

AI Shake Coach (Backend):

chat = LlmChat(
    api_key=os.environ.get('EMERGENT_LLM_KEY'),
    session_id=f"shake-coach-{uuid.uuid4()}",
    system_message="You are a sarcastic Shake Coach..."
).with_model("gemini", "gemini-3-flash-preview")

ai_response = await chat.send_message(user_message)

What I Learned

DeviceMotionEvent API is surprisingly powerful (and hilarious)
Gemini 3 Flash is excellent at generating creative sarcasm
People will question your sanity when they see you violently shaking your phone
Making something intentionally useless is harder than making something useful
AI makes everything better, even roasting your terrible shaking technique

Conclusion

This project perfectly embodies the April Fools spirit: it's technically impressive, completely useless, and will frustrate anyone who tries it. The AI integration adds an extra layer of absurdity by judging and roasting your attempts to use an already ridiculous interface.

Try it, shake it, tilt it, and get roasted! 🎉

Built with frustration, bad ideas, and Google AI by @sushanshetty2601-pixel

This app is intentionally useless. Created for DEV April Fools Challenge 2026.