Project Index
Liquid Galaxy RPG — A Multiplayer 2D Fantasy Game
Author: Shailesh Kumar Shukla
A short description of the goals of the project: Liquid Galaxy RPG is a real-time multiplayer pixel-art RPG built for the Liquid Galaxy rig. The goal was to turn the rig's multi-screen panoramic display into a shared game world: a Node.js + Socket.IO server renders one continuous Phaser world across all LG screens, while players join from their Android phones through a Flutter controller app.
What you did: I built LG RPG end to end — an authoritative Node.js + Socket.IO server that renders one continuous Phaser world across all Liquid Galaxy screens, and a Flutter controller app that both plays the game and manages the rig over SSH. The server handles the game loop, collision, pathfinding, spawning and projectiles, and supports two modes: co-op Zombie wave survival ending in a dragon boss fight, and PvP Zone Capture with real-time scoring and forfeits. Also added two playable characters with ranged and melee specials, a loadout system of power-ups and health items on cooldowns, four Tiled maps redrawn for the multi-screen world with an in-map leaderboard, and an optional AI cheerleader that commentates matches using Gemini and free Edge TTS. The controller app was rebuilt on clean architecture with pages for the lobby, loadout, Google Earth control and LG tasks, alongside install docs, a networking troubleshooting guide and asset credits.
The current state: Project finished and latest code on github.
What's left to do: Nothing.
What code got merged (or not) upstream: My project is fully developed and published on the Liquid Galaxy Go Store.
Any challenges or important things you learned during the project: The hardest part was port and firewall handling on the rig, where firewalls rules are strict. I fixed this by using whitelisted port 8111. Another challenge was panoramic map scaling; designing at full resolution made characters look tiny and it is also difficult to design. I solved it by designing a smaller map and scaling it up to fit the screens. Finally, the live AI commentator threatened to lag the 60Hz game loop. I resolved this by keeping the entire AI generation and text-to-speech completely optional and asynchronous.
Links: - Github - Go Store URL
↑ Back to menuEcoGrid Intelligence
Author: Bhoomi Shivhare
A short description of the project: EcoGrid Intelligence is a Flutter-based application that enables users to explore global power infrastructure through the lens of climate resilience. It combines global power plant data with historical climate data to calculate a Climate Vulnerability Score (CVS), provides AI-powered insights using Gemini, and visualizes climate-risk information on Liquid Galaxy through dynamic KMLs, synchronized Google Maps navigation, camera tours, orbits, and 3D visualizations.
What you did: - Global Power Plant Exploration: Developed the application using Flutter, Clean Architecture, and BLoC state management. Integrated the Global Power Plant Database containing 35K+ power plants. Implemented pagination, smart loading, search, and filtering for efficient exploration. - Climate Vulnerability Analysis: Integrated Open-Meteo for historical climate data. Developed climate anomaly and Climate Vulnerability Score (CVS) calculation engines. Added caching and optimized data fetching. - AI-Powered Insights: Integrated Gemini to generate contextual insights based on climate and power plant data. Connected AI insights with the application's climate analysis workflow. - Liquid Galaxy Visualization: Built the Liquid Galaxy visualization flow using dynamic KML generation. Implemented plant and regional orbits, camera tours, synchronized Google Maps navigation, and 3D visualizations. Tested and refined the visualization flow on physical Liquid Galaxy hardware. - Additional Features: Plant comparison and climate scenario simulation, speech-to-text and localization, theming and secure API-key management.
Current State: The project is fully developed and functional, with the core features implemented and tested on physical Liquid Galaxy hardware. It supports: Exploration of 35K+ power plants, Climate vulnerability analysis and AI-powered insights, Dynamic KML visualization, tours, and plant/regional orbits, Synchronized Google Maps and Liquid Galaxy navigation, 3D visualization, plant comparison, and climate scenario simulation.
What's Left to Do: Final validation on the Liquid Galaxy HQ rig.
What code got merged (or not) upstream: The project was developed in the official Liquid Galaxy repository. All significant GSoC implementation work, including features added beyond the original proposal, was merged upstream through reviewed pull requests.
Any Challenges or important things you learnt during the project: - Making dynamic Liquid Galaxy visualizations reliable on physical hardware, particularly with large KML files, regional orbits, and tours. - Efficiently handling 35K+ power plants while maintaining application performance through pagination, caching, and optimized processing. - Gained practical experience with Flutter architecture, climate-data processing, AI integration, geospatial visualization, and hardware-based testing.
Links: - GitHub - Go Web Store
↑ Back to menu
Rice Farm Agriculture India Visualization
Author: Vinayak Dhaka
Project Description: Build a Flutter controller application for the Liquid Galaxy multi-screen rig that visualizes India's rice agriculture on Google Earth. The app turns rice production statistics, seasonal crop cycles, irrigation and rainfall data, and guided narrated tours into an immersive multi-screen experience, fully controllable from a phone over SSH.
What i did: - Colored 3D production visualization — extruded polygon KMLs for all Indian states, color-scaled by rice output, rendered across every screen of the rig. - Major Rice Regions — fly-to any state with TTS narration, a full 360° orbit camera, and a production-tier color scale. - Irrigation & Rainfall — per-state rainfall polygons, irrigation-source markers, orbit, and water-source dashboards. - Seasonal Crop Cycle — Kharif/Rabi stage visualization with auto-play narration. - Three guided tours — automated storytelling with TTS narration and matching data dashboards per step. - Synced Navigation — real-time Google Maps pan/zoom drives the Liquid Galaxy camera live. - Side-screen dashboards — HTML balloon dashboards with live statistics and narration text. - KML self-diagnostic — a built-in on-screen verifier that checks delivery end to end.
The current state: The application is complete and working on the Liquid Galaxy rig. All features — production maps, fly-to, orbit, irrigation, crop cycle, guided tours, side-screen dashboards, branding, and synced navigation — are functional and have been reviewed and approved by mentors.
What code got merged upstream: All work was contributed and merged through 12 pull requests to the official Liquid Galaxy repository. The final GSoC commit is the last commit merged in PR #12.
Important things i learned during the project: The most significant challenge was a colored-KML rendering issue: the visualizations worked perfectly on my local 3-screen rig but failed on the physical test rig. I studied the Liquid Galaxy core source code and past GSoC projects to understand the rig's KML sync mechanism, then built an on-screen self-diagnostic that verified each stage of delivery and reported the exact failure remotely. Resolving it taught me disciplined debugging of distributed systems I could not directly observe, and the value of building instrumentation instead of guessing.
Links: - GitHub - Documentation - Go Store
↑ Back to menuLG Interactive Onboarding: Guided Tutorial & Learning System
Author: Darpan Baviskar
A short description of the goals of the project: The primary goal of this project is to lower the barrier to entry for using Liquid Galaxy rigs by providing an intuitive, interactive interface. Specifically, it aims to: - Educate: Offer a comprehensive, gamified curriculum to teach users about Liquid Galaxy features and capabilities. - Empower Creation: Provide an easy-to-use 3D model builder that allows users to seamlessly import, manipulate, and project custom models onto the rig without deep technical knowledge of KML. - Simplify Control: Act as a centralized dashboard to easily manage the rig's state and execute complex commands.
What you did: I developed a secure SSH communication layer to remotely manage the rig’s power state, alongside implementing a persistent navigation system and a modern dark theme. A significant portion of the development was dedicated to engineering a 3D Model Builder. The goal was to allow users to interactively position 3D assets on an OpenStreetMap interface and seamlessly project them onto the rig. I designed a targeted import solution that reliably processes .DAE (COLLADA) and packaged .ZIP files. I built an execution queue to prevent SSH connection overloads under heavy use and migrated the app to a robust Riverpod state management architecture. I also developed a dedicated KML Playground as an interactive testing environment, and an interactive Curriculum Engine equipped with Text-to-Speech (TTS) narration and an auto-verification polling system.
The current state: The current state of the main branch is a fully functional, Riverpod-managed Flutter application designed specifically for the Liquid Galaxy (LG) ecosystem. It features: Core Infrastructure, LG Management Module, 3D Model Builder, Curriculum Engine, Architecture Explorer, and KML Playground.
What's left to do: Enhanced 3D Parsing Engine: Optimize the underlying logic to autonomously identify and mitigate discrepancies within non-standard, corrupted, or incorrectly scaled .DAE assets.
What code got merged (or not) upstream: My project is fully developed and published on the Liquid Galaxy Go Store and onto the GitHub Releases page on the project repository.
Any challenges or important things you learned during the project: A significant challenge involved reconciling 3D COLLADA model rendering inconsistencies between modern development environments and the legacy Liquid Galaxy hardware. I systematically replicated the rig's hardware constraints by downgrading Google Earth and artificially limiting virtual machine resources. This iterative process was instrumental in deepening my understanding of legacy hardware limitations.
Links: - GitHub - GO Store Link
↑ Back to menuUNESCO World Heritage
Author: Saumya Bhattacharya
Project Goals: The goal of this project was to build a Flutter Android application that helps users explore UNESCO World Heritage Sites through a Liquid Galaxy rig. The app connects mobile exploration with immersive multi-screen visualization, allowing users to search heritage sites, view site information, render KML boundaries, run orbit tours, and learn through AI-assisted storytelling.
What I Did: I developed the application using Flutter and Dart, with Liquid Galaxy integration through SSH. I implemented UNESCO site discovery, search and filters, site detail pages, Google Maps previews, KML generation, boundary rendering, orbit tours, LG commands, Gemini-powered chat, AI-generated site narration, weather information, and local settings for API keys and rig connection details.
Current State: The main application features are implemented and working. Users can connect to a Liquid Galaxy rig, explore UNESCO World Heritage Sites, send KML visualizations to Google Earth, run orbit tours, view site information panels, use Gemini features, and control common LG commands from the app.
What's Left to Do: Adding more languages, improving offline support, expanding the dataset, and refining some edge cases in map/API loading when network services are slow or unavailable.
Code Status: The completed application code and documentation have been merged into the main project repository.
Challenges and Learnings: The most challenging parts were handling Liquid Galaxy SSH communication reliably, generating KML that renders well across multiple LG screens, tuning camera angles and orbit ranges, and combining live map interaction with Google Earth visualization. During the project I learned a lot about KML generation, Flutter architecture, AI API integration, and debugging real Liquid Galaxy rig behavior.
↑ Back to menuLocal AI with Gemma by Google
Author: Harsh Mehta
A short description of the goals of the project: This project introduces the world of Agentic AI based on Hermes and Gemma or any other LLM models for the Liquid Galaxy Project. The goal was to develop an AI-powered assistant for Liquid Galaxy, capable of understanding natural-language requests and converting them into actions and interactive visualizations on a Liquid Galaxy rig. Instead of pre-programmed topics and pre-built KMLs, this project could make dynamic KMLs based on different sources of data, such as multiple APIs and internet searches, at runtime.
What you did: During the GSoC period, I developed Nara, an AI assistant for Liquid Galaxy built on top of Hermes Agent. The main focus was to make the agent capable of understanding natural-language requests, using modular skills, and turning those requests into actions and visualizations on a real Liquid Galaxy rig. The project started with the core Liquid Galaxy control layer. Nara can establish an SSH connection with the rig and perform administrative operations. A major part of the work was designing and developing a modular skill architecture for Liquid Galaxy covering: Weather monitoring, Geography education, History education, Natural disasters, Maritime awareness, Aviation tracking, Cyber infrastructure, Economic markets, Armed conflicts, Energy monitoring, Animal migration, Coral reef monitoring, Deforestation tracking, Satellite orbital tracking, Prediction Markets, Global Progress, etc.
The current state: The core agent runtime and Liquid Galaxy integration are working. Nara can connect to the Liquid Galaxy rig, execute control commands, generate KML, deploy it to the LG rig, replace or clear existing visualizations, and support multiple advanced domain-specific visualization skills.
What's left to do: This project can be used as a base to further extend Agentic AI applications for the Liquid Galaxy Project.
What code got merged (or not) upstream: The major parts of the codebase are Detailed documentation and Demonstration applications developed using the agent.
Any challenges or important things you learned during the project: My mentors helped me make an end user focused project. I realized how a very well written documentation is important which can be understood by different types of people. This project involves natural language to operate so a good documentation is such that even a non technical user can read and understand most about the project and can use it properly. I also learnt that the architecture of a project must be planned to be scalable from the very start.
↑ Back to menuCity Historic Engine for Liquid Galaxy
Author: Yasmina Ramadan Sevdanova
A short description of the goals of the project: The main goal of the City Historic Engine project is to develop an application that provides historical and cultural information about a city. The application aims to offer users a simple and interactive way to discover important historical places, such as churches, cathedrals, museums, and other points of interest. For this project, the chosen city is Lleida.
What you did: I developed an application focused on the historical and cultural heritage of Lleida. The application allows users to explore different historical locations in the city and access information about their history, meaning, and importance. The objective was to make the information clear, accessible, and easy to explore.
The current state: The application is fully finished.
What's left to do: Nothing.
What code got merged (or not) upstream: All the code was merged.
Any challenges or important things you learned during the project: Challenges: I have a lot of problems just putting the logos. Things learned: Specially I got deeper knowledge about Flutter. Learned how to structure information within an application and how to create an interactive experience focused on discovering historical places. I also learned more about developing an application that combines technology with cultural and educational content.
↑ Back to menuGeoSaurio
Author: Josep Miquel Sert Esteban
A short description of the goals of the project: GeoSaurio is an educational and interactive application developed for Liquid Galaxy that allows users to explore dinosaurs from different geological periods around the world. The main goal of the project is to combine paleontology, geographic exploration and the immersive capabilities of Liquid Galaxy. Users can navigate through geological periods, continents and countries, select dinosaur species, explore their fossil locations on Google Earth and access additional information about each dinosaur.
What you did: During the project, I developed the GeoSaurio Flutter application and integrated it with Liquid Galaxy through SSH and KML commands. The main features implemented include: Navigation by geological period, continent, country and dinosaur; Google Earth FlyTo navigation for continents, countries and individual dinosaur locations; Dinosaur location markers based on geographic coordinates; Information panels displayed on the Liquid Galaxy side screens; Dinosaur images and scientific information including period, year, diet, habitat, size, weight, location, author, status and geological formation; Dinosaur skeleton visualization; Dinosaur size comparison visualization; Audio narration for dinosaur information; Light and dark themes.
The current state: GeoSaurio is currently finalized. The main features of the application have been implemented and tested with Liquid Galaxy. The complete navigation flow from geological period to continent, country and dinosaur is working.
What's left to do: Nothing.
What code got merged (or not) upstream: All the code was merged with the Liquid Galaxy
Any challenges or important things you learned during the project: One of the main challenges was learning how to communicate with and control a Liquid Galaxy installation from a Flutter application. Working with SSH, KML and Google Earth camera parameters such as latitude, longitude, altitude, range, tilt and heading required a lot of testing on the actual Liquid Galaxy system. Throughout the project, I improved my knowledge of Flutter and Dart, asynchronous programming, SSH communication, KML generation.
Links: - GitHub - Go Store Link
↑ Back to menuLiquid Galaxy Time Machine
Author: Jan Sanchez
A short description of the goals of the project: Liquid Galaxy Time Machine is a mobile app that lets you travel through time across global landmarks using the Liquid Galaxy multi-display system. It allows users to explore the Past, Present, and Future through satellite imagery and AI-driven projections such as future estimation images. Featuring visual comparisons, visual images of those landmarks, and smart narration, the app provides an immersive, educational journey through our world’s evolution.
What you did: I’ve developed next to Gemini the application with Flutter using Android Studio to manage the code and test the application around the times of developing. My app has always been centered not only to educate and display a fun topic for liquid galaxy projects, but also to show a visually aesthetic application and attractive imagery and displays on the liquid galaxy. Also I’ve been investigating around how to implement an AI Library for flutter, specifically for the AI Image generation for the future estimation, being able to choose your own API and model for the Image.
The current state: The project is successfully completed
What's left to do: The project is fully completed. I think that better 3D boundaries have to be created, the ones that are being displayed in each landmark delimiting the area around it, at least visually for a more immersive and professional view of the area itself.
What code got merged (or not) upstream: My project is fully developed and ready to be published on the Liquid Galaxy Go Store.
Any challenges or important things you learned during the project: I’ve faced a lot of challenges during the project development, specially in the KML display in the liquid galaxy screens. KMLs showing on screens that weren’t supposed to be showing, or implementing 3d KMLs that didn’t display in the master screen. I’ve learned a lot about organization around things I’ve had to do on the project and lots of coding methods for Flutter.
↑ Back to menuAI Tour Director
Author: Kabir Khanuja
A short description of the goals of the project: AI Tour Director is a phone app that turns a single sentence into a cinematic tour of the world on a Liquid Galaxy rig. You type something like "the historic forts of India" or "the most beautiful places in Switzerland", and the app figures out real places that fit, then flies the giant screens through each one with narration, like a documentary that writes itself. On top of that, the app can also create a short AI movie of the tour you just watched.
What I Did: I built the whole thing as a single Flutter app for Android, with no separate server behind it. When you enter a prompt, the app asks an AI model to pick four to six real, well known landmarks and write a short, spoken description for each. It connects to the Liquid Galaxy screens over a secure network link and drives Google Earth to fly into each landmark, circle all the way around it smoothly, and show an information card with the photo and description on the side screen, while the phone reads the narration out loud. The second half of the project is the AI Film feature. After a tour, the app can ask an AI video model to generate a short cinematic clip for each place, then stitch all those clips together into one film right on the phone.
Current State: The project is complete and published on the Liquid Galaxy GO Web Store. I tested it end-to-end on a real Liquid Galaxy rig: the tours fly correctly, the smooth orbit works, the information cards and logo appear on the right screens, the narration stays in sync, and the AI Film feature was verified end-to-end on a real Android phone.
What's Left To Do: The app is fully usable today, and the remaining items are enhancements rather than missing pieces. The most exciting one is generating the AI film in the background while the tour is still playing on the rig, so the film is ready the moment the tour ends instead of afterwards.
What Code Got Merged: The project is fully developed and published on the Liquid Galaxy GO Web Store.
Challenges and Learnings: The hardest and most rewarding part was the smooth circular flight around each landmark. My early attempts sent hundreds of tiny camera commands over the network, which made the motion stutter badly. The breakthrough was to send one small instruction that runs directly on the rig and rotates the camera by itself, carefully paced to match how fast Google Earth can actually turn. The AI Film feature taught me to design for failure. Video generation costs real money per clip, so I made sure that if a user runs out of credit halfway through, the app stops spending immediately, still stitches together the clips it already made, and explains clearly what happened.
Links: - GitHub - GO Store URL
↑ Back to menu
