CloudStudio turns massive point clouds into browser-native project links.

Point clouds should not be trapped inside specialist desktop software. A scan can capture a mine, road, building, stockpile, facade, forest plot, or construction site in extraordinary detail, but the result is often locked inside multi-GB files that only trained operators can open. Clients cannot inspect it. Sales teams cannot share it easily. Project owners cannot measure it without asking a specialist. Valuable 3D information becomes a file-transfer problem.

CloudStudio is built to change that.

It is a self-hostable web platform for uploading, converting, viewing, measuring, sharing, and extending LiDAR point cloud workflows directly in the browser. The goal is simple and ambitious:

Make point cloud sharing as easy as sending a web link.

Try CloudStudio directly in the browser before cloning or deploying anything:

• CloudStudio demo homepage - browse published scenes, upload workflow, feature overview, and project sharing UI.
• 3D Gaussian Splatting demo
  • open a SuperSplat-powered SOG scene with a share-ready browser link.
• LiDAR point cloud demo
  • inspect a browser-ready Potree scene with viewer and measurement tools.

The public demo is read-only for visitors. Upload and administration workflows remain protected on the hosted server; use the deployment guide to run your own self-hosted instance.

With CloudStudio, a team can move from "please install this point cloud tool first" to "open this link and inspect the site." It brings together the pieces that are usually scattered across desktop software, conversion tools, manual exports, screenshots, and large file transfers:

• upload LAS/LAZ point cloud data
• convert data into browser-ready Potree projects
• open and review scenes from a standard web browser
• measure distance, height, XYZ coordinates, profiles, and clipped regions
• run practical stockpile and earthwork volume workflows
• draw and export DXF linework
• browse, download, and share project outputs
• view 3D Gaussian Splatting assets through the bundled SuperSplat integration
• deploy the entire stack privately on your own server

This is not just another point cloud viewer. It is a starting point for a browser-based 3D work platform: scanner to server, server to browser, browser to decision. For dealers, survey teams, SLAM and LiDAR product groups, mining and construction users, BIM teams, and reality-capture developers, CloudStudio offers a practical path toward making 3D data easier to understand, easier to share, and easier to turn into action.

Modern reality capture hardware is advancing quickly, but sharing the result is still painful. A beautiful scan is not useful if the people who need it cannot open it, measure it, or discuss it without installing professional tools.

CloudStudio targets that gap:

• For customers: open a link, inspect the result, and understand the project.
• For dealers and sales teams: demonstrate scan value without sending giant files.
• For field teams: publish site data to a private server and review it from anywhere.
• For developers: extend a real point-cloud workflow instead of starting from zero.
• For AI agents: use the deployment guide to install and validate the platform step by step.

The long-term opportunity is large: self-hosted digital twin review, web-based LiDAR collaboration, 3D Gaussian Splatting publishing, browser-side measurement, private point cloud portals, dealer demo servers, and AI-assisted geospatial workflows can all build on the same foundation.

• Run a local CloudStudio viewer for development, testing, and demos.
• Deploy CloudStudio to an Ubuntu server with the included installer.
• Upload LAS/LAZ point clouds and convert them to Potree format.
• Open and inspect point cloud projects in a browser.
• Review measurements, profiles, clipping, display settings, and export tools.
• Use volume workflows for stockpile and earthwork-style calculations.
• Use DXF drawing/export workflows for lightweight linework handoff.
• Use 3D Gaussian Splatting viewer assets through the bundled SuperSplat integration.
• Give the repository and deployment guide to an AI coding agent to assist setup.
• Extend the frontend modules and backend processing scripts for your workflow.

CloudStudio is now prepared as an open-source, self-hostable foundation for browser-based point cloud workflows. It is functional and deployable today, while still evolving as a platform that developers, survey teams, dealers, and 3D software builders can adapt for their own field, sales, demo, research, or production workflows.

This repository contains source code, setup scripts, public documentation, license information, and lightweight static assets. Runtime datasets, customer files, credentials, generated point clouds, deployment secrets, and internal development notes are intentionally excluded.

cloudstudio-platform/
├── web-uploader/          # Main Express app, frontend shell, APIs, scripts, tests
├── potree/                # Potree viewer source and static runtime assets
├── PotreeConverter/       # PotreeConverter source, compiled on the target machine
├── docs/                  # Public technical and deployment documentation
├── setup.sh               # Ubuntu server installer
├── DEPLOY.md              # Human-readable deployment guide
├── LICENSE                # Open-source license for CloudStudio-specific work
├── CONTRIBUTING.md        # Contribution guidelines
├── SECURITY.md            # Security reporting and deployment notes
├── THIRD_PARTY_NOTICES.md # Third-party license summary
└── README.md              # Repository overview

• Technical overview
• Deployment guide
• AI agent deployment guide
• Roadmap
• Repository and GitHub workflow
• Web app notes
• Contributing
• Security
• Third-party notices

Requirements:

• Node.js 18 or newer
• Python 3.10 or newer
• npm
• A local build of PotreeConverter for full point cloud conversion support

Start the local app:

cd web-uploader
npm install
cp .env.example .env
npm run setup:local
npm run start:local

Open:

• http://localhost:8090/
• http://localhost:8090/viewer
• http://localhost:8090/health

If /health reports converter:false, build PotreeConverter locally or set CONVERTER_PATH in web-uploader/.env.

CloudStudio is designed for an AI-native deployment workflow.

In the AI era, deployment should not mean manually interpreting every command in a long operations guide. Give your AI coding agent this repository, share the deployment documents, provide the approved SSH/login method for your server, and let the agent walk through the setup, validation, and troubleshooting process with you.

The recommended flow is:

1. Share this repository link with your AI agent.
2. Point it to DEPLOY.md and docs/AGENT_DEPLOYMENT_GUIDE.md.
3. Provide server access in your normal secure way.
4. Ask the agent to install CloudStudio, run /health, inspect PM2/Nginx, and fix any missing dependency or configuration issue it finds.

For a fresh Ubuntu server, the core command is still simple:

sudo bash setup.sh

The setup script installs system dependencies, builds PotreeConverter, creates the Python environment, installs Node dependencies, configures Nginx, and starts the app with PM2.

If something fails, do not treat it as a dead end. Ask your AI agent to read the error output, check the health endpoint, inspect logs, rebuild PotreeConverter, repair Python dependencies, or adjust Nginx. The platform is intentionally documented so an agent can reason through the deployment instead of requiring a specialist to start from zero.

The following paths are runtime data and must not be committed:

• web-uploader/uploads/
• web-uploader/projects/
• web-uploader/pointclouds/
• web-uploader/gaussians/
• web-uploader/exports/
• web-uploader/cache/
• web-uploader/*_jobs/
• web-uploader/.env
• web-uploader/node_modules/
• web-uploader/.venv/
• PotreeConverter/build*/
• potree/pointclouds/*

Large source datasets such as LAS, LAZ, PLY, SOG, E57, ZIP archives, customer projects, and generated Potree outputs should live in server storage or object storage, not Git.

For a colleague, dealer, developer, or AI coding agent:

1. Clone or fork this repository.
2. Start with README.md, DEPLOY.md, and docs/AGENT_DEPLOYMENT_GUIDE.md.
3. Give your AI agent the repository URL, the deployment guide, and the approved SSH/login method for your own server.
4. Ask the agent to deploy CloudStudio, validate /health, inspect service logs, and fix missing dependencies or configuration issues.
5. Keep real customer data, generated point clouds, credentials, server logs, and private deployment notes outside Git.
6. Before pushing changes, run cd web-uploader && npm run check:public-release.
7. Open issues or pull requests for improvements that should benefit the wider project.

CloudStudio-specific code and documentation are released under the BSD 2-Clause License. Bundled third-party components keep their own licenses; see THIRD_PARTY_NOTICES.md.

Before using CloudStudio for a public or customer-facing deployment, review your own HTTPS setup, authentication requirements, API exposure, storage plan, retention policy, sample data policy, and production hardening needs.