mrnerf/lichtfeld-studio

Train, inspect, edit, automate, and export 3D Gaussian Splatting scenes from a single native application.

View on GitHub ↗Jump to charts ↓Open shareable report

Summary Information

Updated 19 minutes ago
Added to GitGenius on September 18th, 2026
Created on July 30th, 2023
Open Issues & Pull Requests: 221 (-2)
GitHub issues: Enabled
Number of forks: 461
Total Stargazers: 3,725 (+0)
Total Subscribers: 42 (+0)

Repository Insights (GitGenius)

Median issue/PR response: 18.1 hours
Mean response time: 21.2 days
90th percentile: 58.5 days
Tracked items: 675

Most active contributors

Sign in to see contributor activity.

How this project is maintained

Roughly one issue in five opened in the past year never receives a reply. 31% of open issues come from outside the core team, a mix of external reports and the maintainers' own roadmap. Work labelled "GUI" is answered fastest, typically in about 14 hours, while "editing" waits about 13 days. 22% of tracked open issues have had no activity in three months. 70% of issues opened in the past year have been closed, leaving a working backlog.

Charts & Analytics

Fetching additional details & charts...

Issue Activity (beta)

Open issues: 207
New in 7 days: 45
Closed in 7 days: 37
Avg open age: 88 days
Stale 30+ days: 131
Stale 90+ days: 51

Recent activity

Opened in 7 days: 41
Closed in 7 days: 29
Comments in 7 days: 12
Events in 7 days: 102

Top labels

  • enhancement (185)
  • bug (156)
  • UI/UX (98)
  • training (58)
  • good first issue (48)
  • Project Manager (46)
  • GUI (42)
  • help wanted (42)

Detailed Description

LichtFeld Studio is a native application workstation for training, inspecting, editing, and exporting 3D Gaussian Splatting scenes.

The tool addresses the fragmentation of Gaussian Splatting workflows by consolidating training, visualization, editing, and export into a single application. Users can train new scenes from COLMAP datasets or resume from checkpoints, inspect reconstructions in real time during or after training, select and transform gaussian subsets with undo/redo support, and export to multiple formats including PLY, SOG, SPZ, and standalone HTML viewers. The application is built in C++ with CUDA support for efficient GPU computation.

Researchers benefit from the ability to iterate on reconstruction quality and inspect training progress interactively without switching between separate tools. Production teams can visually inspect scenes, edit gaussian selections, and deliver portable exports as a complete package. Tool builders can extend the application through a Python plugin system with plugin-local dependencies, embed Python directly for automation, and integrate workflows via Model Context Protocol resources and tools. The workstation suits projects where a unified interface for the full Gaussian Splatting pipeline reduces friction compared to assembling multiple standalone scripts and viewers.

The project maintains active development with regular feature additions and bug fixes. The codebase shows ongoing refinement of core functionality across training, visualization, and editing systems. Community engagement is supported through Discord and documented contribution guidelines. The plugin system and MCP integration indicate a deliberate architecture designed for extensibility, suggesting the maintainers prioritize enabling users to customize and automate their workflows rather than building every feature into the core application.