Open Source
Explore the latest AI open-source projects from GitHub and HuggingFace.
Explore the latest AI open-source projects from GitHub and HuggingFace.
SuperSplat is a free, open-source browser-based editor for inspecting, editing, optimizing, and publishing 3D Gaussian Splats, developed by PlayCanvas under an MIT license. With version 2.26.1 released on May 14, 2026, the project has grown to 8.4k stars and 903 forks, becoming the de facto open tooling for the rapidly expanding 3D Gaussian Splatting ecosystem. ## The Editor Gap in 3DGS 3D Gaussian Splatting has become the dominant technique for photoreal radiance-field reconstruction, but most tools in the ecosystem focus on training (gsplat, nerfstudio) or research (vanilla 3DGS implementations). Production workflows demand something different: an editor that can clean noisy splats, crop unwanted geometry, optimize file size for web delivery, and export to standard formats. SuperSplat fills exactly that gap, providing a Photoshop-class editing surface for what was previously an unwieldy point-cloud-like representation. ## Browser-Native Architecture The entire editor runs in the browser at https://superspl.at/editor — no installation, no GPU drivers to manage, no platform compatibility matrix. SuperSplat is built on TypeScript (92.6% of the codebase) using WebGL and WebGPU for hardware-accelerated rendering of millions of Gaussian primitives. The PlayCanvas engine handles scene graph management, while PCUI provides the editor's interface widgets. This stack means a creator on a Chromebook can load, edit, and republish a 20-million-Gaussian capture with the same fidelity available on a desktop workstation. ## Core Editing Capabilities SuperSplat supports the workflows that production 3DGS pipelines actually need. Box and sphere selection tools let editors isolate regions for deletion, color adjustment, or transformation. A rectangle tool with pixel-precise marquee selection complements bulk operations across the splat cloud. The optimizer compresses and reorders splats to dramatically shrink file size for web delivery, while preserving visual quality through perceptual ranking. Background removal, ground-plane alignment, and rotation tools handle the most common post-capture cleanup tasks. ## Publishing Pipeline The editor exports to multiple standardized formats including PLY, compressed PLY, and SOG (Self-Organizing Gaussians) — the emerging format optimized for web streaming. SuperSplat also provides direct publishing to PlayCanvas's hosting service, generating shareable URLs with embedded viewers that work across desktop and mobile browsers without requiring app installs. This closes the loop from capture to public-facing 3D content. ## Localization and Accessibility The project ships with multi-language localization, broadening adoption beyond English-speaking developers. Combined with the zero-install browser deployment, SuperSplat lowers the barrier to 3D Gaussian Splatting more than any other tool in the ecosystem — a high schooler with a phone capture and a free browser tab can now produce publishable 3D scenes. ## Position in the 3DGS Ecosystem While training frameworks like gsplat and FastGS push the research frontier, SuperSplat serves the equally important post-production layer. It complements rather than competes with these tools, sitting downstream as the polish-and-publish surface. For creators, studios, and developers shipping 3D content to the web in 2026, SuperSplat has become an essential link in the pipeline.
graphdeco-inria
Original reference implementation of 3D Gaussian Splatting for real-time radiance field rendering
ahujasid
Connect Blender to Claude AI via MCP for natural-language-driven 3D scene creation and manipulation.
Tencent Hunyuan
Tencent's open-source 3D asset generation system with 13k+ GitHub stars, creating high-resolution textured 3D models from a single image using a two-stage diffusion pipeline.