Gaussian Splatting Workflow: Tools for Web, Unity, Unreal & 4DGS

3D / GAUSSIAN SPLATTING / PRODUCTION PIPELINE Gaussian Splatting Workflow: Tools for Web, Unity, Unreal & 4DGS Which tools are needed

Gaussian Splatting Workflow: Tools for Web, Unity, Unreal, and 4DGS
3D / GAUSSIAN SPLATTING / PRODUCTION PIPELINE

Gaussian Splatting Workflow: Tools for Web, Unity, Unreal & 4DGS

What tools are needed today for a professional Gaussian scattering workflow? This overview is deliberately not focused on beginner apps, but on the production chain: Capture and training, editing and compression, web output, game engines, as well as 4D Gaussian Splatting.
CLARENCE DADSON · UPDATED SEPTEMBER 2026
Gaussian Splatting Workflow: Tools for Web, Unity, Unreal, and 4DGS
01

Not a single tool, but a toolchain

Gaussian Splatting is no longer a single software workflow in 2026. Depending on the target, the tools differ significantly: a local high-end scan requires different tools than a quick smartphone capture; a web experience requires different formats than a Unity or Unreal project.
For the practice, a division into five steps therefore helps: 1. Capture & Training, 2. Editing & Optimization, 3. Web & Streaming, 4. Engine Integration and 5. 4DGS. This overview is built along this production chain.
Those looking for a simple entry point and general capture tools will find them in our Introductory guide to Gaussian scattering software. For the specific web integration, there is also our separate article Integrating Gaussian Splatting into websites.
02

Capture & Training: Postshot, KIRI Engine, Polycam

Postshot For us, Jawset is one of the most interesting tools when control and local processing are more important than maximum simplicity. Images or videos are processed locally, training runs can be configured, and the results can be exported as PLY or compressed SPZ. In addition, there are features for editing, camera animations, and integrations with After Effects and Unreal Engine.
KIRI Engine This is the pragmatic cloud/mobile variant. 3DGS can be generated from image material and exported as a PLY file; with appropriately configured projects, a mesh can be derived as well. This is practical when capturing should operate quickly and without its own GPU infrastructure.
Polycam It combines smartphone capture, cloud processing, and an accessible editing interface. This is particularly convenient for quick room and object shots. For automated or fully controlled production pipelines, a local workflow like Postshot is usually more flexible.
In a nutshell: Postshot for local control and production quality, KIRI Engine or Polycam for fast capture and cloud workflows.
03

Editing & Compression: SuperSplat, SOG and SPZ

SuperSplat is a browser-based editor for Gaussian splats and is particularly powerful for refining and optimizing existing scenes. Floaters can be removed, areas can be trimmed, splats can be selected, and data can be prepared for delivery.
For web projects, it is primarily SOG (Spatially Ordered Gaussians) Interesting. PlayCanvas recommends this format because of its efficient compression and short loading times. For larger scenes, streamed SOG octrees with multiple LOD levels can be used, so that the entire scene does not have to be loaded in full detail level.
SPZ It is another important option. The open format developed by Niantic greatly compresses Gaussian-Splat data and is supported by Scaniverse and, as of now, also by Postshot. Which variant is more useful depends on the target system: SOG fits particularly well into the PlayCanvas stack; SPZ is increasingly becoming an interoperable exchange format.
04

Web & Streaming: PlayCanvas, StorySplat, and JavaScript Viewer

PlayCanvas has evolved into one of the most complete web engines for Gaussian Splatting. Splats can be combined as regular scene components with UI, classic 3D objects, animations, and custom logic. SOG is natively supported; for streamed SOG, LOD and splat budget controls are available.
StorySplat It builds on a level above that. The platform offers viewers, authoring, hotspots, waypoints, WordPress/Webflow integrations, and now also tools for 4DGS. For editorial or marketing-oriented experiences, this is interesting if not every project is to be programmed completely individually.
gsplat.js is an open JavaScript library for developers who want to build a slimmed-down own viewer. For Three.js projects, mkkellogg/GaussianSplats3D was also a widely used option for a long time; the project now even indicates that it is no longer actively developed. For new projects, one should therefore prefer current alternatives.
A detailed guide to file formats, performance, and CMS integration is available in Integrating Gaussian Splatting into websites.
05

Unity, Unreal, and XR: Splats in Real-Time Engines

For Unity There are now several open-source renderers and packages available. The well-known UnityGaussianSplatting project supports PLY and SPZ and offers VR support, but is more of a developer tool than a finished product. Newer projects like UnitySplats expand the approach to include Unity 6, URP/HDRP, runtime loading, and other formats.
Unreal Engine It is especially interesting when splats are combined with classic real-time content, virtual production, or film pipelines. Postshot offers its own Unreal integration for this purpose and has further updated its support in 2026.
Therefore, for professional XR projects, one should not only ask whether a plugin can render a splat. Platform support, mobile/XR performance, streaming, interaction, format strategy, and the question of how splats can be combined with traditional 3D geometry are also crucial.
For pure browser-based output, PlayCanvas is currently the more straightforward option; for native VR/MR applications, Unity and Unreal remain important options.
06

4D Gaussian Splatting: Gracia AI and volumetric video

At 4D Gaussian Splatting The time dimension is added: Instead of a static scene, motion is recorded volumetrically. This is technically significantly more demanding and currently not yet an open standard workflow like static 3DGS.
Gracia AI It is one of the most visible production platforms here. The company offers a closed processing and compression pipeline for 4DGS, including streaming, WebXR, and output for Quest, Apple Vision Pro, and PCVR. According to its own statements, the pipeline achieves strong, near-lossless compression and provides SDKs for Unity, Unreal, and WebXR.
What matters is the classification: Gracia is currently more of a Studio and platform as a freely accessible open-source toolchain. For commercial volumetric projects, this can be very interesting; for those looking for a fully open workflow, 4DGS currently offers significantly fewer advanced options than static Gaussian Splatting.
→ Gracia AI
07

Which workflow fits which project?

Faster smartphone scan: KIRI Engine or Polycam → Export → SuperSplat if needed for cleaning.
Local high-quality production: Camera/Video → Postshot → PLY or SPZ → Further optimization and target platform.
Interactive website: Capture/Training → SuperSplat → SOG → PlayCanvas or StorySplat. For individual web apps, a dedicated JavaScript viewer can also be useful.
Native VR/MR application: PLY/SPZ → suitable Unity or Unreal renderer → performance and interaction optimization for the target headset.
Volumetric video / 4DGS: Multi-camera capture → specialized pipeline like Gracia AI → streaming, XR or VFX output.
08

Conclusion: The workflow is becoming more modular

Gaussian Splatting is now much more mature than it was two years ago. At the same time, there is no longer a single tool that dominates the entire process. Capture, training, editing, compression, and delivery are evolving into specialized components.
For static 3DGS projects, the pipeline is already very production-ready today: Postshot or Cloud-Capture for production, SuperSplat for editing and optimization, SOG/SPZ in compact formats, and PlayCanvas, StorySplat, or engine plugins for output.
In 4DGS, the market is even more closed and experimental. It is precisely there that a lot of movement is expected to take place in the coming years.

Frequently asked questions about Gaussian-Splatting tools

It depends on the use. Postshot is strong for local production and control, SuperSplat for editing and optimization, PlayCanvas for web experiences, and Unity or Unreal for native real-time and XR applications.

SOG is a highly compressed format that is particularly tightly integrated into the PlayCanvas ecosystem and also supports streamed LOD structures. SPZ is an open, compressed exchange format developed by Niantic that is supported by an increasing number of tools.

For more complex interactive applications, PlayCanvas is a strong option. StorySplat also offers authoring and CMS integrations. Developers can alternatively build their own viewers using JavaScript libraries like gsplat.js.

Yes. Plugins and integrations exist for both engines. However, the maturity differs depending on the platform and target device, so XR, mobile, and streaming requirements should be examined early on.

4DGS extends Gaussian Splatting to include motion over time and enables volumetric videos. Production usually requires synchronized multi-camera footage and specialized processing and compression pipelines.

Clarence Dadson

Interested in an XR project?

Whether Virtual Reality, Augmented Reality, 360° Video or 3D scan — we accompany you from the initial idea to the finished application.

Clarence Dadson · CEO Design4real

+49 179 3927971
·
info@design4real.de
·
Lindwurm St. 95a, Munich