SkillFoxx

World to World Mapping skill

A skill that guides an agent from a Gaussian splat to a Blender scene on Geometry Nodes and Shader Nodes, checked on views.

  • 83 ★ on the repository
  • Last commit
  • MIT
Skill

Extended access

Why this level

  • The skill runs Python and Blender scripts on your machine and writes scene, render and manifest files.
All reasons and checks

gokayfem/gaussian-splat-to-blender-nodesAuthor:gokayfem

Install

In your terminal, with SkillFoxx CLI

npx skillfoxx add skills/world-to-world-mapping

Detects the agents on your machine, shows the access level and pins the version.

Other ways to install

Access and checks

How we review

Extended accessThe tool runs code or makes network calls and only reads external data.

  • The skill runs Python and Blender scripts on your machine and writes scene, render and manifest files.

This is third-party code. Review the repository files before installing.

What it does

World to World Mapping is a skill and toolkit for reconstructing a scene from a Gaussian splat in Blender. The result is not a one-click mesh conversion: the agent describes objects through meaningful geometry, parameters and materials on Geometry Nodes and Shader Nodes, and you can edit them in Blender. The workflow covers capturing reference views, a measured coordinate frame and cameras, splitting the scene into parts, building them one by one, review and direct critique of mismatches. The kit includes prompt templates, an inspector for PLY, SPLAT and KSPLAT files with a SHA-256 hash, a camera generator, a native scene audit, rendering from a camera manifest and frame comparison. Texture projection tools help map captured appearance.

Who it is for. 3D artists and developers who need an editable Blender scene from a splat instead of a dense mesh with no structure.

Good fit when

  • A splat of a garden or a room exists and the scene is rebuilt from tables, paving and plants with parameters.
  • You want to start with a trial template without your own scan: the kit includes a synthetic table.
  • You must confirm that the final file keeps no hidden reference point cloud.
  • A scene is compared with the splat on frames and video from several angles.

Not a fit when

  • You need an automatic splat-to-mesh converter: the skill drives stepwise agent work.
  • You have no rights to the splat source: the skill assumes authorized source data.
  • Blender below 5.0 or Python below 3.10.

Example request

Use gaussian-splat-to-blender-nodes. Reconstruct this scene as editable native Blender geometry and materials, start with a measured coordinate frame and cameras and be candid about mismatches.

Limitations

Blender 5.0 or newer and Python 3.10 or newer are required; verification used Blender 5.0.1, so validate other versions yourself. The source scan and the fitted assets of the Garden reference example are not bundled. Reference frames must come from a compatible viewer using the same camera manifest, and Blender renders only the native side. Appearance mapping produces a hybrid scene: some objects stay node-based, others are baked into meshes with textures. This is reference view capture, not splat training.

How to disable. Delete the `gaussian-splat-to-blender-nodes` folder from the agent skills directory or run `/plugin uninstall` for the plugin. Scenes, renders and manifests in `outputs/` are deleted by hand.

FAQ

How do I install the skill?

Run `npx skills add gokayfem/gaussian-splat-to-blender-nodes`. A Claude Code plugin exists: `/plugin marketplace add gokayfem/gaussian-splat-to-blender-nodes`. For Codex copy the skill folder into `~/.codex/skills`.

Can I try it without my own scan?

Yes. The `starter.py` template creates a small editable table on Geometry Nodes, and `audit_blend.py`, `cameras.py` and `render_manifest.py` run the whole pipeline.

What is the native scene audit?

The script deletes the hidden reference cloud in memory without changing the saved file and compares mesh and instance signatures.

How do I brief the agent?

Give an authorized source, intended views and a quality target, then name the skill. A ready starting message is in the Reconstruct your scene section.

Do I need Blender installed?

Yes, version 5.0 or newer. Verification ran on 5.0.1.

Discussion

No questions or reviews yet.

Skills

A loop skill: the agent generates a target screenshot, builds toward it, and uses a critic subagent to close the gap until they match

SkillExtended access2K

Agent skill for Blender: builds a procedural node system from a video, image or idea and checks the finished graph.

PluginExtended access95

A Claude Code skill that builds a game-ready 3D asset in Blender from concept art and checks each phase by measurement.

SkillExtended access182

A skill rebuilds the object in a reference image as a procedural Three.js model written in code, not downloaded art

SkillMinimal access21.3K

A Claude Code workflow that builds a dressed, lit Blender scene from reference images or one paragraph, with graded renders.

WorkflowExtended access369

A skill collection guides AI agents through Blender modeling, materials, animation and asset checks using supplied Python scripts.

SkillBroad access1K

README in short

The description explains the approach (agent reconstruction rather than a converter), lists the kit contents, gives three install options and walks through a trial without a scan, scene reconstruction and appearance mapping.

README badge

Original SKILL.md

---
name: gaussian-splat-to-blender-nodes
description: Reconstruct Gaussian splat scenes as editable native Blender Geometry Nodes meshes and Shader Node materials. Use for splat-to-procedural reconstruction, matched-camera analysis, semantic scene decomposition, detailed environment recreation, native-scene audits, optional hybrid capture-texture mapping onto reconstructed surfaces, and reference-versus-Blender comparison videos. This is an agent-guided measured workflow, not automatic splat-to-mesh conversion.
license: MIT
---

# World to World Mapping — Gaussian Splat to Blender Nodes

Build an editable scene that explains the reference through geometry, materials and light.
Use the splat as measurement and visual evidence. Deliver native source, parameters and
honest multi-view comparisons. Default to procedural materials. When captured appearance is
requested, use the optional hybrid workflow below and disclose its baked-surface limits.

## Prompting a complete project

For a full-scene task, use the [prompting workflow](references/prompting-workflow.md) and
[master project brief](templates/prompts/project-brief.md). They supply the actual brief,
capture, part-contract, detail, review and delivery prompt structure with project-specific
values left to measure. Read [whole-scene capture](references/whole-scene-capture.md) when
building the reference package: it explains the source-viewer, camera, coverage and close-up
capture procedure. These prompts work with one agent; delegation and recurring review are
optional, separately authorized modes.

## Start from evidence

1. Inspect current artifact versions and preserve user work.
2. Record asset URL, author/dataset, usage terms, byte size and SHA-256. Viewer software
   licensing does not establish dataset rights. Keep scans and decoded arrays local.
3. Identify the actual container and producer. `scripts/inspect_asset.py` fingerprints
   assets and reads limited metadata; it does not decode Gaussians into surfaces.
4. Establish one scan-to-world transform, scale, handedness and camera convention. Save
   position, target, up, vertical FOV and dimensions. Register stable landmarks first.
5. Read [acquisition and cameras](references/acquisition-and-cameras.md).

## Reconstruct coarse to fine

1. List major objects and occlusions. Separate measurements, estimates and invented hidden
   surfaces. Prioritize silhouette and spatial relationships.
2. Create collections: `GS_REFERENCE` (hidden from rendering), `SCENE_PROCEDURAL` (or a
   project-specific equivalent), `CAMERAS`, `LIGHTING`, and `QA`.

Report a problem