---
title: "A generated 3D world whose objects remain movable"
locale: "en"
url: "https://irz.fr/en/articles/worldclaw-editable-3d-worlds-en"
markdown_url: "https://irz.fr/en/articles/worldclaw-editable-3d-worlds-en.md"
category: "creative"
tags: ["3D", "generation", "AI", "games"]
published_at: "2026-08-11T14:05:00.000Z"
author: "Camille Morel"
translation: "https://irz.fr/fr/articles/worldclaw-mondes-3d-editables-fr.md"
---

# A generated 3D world whose objects remain movable

WorldClaw turns a prompt into regions, builds global terrain, then generates local objects as separate meshes. The important difference is structural: the result remains an explicit 3D scene instead of a fixed view.

A generated video of a world can look spectacular and become useless the moment you want to move one house three meters to the left.

**WorldClaw**, a Tencent Hunyuan3D research project released in early August, is aimed at that exact gap: generating a world from text while preserving explicit terrain, objects and transforms.[1](https://arxiv.org/abs/2608.05248)

The goal is not merely a camera that can move through a convincing visual illusion. Generated objects are supposed to remain independently manageable so they can be moved, reused or handed off to downstream 3D workflows.[1](https://arxiv.org/abs/2608.05248) [2](https://tencent-hunyuan.github.io/Hunyuan3D-WorldClaw/)

That distinction is less exciting in a teaser. It matters much more once somebody needs to keep working after the teaser ends.

## Two resolutions

WorldClaw does not try to generate every rock at full detail from the beginning.

A planning stage first turns an open-ended prompt into a structured specification describing regions, terrain, objects, appearance and spatial relationships.[1](https://arxiv.org/abs/2608.05248)

The system then builds global structure. A semantic layout divides the terrain into regions, while a height field combines elevation, noise and landform operators to create mountains, dunes, terraces and related shapes. Materials and terrain assets follow those regional semantics.[1](https://arxiv.org/abs/2608.05248)

Local detail comes afterward.

For regions that need it, WorldClaw generates a terrain-conditioned 2D composition, segments individual objects, reconstructs them as textured meshes and recovers their placement and orientation in the world.[1](https://arxiv.org/abs/2608.05248)

That keeps a coherent global terrain while spending expensive local generation where detail is actually needed.

## A tree, not pixels

The paper repeatedly emphasizes explicit representation.

Each local object has its own mesh, attributes and placement transform. Render-guided refinement agents inspect the result and can adjust pose, size, orientation or terrain contact before rendering again.[1](https://arxiv.org/abs/2608.05248) [2](https://tencent-hunyuan.github.io/Hunyuan3D-WorldClaw/)

A floating object can be reseated. An oversized building can be scaled. A badly oriented prop can turn without regenerating the entire world.

Traditional 3D has always had this advantage: a scene is a manipulable structure, not only an appearance.

Image and video generation made appearance dramatically easier. WorldClaw is trying to put structure back into the loop.

## The spatial contract

The starting prompt can remain broad: a medieval village spanning snowy mountains, plains, water and desert, for example.[2](https://tencent-hunyuan.github.io/Hunyuan3D-WorldClaw/)

WorldClaw first converts that intent into an intermediate representation that later stages share. The paper describes regions, terrain constraints, object categories and spatial relationships.[1](https://arxiv.org/abs/2608.05248)

That specification behaves somewhat like a conceptual scene file. Each downstream agent does not need to reinterpret the original prose from scratch.

This may be the most transferable part of the design.

When several agents participate in a complex build, sharing an **explicit structure** can be more useful than sharing only the original natural-language instruction.

Terrain becomes a spatial contract. Regions become work units. Local objects arrive later as separate components.

## Paper, no tool

This is where the demo needs restraint.

WorldClaw is currently a **research project**. The public GitHub repository contains a README, pipeline image and teaser. At the time we checked, it did not contain the implementation required to reproduce the system.[3](https://github.com/Tencent-Hunyuan/Hunyuan3D-WorldClaw)

The paper's comparisons are primarily qualitative and the reported results come from the team proposing the method.[1](https://arxiv.org/abs/2608.05248)

An artist cannot currently install WorldClaw from that repository and build a game with it.

Even the official project visuals are not reusable press assets by default: Tencent states that its visual content remains owned by Tencent unless otherwise noted and grants no general license to it.[2](https://tencent-hunyuan.github.io/Hunyuan3D-WorldClaw/)

The project demonstrates an architecture, not yet a reproducible creative pipeline.

## View or world?

That limitation does not make the direction less interesting.

An image is excellent for exploring a mood. A video can suggest a walkthrough. A 3D production eventually needs much less glamorous operations: move an object, swap a material, change a path, export to a physics engine or ask a character to walk around a rock.

For that, the rock has to **exist as an object**.

WorldClaw is organized around that requirement. Global terrain provides coherence. Regions provide decomposition. Separate meshes give the creator local handles again.[1](https://arxiv.org/abs/2608.05248)

That shift matters more than screenshot quality.

The next useful question for generative 3D is not only whether something looks like a world.

It is whether you can still work inside it after generation is finished.

## References

1. [Guo et al., “WorldClaw: Agentic 3D Open-World Generation at Scale”, arXiv, August 2026](https://arxiv.org/abs/2608.05248)
2. [Tencent Hunyuan3D, official WorldClaw project page](https://tencent-hunyuan.github.io/Hunyuan3D-WorldClaw/)
3. [Tencent-Hunyuan, Hunyuan3D-WorldClaw repository](https://github.com/Tencent-Hunyuan/Hunyuan3D-WorldClaw)