Tencent Hunyuan Introduces WorldClaw for Agentic 3D World Generation
智东西 · wechat · 2026-08-13
Tencent's Hunyuan 3D team introduced WorldClaw, a framework that generates large-scale, editable 3D open worlds from a single text prompt.
WorldClaw uses a coarse-to-fine, agentic approach across four main stages:
- Intent Analysis & Planning: Multiple agents parse user intent and generate a comprehensive scene layout.
- Global Terrain Generation: Builds continuous 3D heightfields using procedural content generation (PCG) based on semantic layouts.
- Regional Object Generation: Renders specific areas to place objects like buildings, converting them into editable 3D meshes using SAM3D.
- Scene Refinement: Agents automatically detect and fix geometric issues like floating objects or misaligned poses via multi-view rendering.\n
The team demonstrated complex scenes like tropical islands and desert battlefields, noting that the generated assets can be freely edited and exported to game engines.
Related event: Tencent Hunyuan Introduces WorldClaw Text-to-3D World Framework(3 posts)→
More from Multimodal
- StableVQ: three lightweight fixes for stable vector-quantized tokenizer training — Kwai-Kolors · 2026-09-23
- Comfy Desktop ships early-stage performance testing for local ComfyUI instances — Lexius2129 · 2026-09-23
- MiniMax H3 powers retro-style short film honoring Earth, Wind & Fire — TheChuckTone · 2026-09-23
- Swarm-built inference engine runs Qwen Image-2.1: 1K images in under 0.5s — bingxu_ · 2026-09-23
- Gemini image generation stops refining and regenerates whole scenes on follow-up edits — asm99 · 2026-09-23
- Abliterated Qwen-Image text encoder GGUF quantization trends on Hugging Face — pottokao · 2026-09-23