Higgsfield Launches 3D Jutsu: AI-Native Browser Workspace for Scene Blocking and Previz

Higgsfield has launched 3D Jutsu, an AI-native 3D workspace that generates fully editable 3D scenes — including objects, layout, lighting, cameras, and animation — from a text prompt or reference image, entirely in the browser. An AI agent assembles the scene on Higgsfield's Supercomputer infrastructure, and users can then refine individual scene objects without regenerating the whole scene, before rendering directly to video from the same chat interface.

The tool targets the blocking and previsualization stages of production, stages traditionally handled in desktop 3D software. 3D Jutsu supports timelines, camera trajectory editing, a curated GLB asset library with Mixamo characters, revision history, and export to MP4, static frame, or GLB. The agent accepts plain-language instructions and can be driven by a range of LLMs, with collaboration features allowing scene sharing with role-based access.