Higgsfield launches 3D Jutsu for editable AI-generated 3D scenes

Higgsfield has added a new 3D creation surface called 3D Jutsu. The browser workspace starts from a text prompt or reference and produces a scene whose objects, layout, lighting, cameras and animation can be edited directly before the result is rendered to video.\n\nThat distinction matters because many generative-video workflows treat each change as another full generation. Higgsfield says 3D Jutsu keeps scene elements editable, so a user can move objects, adjust cameras or lighting and refine the environment without rebuilding the entire scene. The company positions the workspace for scene blocking and previsualization as well as final video generation.\n\nThe workflow is agent-driven through Higgsfield Supercomputer. Users can choose an LLM for the scene-building chat, with GPT-6 Astra among the supported options, while the agent assembles and revises scene data in response to natural-language instructions. Finished scenes can move into video generation from the same workspace.\n\nHiggsfield's September 4 changelog is the primary evidence for the release and feature scope. Same-day independent launch tracking also corroborates that 3D Jutsu is newly live. Those sources establish availability and workflow, not independent output-quality benchmarks.\n\nAiToolMap should evaluate 3D Jutsu as a distinct product surface before cataloging it: the next research step is to verify pricing, export and collaboration limits, data handling, exact account requirements and whether it should be represented as a standalone tool or a Higgsfield suite feature.