UE5 Experimental AI Control via MCP Proves Practical in Test
The test shows AI-driven Unreal Engine operation has moved past the experimental stage into practical production use, shortening work that would be time-consuming for human operators.
Reporting from 1 source: ASCII.jp.
Epic Games' Unreal Engine 5.8 added MCP support as an experimental feature in June, letting AI operate the engine. Shin Seishi tested it with GPT-5.6 Sol in the Codex environment and found it practical for event scene creation. The test produced a demo video for Math Magic's Hi3D v3.0 AI 3D modeling service, showing tasks that take humans time and effort can be shortened significantly.
Unreal Engine 5.8 shipped with MCP as an experimental feature in June, and the Codex application environment let Shin Seishi put it to work. Operating the engine through GPT-5.6 Sol proved practical for building event scenes, cutting down tasks that would normally take humans considerable time and effort.
The test was a promotional video for Hi3D v3.0, Math Magic's updated AI 3D modeling service. A single image generates 3D models that infer invisible parts, with higher accuracy from multi-view input planned for launch. Seishi generated a robot, tank, and ruined building with Hi3D v3.0, then brought them into UE5 real-time CG to compare against an earlier Blender MCP previs battle video.
Synthesized by Yomimono from the 1 cited source below, including Japanese-language reporting where cited, then editorially reviewed before publishing.