Microsoft has released a new 4B parameter model that can convert any image into a 3D asset in just three seconds, utilizing a new geometry format called O-Voxel. This technology outputs GLB files complete with full PBR textures, ensuring compatibility with popular tools like Blender, Unity, and Unreal. The model is fully open source, aimed at encouraging experimentation among indie developers and researchers, ultimately addressing a significant gap in content creation workflows where accurate object orientation is crucial for video and virtual production pipelines.

GLB: GLB is the binary version of the glTF 3D file format, commonly used for compact, efficient transmission and loading of 3D models with embedded materials and textures. Here, Microsoft’s model outputs GLB files with full PBR textures, making the generated 3D assets immediately usable in tools like Blender, Unity, and Unreal without additional conversion steps.
Unity: Unity is a widely used real-time 3D engine and development platform for games, interactive experiences, and simulations across many devices. The news notes that Microsoft’s model produces GLB assets ready for Unity, meaning developers can quickly bring AI‑generated 3D objects into their projects for prototyping, level design, or content production.
Unreal: Unreal Engine is a high-end real-time 3D engine used for games, virtual production, and immersive experiences, known for its advanced rendering and cinematic tools. In this context, Unreal is one of the engines that can directly use the GLB files from Microsoft’s new model, enabling creators to drop AI‑generated, PBR‑textured assets into Unreal scenes with minimal setup.
Blender: Blender is a popular open-source 3D creation suite used for modeling, animation, rendering, and visual effects in both indie and professional pipelines. In this news, Blender is one of the primary target tools for the GLB outputs produced by Microsoft’s image‑to‑3D model, allowing artists to import and refine the auto‑generated assets directly in their existing workflows.
HowToAI: HowToAI is an online creator and news-style account focused on explaining, reviewing, and showcasing new artificial intelligence tools, models, and workflows for a broad audience. In this news item, HowToAI is the source being quoted about Microsoft’s open-source image‑to‑3D model, summarizing its capabilities and performance for followers interested in generative AI and 3D content creation.
O-Voxel: O-Voxel is a newly introduced geometry representation format designed to efficiently convert images into 3D assets and then into textured meshes suitable for real‑time rendering and content creation. In the context of this news, O‑Voxel is the core format used by Microsoft’s new model to produce GLB files with full PBR textures extremely quickly for use in common 3D engines and tools.
Microsoft: Microsoft is a global technology company that develops software, cloud services, and AI infrastructure, and has been increasingly open-sourcing AI models and tooling for developers and creators. In this news, Microsoft is highlighted for releasing an open-source 4B-parameter model that can rapidly convert 2D images into 3D assets using a new O-Voxel geometry format, enabling faster, more controllable workflows for video and 3D content studios.

Open_source_3D_AI: Recent coverage of Microsoft’s image‑to‑3D release emphasizes that making the model fully open source is intended to spur experimentation by indie developers, researchers, and 3D tool vendors rather than locking the capability into a single ecosystem.
Ecosystem_integration: Developer commentary over the past few weeks has noted a growing trend for new 3D and generative models to output in formats like GLB and integrate smoothly with Blender, Unity, and Unreal to reduce friction between AI tools and existing production environments.
Content_creation_workflows: AI researchers and creators have recently highlighted that controllable 3D asset generation, especially with accurate object orientation and material detail, is becoming a key missing piece in many video and virtual production pipelines.