Asset3D Factory

Asset3D Factory turns a single image into an import-ready 3D asset through a GPU-accelerated generation pipeline. It combines containerized AI services with a Unity client, allowing artists and developers to generate, inspect, and compare GLB models without leaving their workflow.

Image-to-3D Generation Pipeline

• Supports Stability AI Stable Fast 3D and Microsoft TRELLIS.2 generators
• Removes image backgrounds automatically before model generation
• Produces textured GLB assets ready for use in Unity
• Runs each model service in a CUDA-enabled Docker container

Unity Workflow

The Unity module provides a practical interface for choosing an input image and configured generator, then importing the completed model directly into the scene. It also tracks generation details including service readiness, duration, file size, and triangle count.

Model Review and Comparison

Generated assets can be browsed with their metadata and source-image thumbnails. The client supports reloading saved results, inspecting models with Unity's Input System, and comparing a saved model against the current generation side by side.

Service Architecture

FastAPI services expose consistent health, readiness, and status endpoints so the Unity client can monitor both generators through the same integration. The backend handles asynchronous requests and reports loading, ready, busy, and error states for clear feedback during generation.

Technical Focus

Built as a local, GPU-optimized tool for experimental and production-oriented 3D asset workflows. The project brings together Python AI services, Docker-based deployment, and Unity tooling into a single image-to-3D workflow.