Discover and explore top open-source AI tools and projects—updated daily.
RareSenseEditable 3D asset generation with procedural part control
New!
Top 83.0% on SourcePulse
Nova3D generates structured 3D assets with distinct, editable parts from text prompts or reference images. It targets 3D artists, game developers, and engineers who require actionable models rather than monolithic meshes. The primary benefit is the preservation of part boundaries, enabling precise downstream editing, regeneration, or articulation of individual components.
How It Works
Nova3D employs a "Prompt-to-Code" or "Image-to-Code" paradigm, generating Blender-native Python scripts as its core output. This approach leverages Blender's scene graph to create assets with logical hierarchies, named parts, and proper parenting. Unlike diffusion-based generators that produce single, merged meshes, or CSG systems limited by organic shape complexity, Nova3D's script-native method is a superset, offering greater expressiveness, material support (PBR), and editability.
Quick Start & Requirements
git clone https://github.com/RareSense/Nova3D.git
cd Nova3D
flutter pub get
flutter run -d web-server --web-hostname 0.0.0.0 --web-port 5555 (access via http://127.0.0.1:5555).Highlighted Details
Maintenance & Community
Community channels include discord and twitter/X. Development activity can be tracked via github issues.
Licensing & Compatibility
The provided README does not specify a software license. This omission requires clarification regarding usage rights, particularly for commercial applications or integration into closed-source projects.
Limitations & Caveats
The client connects to a currently closed-source hosted backend service (nova3d.xyz). While self-hosting the backend is theoretically possible, it necessitates building the server infrastructure from scratch. The project relies on Flutter for its client-side UI.
1 day ago
Inactive