NVIDIA launches AI Blueprint for faster 3D object generation
NVIDIA has rolled out a new AI Blueprint for 3D object generation, designed to remove the heavy technical lifting from 3D prototyping workflows. Traditionally, artists spend hours building placeholder assets just to populate scenes, but the new pipeline uses generative AI to automate much of that process through the in
NVIDIA has rolled out a new AI Blueprint for 3D object generation, designed to remove the heavy technical lifting from 3D prototyping workflows.
Traditionally, artists spend hours building placeholder assets just to populate scenes, but the new pipeline uses generative AI to automate much of that process through the integration of several tools from Team Green and Microsoft.
First, an LLM powered by Llama 3.1 8B NIM generates ideas and object suggestions, while NVIDIA SANA produces high-quality image previews. Once approved, the Microsoft TRELLIS NIM microservice converts previews into production-ready 3D assets, running 20% faster thanks to PyTorch optimizations.
Artists can generate up to 20 objects per prompt and export them directly into Blender or other popular 3D applications for further refinement. By pre-packaging these tools, NVIDIA simplifies what would otherwise be a complex and time-intensive setup, making it easier for professionals in gaming, architecture, and digital media to accelerate their workflows.
The Microsoft TRELLIS NIM microservice is supported on NVIDIA GeForce RTX 50 Series and 40 Series GPUs with at least 16GB of memory, delivering measurable time savings that add up significantly across large-scale projects.
An ASUS gaming PC can run games, everyday work and local AI agents on the same hardware. Here is where local execution can improve data control and cost predictability — and where...
GIGABYTE’s WRX90 AI TOP combines seven PCIe 5.0 x16 slots, up to 2TB of ECC memory, dual 10GbE and IPMI management for expandable local AI workstations.
Decathlon China and Tencent Smart Retail are exploring AI applications for customer service, business decisions, employee productivity and store operations.
Traditionally, artists spend hours building placeholder assets just to populate scenes, but the new pipeline uses generative AI to automate much of that process through the integration of several tools from Team Green and Microsoft.
First, an LLM powered by Llama 3.1 8B NIM generates ideas and object suggestions, while NVIDIA SANA produces high-quality image previews. Once approved, the Microsoft TRELLIS NIM microservice converts previews into production-ready 3D assets, running 20% faster thanks to PyTorch optimizations.
Artists can generate up to 20 objects per prompt and export them directly into Blender or other popular 3D applications for further refinement. By pre-packaging these tools, NVIDIA simplifies what would otherwise be a complex and time-intensive setup, making it easier for professionals in gaming, architecture, and digital media to accelerate their workflows.
The Microsoft TRELLIS NIM microservice is supported on NVIDIA GeForce RTX 50 Series and 40 Series GPUs with at least 16GB of memory, delivering measurable time savings that add up significantly across large-scale projects. 
