NVIDIA has introduced a new AI Blueprint designed to revolutionize how 3D artists populate digital scenes, dramatically reducing the time and effort required for prototyping. The AI Blueprint for 3D object generation enables artists to create up to 20 fully realized 3D assets from a simple text prompt, streamlining the creative process from concept to usable model. Traditionally, 3D artists spend significant time building low-fidelity placeholders to test scene layouts, often discarding much of this early work. This repetitive process slows down creativity and diverts attention from artistic exploration. The new NVIDIA AI Blueprint addresses this by integrating generative AI directly into the workflow, automating the initial phase of asset creation. The blueprint begins with a text prompt—such as “night market” or “sunny day at the beach”—which is processed by a built-in large language model powered by the Llama 3.1 8B NVIDIA NIM microservice. This model generates creative ideas and suggests relevant objects to include in the scene. Artists can then view 20 high-quality image previews of potential assets, instantly generated using NVIDIA SANA, a text-to-image framework that produces detailed, resolution-optimized visuals. Each preview can be refreshed, replaced, or deleted with a simple click. Once satisfied, artists can convert any image into a 3D model using the Microsoft TRELLIS NVIDIA NIM microservice. This cutting-edge model, developed by Microsoft Research, generates richly detailed 3D assets with complex geometry and textures from text or image inputs. When used within the AI Blueprint, TRELLIS runs 20% faster thanks to PyTorch optimizations, saving around 6 seconds per object on an NVIDIA GeForce RTX 5090 GPU. For artists creating hundreds of assets, this efficiency gain is substantial. The generated 3D models are automatically exported to Blender, an open-source 3D platform, allowing immediate refinement. Artists can also export to other major 3D applications, ensuring seamless integration into existing pipelines. Setting up such a workflow manually would require deep technical expertise, model selection, and configuration. NVIDIA AI Blueprints eliminate this complexity by providing pre-tested, optimized pipelines that run efficiently on NVIDIA GeForce RTX and RTX PRO GPUs. The Microsoft TRELLIS NIM microservice is compatible with all NVIDIA GeForce RTX 50 Series and 40 Series GPUs, both desktop and laptop, provided they have 16GB or more of memory. To get started, users simply load the AI Blueprint, enter a scene idea, and let the system generate 20 visual concepts. From there, they can refresh, remove, or convert individual assets—or use the “Convert all to 3D” button to process the entire batch. The final models are ready for export to Blender with a single click. Detailed setup guides and downloadable resources are available on NVIDIA Build and GitHub. Additional AI Blueprints, including those for 3D-guided generative AI, are also available for exploration. At the upcoming IFA trade show on September 6 at 2pm CET, attendees can join a session titled “How AI Is Changing Content Creation” to hear NVIDIA’s Sean Kilbride and other experts discuss real-world AI applications in creative workflows. The RTX AI Garage blog series continues to spotlight community-driven AI projects, covering everything from AI agents and productivity tools to creative apps built on AI PCs and workstations. Stay updated by following NVIDIA on Facebook, Instagram, TikTok, and X, subscribing to the RTX AI PC newsletter, or joining the NVIDIA Discord community. For industry insights, follow NVIDIA Workstation on LinkedIn and X.