NVIDIA's 'Bonsai Diorama' Gets Live with DLSS 4 and RTX Mega Geometry
In its latest tech demo, 'Bonsai Diorama', NVIDIA goes one step further in laying bare the new-age power of graphics technologies under its belt, such as the new DLSS 4, path tracing, and especially the image quality improvements brought along with RTX Mega Geometry. You can find it in the latest update for Unreal Engine's RTX branch (version 5.6.1).
Enjoy the 'Bonsai Diorama' Demo
If you're interested in running the demo, NVIDIA has indicated the minimum hardware and software requirements:
- Minimum GPU: GeForce RTX 2080 Ti
- Recommended GPU: GeForce RTX 5060
- Driver Version: GeForce driver 581.29 or newer
In case of loading the demo level, NVIDIA recommends setting resolution and window mode from the main menu. For GPUs equipped with 8 GB of VRAM, 1080p or 1440p resolution is supported, while those with 11 GB or more can handle 4K. Once inside, users can press 'H' to access the UI and toggle various elements, including RTX Mega Geometry, DLSS Super Resolution, Ray Reconstruction, and Frame Generation.
The Effects of RTX Mega Geometry
Previously, it has been seen mostly in games such as Alan Wake 2 as advantages in performance, the 'Bonsai Diorama' showcase mostly handles the imagery side of RTX Mega Geometry. NVIDIA articulates that this way Nanite geometry is also into beam-traced space: "BVH." This allows for the avoidance of many visual artifacts in ray-traced shadows and reflections that would have been there by default with Unreal Engine 5, as a result, leading to much clearer images and even more realistic scenes.
To elaborate on these technologies, NVIDIA is organizing a webinar for developers on Thursday, October 28, at 10 AM Pacific Time.
NVIDIA Avatar Cloud Engine (ACE) Modernized Updates
In a related development, NVIDIA announced a slew of impressive improvements to its Avatar Cloud Engine space. An open-source Qwen3-8B AI model is now available as a plug-in for the In-Game Inferencing (IGI) SDK. This incorporation enables developers to create more dynamic and intelligent game characters that can reason in real-time and react to non-scripted events and actions taken by players.
The core IGI SDK has also been improved in fundamental ways:
- MultiLORA Adapters: Allows for very lightweight tuning of full AI model weights at very low computational expense.
- CUDA Support for Vulkan: Per-pixel multitasking of graphics and AI workloads can be optimized for overall enhanced game performance.
