NVIDIA has officially launched the NVIDIA RTX Global Illumination (RTXGI) SDK. The version one of the SDK will be helpful to multiple agencies, developers, researchers in deploying scalable solutions and help them benefit from the next-generation Ray Tracing without having to wait a long time and spend a lot to access the same.
In addition to the NVIDIA RTXGI SDK, the GPU maker has also released NVIDIA Texture Tools Exporter as well as Deep Learning Super Sampling (DLSS) 2.0. The new version of NVIDIA’s DDS texture compression tool is available both as a standalone application and as a plugin for Adobe Photoshop.
NVIDIA Launches Multiple Tools That Will Help Game Developers, Researchers And Other Benefit From Ray Tracing:
With the NVIDIA RTX Global Illumination (RTXGI) SDK v1.0 game developers, researchers, students, and artists will be able to deploy scalable solutions to leverage Ray Tracing without bake times, light leaks, or expensive per-frame costs. The SDK has several important and much-needed features that should significantly ease the development process. The NVIDIA RTXGI SDK comes with efficient memory layouts and compute shaders, support for multiple coordinate systems, and hooks for engine and gameplay events to prioritize lighting updates.
It is interesting to note that RTXGI uses real-time Ray Tracing to update lighting information, and moreover, the entire process happens in real-time. This completely eliminates the pre-computation and baking steps. Earlier and current-generation illumination solutions or platforms take a significant time to render the details.
— Neowin (@NeowinFeed) March 23, 2020
The NVIDIA RTXGI SDK temporally accumulates and filters lighting and distance information in real-time with its probe-based data structure. This reportedly creates a hyper-realistic, multi-bounce lighting cache, complete with visibility information. NVIDIA assures the new SDK v1.0 ensures no light or shadow leaking out of the box. This means the platform does not need UV parameterization or probe blockers. Developers who avail access to the SDK early on will also receive automatic probe placement and dynamic performance optimization.
NVIDIA RTXGI SDK v1.0 can work on any DXR-enabled GPU. In other words, developers and researchers can use any of their existent NVIDIA GeForce RTX 20 Series, GTX 1660 Series, and GTX 10 series. Although RTXGI doesn’t yet work with Unreal Engine 4 or Unity, NVIDIA has indicated it is working with Epic Games and Unity to bring support for RTXGI to these game engines.
Nvidia launches Deep Learning Super Sampling (DLSS) 2.0 Which Will Boost AI Rendering:
In addition to the NVIDIA RTXGI SDK v1.0, the company also launched Deep Learning Super Sampling (DLSS) 2.0. It is essentially a robust artificial neural network that uses Nvidia RTX Tensor Cores. The primary agenda is to boost frame rates and generate sharp frames. NVIDIA aims to achieve better results than native rendering.
— TendenciasTech (@TendenciasTech) March 23, 2020
NVIDIA claims the DLSS v2.0 was imparted intense training by making it go through “tens of thousands of high-resolution images”. These images were reportedly rendered offline inside a supercomputer at very low frame rates at 64 samples per pixel. Using such input methods, DLSS 2.0 is able to take lower-resolution images and construct high-resolution images. Relying on such a trained model, NVIDIA then distributes the same RTX-based PCs via NVIDIA drivers and OTA updates.
DLSS 2.0 has three image quality modes for any game’s internal rendering resolution: Quality, Balanced, and Performance. The Performance Mode allows scaling from 1080p to 4K on the fly. Turing’s TensorCores are capable of providing up to 110 teraflops. Needless to add, this makes DLSS 2.0 twice as fast as its predecessor. Using such a computational power, developers can effectively run both intensive 3D games alongside a deep learning network at the same time.