NVIDIA's Open-Source Image Scaling SDK 1.0 Released

Last week NVIDIA announced the Image Scaling SDK as an open-source, cross-platform GPU image upscaling implementation that with their own hardware makes use of DLSS. Following the brief exposure over the past week, NVIDIA Image Scaling SDK 1.0 has been formally christened.

The NVIDIA Image Scaling SDK can work on the likes of Intel and AMD Radeon hardware via the SDK's generic compute shaders that are MIT licensed. Integrating the NVIDIA Image Scaling SDK does require integration on part of the game/engine developer.

As the project's code repository puts it, "The NVIDIA Image Scaling SDK provides a single spatial scaling and sharpening algorithm for cross-platform support. The scaling algorithm uses a 6-tap scaling filter combined with 4 directional scaling and adaptive sharpening filters, which creates nice smooth images and sharp edges. In addition, the SDK provides a state-of-the-art adaptive directional sharpening algorithm for use in applications where no scaling is required. The directional scaling and sharpening algorithm is named NVScaler while the adaptive-directional-sharpening-only algorithm is named NVSharpen. Both algorithms are provided as compute shaders and developers are free to integrate them in their applications. Note that if you integrate NVScaler, you should NOT integrate NVSharpen, as NVScaler already includes a sharpening pass."

The only change that landed in this tagged version since the original open-source publishing last week was fixing the indexing precision and scale normalization code.

The formal NVIDIA Image Scaling SDK 1.0 release can be downloaded here.