Intel launches XeSS 1.2 promising visual and performance improvements

Intel has released a new and improved version of XeSS, but has failed to open source their technology as promised

On the company’s Github, Intel has released a new version of XeSS called XeSS 1.2, adding support for dynamic resolution scaling to XeSS alongside new performance optimisations, bug fixes, and stability improvements.

With the release of XeSS 1.2, Intel has improved both the quality and the performance of their Deep Learning based Temporal Super Sampling technique, which should come as great news for users of Intel ARC graphics cards. Unlike Nvidia’s DLSS technology, Intel’s XeSS upscaler works on AMD, Nvidia, and Intel graphics cards, with XeSS running at its best on Intel GPU due to its use of an optimised codepath to deliver stronger performance and better visual results.

As pointed out by Phoronix, Intel has failed to open source XeSS as promised, despite Intel’s promises to open source their technology in the past. This is hugely disappointing for users who want to customise or improve Intel’s AI upscaling technology, or simply want to dive into the code and see how XeSS works.

Intel’s main libraries for XeSS remains closed source, preventing users of XeSS from modifying or viewing Intel’s code. Until Intel open sources this code, Intel will not have lived up to their promises for XeSS.

You can join the discussion on Intel’s XeSS 1.2 AI upscaling technology on the OC3D Forums.