AMD FSR 2.0 Source Code Hits GPUOpen With Samples And Demo, Right here’s How To Strive It
On: June 23, 2022
NVIDIA’s DLSS is a kind of few applied sciences that comes alongside occasionally and legitimately delivers the seemingly unimaginable: extra with much less. AMD’s first try at difficult DLSS wasn’t fairly as superior; whereas the unique FidelityFX Super Resolution had its benefits, the picture high quality wasn’t at all times in the identical ballpark.
FSR 2.0 is an entire totally different story, although. While it nonetheless does not precisely match DLSS in each occasion, it appears improbable, and it runs on any graphics processor able to DirectX 12 or Vulkan. AMD promised that it could take “five minutes” to implement in case your game already helps DLSS, and certainly a number of titles have picked it up already.
The warehouse scene is far much less demanding than the Sponza palace scene.
That mentioned, in case you’re not enjoying God of War, Deathloop, or Farming Simulator 22, you most likely could not have seen FSR 2.0 in motion. If you would like to take action for your self, head over to the GPUOpen mission and download the pattern bundle. It’s the button labeled “FSR 2 sample.” This 917MB bundle contains pattern functions for each DirectX 12 and Vulkan.
Using the pattern apps, you possibly can fly across the acquainted Sponza palace scene or a less complicated deserted warehouse scene, whereas fidgeting with the lighting in addition to the tone mapping and numerous types of upscaling. DLSS is not supported, sadly, however you possibly can evaluate FSR 2.0 towards native decision rendering, FSR 1.0, or easier upscalers utilizing bicubic sharpening, bilinear filtering, or straight-up level sampling.
Deathloop element comparability. Clockwise from left: Native, FSR1, DLSS, FSR2.
This pattern bundle is merely a aspect impact of the actual information at the moment, which is that AMD has launched the source code for FSR 2.0. That implies that builders engaged on DirectX 12 or Vulkan functions can independently combine the sensible upscaler into their very own software program. Doing so is nontrivial, however as talked about above, the identical work applies to implementing DLSS, so it is finally a win for everybody. You can discover the source on Github.
