AMD announced on the availability of the initial texts of the updated implementation of the super-seamping technology fsr 2.2 (Fidelityfx Super Resolution), using spatial scaling algorithms to reduce the image quality loss with increasing scale and transforming into higher resolution. The code is written in C ++ and spreads under the MIT license. In addition to the basic API for C ++, the project provides support for graphic API DirectX 12 and Vulkan, as well as the HLSL and GLSL shader languages. A set of examples and detailed documentation is provided.
FSR is used in games to scaling the output on screens with high resolution and allows you to achieve quality close to the initial resolution, ensuring the preservation of details of textures and clear faces due to the reconstruction of small geometric and raster parts. Using settings, you can balance between quality and speed.
The technology is compatible with various GPU models, including integrated chips.
The new version has significantly increased the quality of the generated images and work was carried out to eliminate artifacts, such as the appearance of flicker and halo around fast -moving objects. Changes in the API have been made, which may require changes to the application code in which masks generation functionality is used. The “Debug Chekker” mechanism is presented, which allows to simplify the integration of FidelityFx Super Resolution with the application in the debugging assembly (after turning on the debt messages are transmitted from FSR Runtime to the game, which simplifies the diagnosis of problems).