As an upgraded version of Spatial Upscaling technology, NVIDIA has recently introduced a new Image Scaling technology. The new algorithm uses a 6-tap filter with 4 directional scaling and adaptive sharpening filters to improve performance. Its sharpening and scaling occur in a single thread. According to NVIDIA, this Makes it more efficient than similar methods.
NIS (NVIDIA Image Scaling) technology optimizes the game screen at the driver layer, so it only supports a single screen and must be in full-screen game mode to take effect. The following is a comparison from Necromunda: Hired Gun, which compares the three zooming techniques with the 4K native rendering of the game. When using NVIDIA Image Scaling and scaling technology, the resolution is reduced to 2955×1622, but the fps value is indeed increased to a higher and playable level.
NVIDIA’s Image Scaling (NIS) technology does not rely on any proprietary software, so regardless of your graphics card manufacturer (AMD, or even Intel), its implementation in the game is available. However, no game currently provides native support for NIS, and AMD or Intel users obviously cannot rely on NVIDIA’s control panel or GeForce Experience software to activate it.
Join RealMi Central on Telegram, Facebook & Twitter
However, this can be circumvented by programs such as Lossless Scaling. The 1.6 version now supports NIS in addition to FSR. Lossless scaling is not free software (there is a demo version, but the full version is priced at $3.99), but if you already have the software, there is now another option for you to try.
By the way, this version of NVIDIA image scaling is not as good as the native one, and the developers of lossless scaling frankly admitted this on the Steam forum.
- It uses the official shader, but FP16 support is not enabled for the time being, because it has been confirmed that it cannot work on RTX 3000 for some reason.
- In addition, I don’t know what implementation method they use in the driver. What is certain is that they will not waste GPU resources on image capture.