Sharing — RTX Video Super Resolution

RTX Video Super Resolution (VSR) by Nvidia

Sik-Ho Tsang
2 min readMar 6, 2023
Left: Blurry Image, Right: Much Sharper Image (Look at the Edges) (Figure from Nvidia)

Pixel Perfect: RTX Video Super Resolution Now Available for GeForce RTX 40 and 30 Series GPUs
28 Feb 2023, by Nvidia

  • Deep learning based VSR has come to end-user applications in a very handy way now. I have also reviewed some of the VSR papers before:
  • 2017 [STMC / VESPCN] 2018 [VSR-DUF / DUF] 2019 [EDVR]


Nvidia RTX VSR (From 1080p to 4K) (Figure from Nvidia)

1.1. What is VSR Doing?

  • According to Nvidia article, RTX VSR reduces or eliminates artifacts caused by compressing video — such as blockiness, and ringing artifacts around edges. It also sharpens edges and details.

1.2. Deep Learning VSR Model

  • It uses deep learning network that performs upscaling and compression artifact reduction in a single pass.
  • The network analyzes the lower-resolution video frame and predicts the residual image at the target resolution. This residual image is then superimposed on top of a traditional upscaled image, matches the output resolution.
  • The AI model is trained on a wide range of content with various compression levels. It learns about types of compression artifacts present in low-resolution or low-quality videos that are otherwise absent in uncompressed images as a reference for network training.
  • Extensive visual evaluation is employed to ensure that the generated model is effective on nearly all real-world and gaming content.
  • (I tried to find what exact algorithm and dataset they use, I cannot find them easily, maybe a VSR literature review is needed, or they even not publicize their algorithm. Please feel free to tell me if you know.)

2. VSR Enabling

NVidia Control Panel to Enable VSR (Figure from Nvidia)
  • To enable VSR, first, update the driver and restart the PC.
  • Then, there is an additional option as shown above.



Sik-Ho Tsang

PhD, Researcher. I share what I learn. :) Linktree: for Twitter, LinkedIn, etc.