Nvidia does not stop working on new features that improve the user’s graphic experience, and the new thing we have is called RTX Video Super Resolution . This image upscaling function takes as its starting point the famous DLSS technology specifically for games. Well, the Tensor cores of the Nvidia RTX 30 and 40 cards are also used to develop the AI that will allow rescaling streaming videos at low resolution to view them with higher quality in the most used browsers and on 1440p or 2160p monitors.
Index of contents
- Que es RTX Video Super Resolution
- How to enable RTX Video Super Resolution
- Pruebas con RTX Video Super Resolution
- Conclusions about Nvidia RTX Video Super Resolution
In this article we will see what this technology is about in detail, how to activate it on our equipment and, of course, performance tests to see if we actually obtain improvements in your application. In the new graphic era, not everything involves consuming the content at the highest possible resolution, consuming a lot of network resources, since solutions like this can help a lot to obtain what we are looking for by taking advantage of our graphics cards.
Que es RTX Video Super Resolution
Nvidia RTX Video Super Resolution is a technology capable of improving the quality of streaming or consumed videos in real time using the image rescaling technique using AI (Artificial Intelligence). This means that the technique will take the video in the resolution that is being broadcast, for example 1080p or even 480p, and will rescale it to the native resolution that we have on our monitor.
The process is done using the Tensor cores of the graphics cards, and in principle it will only be compatible with the Nvidia RTX 30 and RTX 40 , leaving out the RTX 20 despite the fact that they also have this type of core. This feature will be implemented directly in Nvidia GeForce Game Ready drivers starting with version 531.14 . It will have 4 quality levels , with each one the consumption of graphic resources will increase a little.
It will be compatible with Google Chrome browsers starting with version 110.0.5481.105 and Chromium-based Microsoft Edge starting with version 110.0.1587.56 . We can also use it with the main streaming video platforms such as YouTube, Disney+, Netflix, Prime Video, Hulu and of course Twitch.tv , one of the most widely used networks today where broadcasts are at 1080p in most cases. However, it supports resolutions from 360p to 1440p.
The idea of RTX Video Super Resolution is to process streaming video by removing the artifacts that are generated due to compression to be transmitted over the network, improving the sharpness and clarity of the image. This will be especially interesting for users who have a monitor with a higher resolution at which the videos are consumed, to carry out a higher quality scaling. It will also be interesting in lower bandwidth connections where we have to see the content, for example, in 720p on a 1080p monitor or higher.
How to enable RTX Video Super Resolution
To activate RTX VSR we need an Nvidia RTX 40 or 30 graphics card and have Game Ready drivers 531.14 or higher installed, as well as up-to-date supported browsers.
Once the drivers are installed, we will open Nvidia Control Panel , either from the icon on the taskbar, or by right-clicking on the desktop and accessing it.
We are going to Adjust the video image settings , and on the right side we will find the new option RTX Video Enhancement .
We activate the check and the 4 quality levels will be enabled to use the one that best suits us, and then click on Apply to activate the function.
If we are already playing a video through our browser, it is recommended to reload the video so that the effect of the algorithm is applied, or ultimately we can close and reopen the browser.
Pruebas con RTX Video Super Resolution
Once we have RTX Video Super Resolution activated, it is time to test with some videos from the internet to see if the image quality improvement is effective. For this we will use the following test bench:
- CPU: Intel Core i9-12900K
- GPU: Nvidia RTX 4080
- Base plate: Asus ROG Maximus Z790 Hero
- Monitor: Viewsonic VX3211 4K mhd
We have chosen some videos from YouTube and Twitch where there is great detail in the image to see how the algorithm works rescaling from 1080p to 2160p in full screen , using the maximum quality mode 4.
RTX-VSR OFF
RTX-VSR ON
RTX-VSR OFF
RTX-VSR ON
RTX-VSR OFF
RTX-VSR ON
RTX-VSR OFF
RTX-VSR ON
Full image captures
Even with the naked eye in uncompressed images it is difficult to see the differences clearly, so we have cropped an area of these images to look for that difference. We can see a slight improvement in the sharpness of the edges of the trees , but the truth is that it is not noticeable even in static images, so during the playback of the video we will notice little improvement .
Where a large number of textures are concentrated, as is the example of the last image, zooming in we do notice how in the image with RTX Video Super Resolution it maintains greater detail .
We leave you several video captures with the function activated and deactivated, to see how the improvement is once again very subtle and works where there are a large number of textures, for example in trees or water. The capture has been made with the AV1 encoders of the graphics card at a bit rate of 90 Mbps in 4K . We recommend watching this video in 4K so that these differences can be noticed, since in 1440p or 1080p you will not notice practically anything.
In a Twitch broadcast , where the image comes directly in streaming (last part of the video where an IRL is shown in Japan) we can see improvements in the texture of the trees and smoothness in the edges of the objects. The difference between quality level 1 and 4 is very low at least in this first release version.
Seeing these types of videos, we believe that a good complement to VSR would be the creation of intermediate frames as DLSS 3 does. In this way, videos recorded with low framerates, such as the ones we have chosen, could greatly improve their fluidity, being an improvement even more noticeable than the image quality itself.
RTX-VSR-OFF
RTX-VSR-Quality-1
RTX-VSR-Quality-4
Nvidia mentions that RTX Video Super Resolution will consume more resources the higher the quality level chosen . The average consumption with the algorithm deactivated has been 8%, while in quality mode 1 it increases to 11 to 14% and in quality mode 4 to 13 to 16%. Considering the enormous power of the RTX 4080, these are minimal figures , but in lower models it should be more noticeable, for example a 4060 or a 3060.
Conclusions about Nvidia RTX Video Super Resolution
In this first version, the results offer small improvements in the sharpness and detail of the textures, but it clearly does not mark a before and after, since it is barely noticeable with moving videos, at least in 1080p rescaling to 2160p. Where we see the most potential for it is in real-time streaming transmissions, for example Twitch , since it seems to better compensate for low bitrate, improving sharpness and softening edges.
As we have commented before, adding Frame Generation to the algorithm as in DLSS 3 could make a big difference , even more than the quality improvement itself. It would improve the fluidity of the video and eliminate the small jerks that exist in some broadcasts, especially when we have a high ping.
RTX VSR is one more step by Nvidia to bring native high-resolution image quality and performance closer to compressed streaming content and obviously limited in this regard. This function will receive improvements and optimization in successive versions, so we expect a promising future for it. It may be that in low-bandwidth connections or low-resolution videos it will become essential very soon.