Snynet Solution Logo
MON - SUN: 10 AM - 6 PM
+60 11 5624 8319

Blog

Nvidia’s AI-powered DLDSR allows RTX GPUs to make your games look better

Image Description

Nvidia has a new trick for its GeForce RTX graphics cards to help further improve the visual quality of games with only a (hopefully) minor performance hit.

We say new trick, but this is actually a refinement of an old feature, with Team Green now powering up its DSR (Dynamic Super Resolution) tech by adding AI into the mix, to make DLDSR (Deep Learning Dynamic Super Resolution, with Deep Learning meaning AI).

DLDSR (yes, it’s a clunky acronym) will arrive in Nvidia’s Game Ready Driver launching later this week (on January 14).

This is a downscaling tech (as opposed to DLSS which engages in upscaling), so that means DLDSR takes a game and renders it at a higher resolution, then intelligently shrinks it back down to the resolution your monitor is running at – preserving some of the extra detail in the process, and providing smoother edges (fewer jaggies) while reducing unpleasant artifacts like shimmering.

This is what DSR has always done – since it was introduced a good while back with GeForce 900 Series GPUs – but the difference now is adding AI fine-tuning to make the results better (with RTX cards that have the requisite Tensor cores).

Nvidia claims: “DLDSR improves upon DSR by adding an AI network that requires fewer input pixels, making the image quality of DLDSR 2.25X comparable to that of DSR 4X, but with higher performance. DLDSR works in most games on GeForce RTX GPUs, thanks to their Tensor Cores.”

Furthermore, Nvidia said it has also got together with ReShade author and modder Pascal Gilcher to release three advanced Freestyle filters in GeForce Experience, one of which will be SSRTGI (or the ‘Ray Tracing ReShade Filter’) that improves both lighting and shadow effects to make them more realistic. The broad contention is that using both DLDSR and SSRTGI, you can expect what’s effectively a ‘remastered’ version of a game (with Prey graphics shown as an example).

As well as delivering DLDSR, Team Green’s new driver will also provide optimization and DLSS support for God of War (as well as Nvidia Reflex support), the PC version of which launches on January 14, the same day as the driver. The same DLSS and Reflex support is being brought to Rainbow Six Extraction ahead of its January 20 release, too.


Analysis: This all sounds great, but performance levels will be key

DLDSR sounds like a big step forward for DSR, certainly if how well Nvidia has managed to implement and improve its AI-powered DLSS is anything to go by (although it took time and version 2.0 to really nail DLSS, of course).

The strength with this new feature is that unlike DLSS, games don’t need to be programmed to use it – as Nvidia puts it, ‘most games’ will just work with DLDSR to give them a nice visual boost.

That means PC gamers with a beefy graphics card playing games on a 1440p monitor, for example, will be able to turn on DLDSR, and it’ll render at 4K, then downscale back to native resolution, preserving some nice additional details on top of the default 1440p image. In short, this puts the extra power of their GPU (as it were) to good use to achieve higher image quality at that 1440p resolution, effectively taking a step up towards 4K with hopefully not much of a performance hit (that’s a key aspect, obviously).

Nvidia makes it clear that DLDSR will be much better than DSR in terms of performance, but we actually need to see this tech in action before we can judge it properly, naturally.

Date

12 Jan 2022

Sources


Share


Other Blog

  • Windows 10’s big redesign is seemingly pinching bits of Windows 10X

    With Windows 10X rumored to be shelved, Windows 10 could benefit from bits of its interface.

    Read More
  • Think your Zoom meeting is safe from Zoombombing? Think again…

    During the pandemic, Zoom has become an indispensable tool – but it’s also made it a target.

    Read More
  • C-V2X: Connected cars and the future of transport

    Cellular Vehicle-to-Everything, or C-V2X, continues to make strides as the preferred technology for vehicles to communicate with one another and with their surroundings.

    Read More
  • Quantum computing will assist search for life in deep space

    Enterprise quantum software company Zapata Computing has partnered with UK’s University of Hull to leverage each other’s expertise to detect signatures of life in deep space.

    The partnership will support research to repurpose Zapata’s quantum workflow platform Orquestra in order to assist in the development of highly accurate astrophysical models and applications.

    “Although quantum computers are an emergent technology and cannot yet outperform classical hardware, Zapata has made it possible to generate valuable insights from the Noisy Intermediate-Scale Quantum (NISQ) devices currently available,” said Dr. David Benoit, senior lecturer in Molecular Physics and Astrochemistry at the University of Hull.

    Dr Benoit added that Orquestra enables the researchers to build future-proof applications that don’t just work with NISQ devices today, but are also capable of leveraging the more powerful quantum computing devices of the future.

    Improving model accuracy

    Sharing details about hope the researchers plan to leverage Zapata’s quantum expertise, the researchers explain that they want to build on top of the work of MIT researchers who in 2016 drew up a list of over 14,000 molecules that could indicate signs of life in exoplanets’ atmospheres. 

    The University of Hull researchers now aim to generate a database of detectable biological signatures of these molecules by using new computational models of molecular rotations and vibrations. However, little is currently known about how these molecules vibrate and rotate in response to infrared radiation generated by nearby stars. 

    In order to detect them, the researchers need to build highly accurate models based on extremely accurate calculations, which is touted as one of the forte’s of quantum computing.

    “The research being done by Dr. Benoit and his colleagues has the potential to redefine our place in the universe, and we’re humbled that Orquestra will have a supporting role,” said Christopher Savoie, CEO and co-founder of Zapata Computing.

    Orquestra evaluation for the research is currently scheduled to run for eight weeks before the team publishes an analysis of the research. 

    Read More

Find Out More About Us

Want to hire best people for your project? Look no further you came to the right place!

Contact Us