Do not take Nvidia’s word for it – try Nvidia’s amazing new image comparison tool instead

Nvidia wants you to know that AMD’s FidelityFX Super Resolution (FSR) is a pure pretender, nothing like the Nvidia Deep Learning Super Sampling (DLSS) technology that its GPUs can use to increase the frame rate and / or image quality of Your games. To prove it, Nvidia is trying an interesting new tactic by releasing a remarkably powerful image comparison tool called Nvidia ICAT, for free, so everyone can see for themselves.

ICAT, an acronym for “Image Comparison And Analysis”, is a very simple program that allows you to view a stack of gameplay screenshots and even recorded videos side by side, automatically creating the frames and allowing you to pan smoothly over each scene, zoom in and out, on all of these photos or videos at the same time. It has a drag-and-drop interface that makes pixel-looking a child’s play, and like well-known pixel-peepers, you’d better believe The edge going to use it for more than just gaming graphics!

Nvidia’s new ICAT tool in action. There is also a slider mode.
Screenshot of Sean Hollister / The Verge

But Nvidia’s hope is again to show how bad AMD’s FSR is compared to DLSS – or even Nvidia’s own, separate spatial upscaling technology.

If you need a short primer: Spatial upscaling Techniques such as AMD FSR run each of your game’s video frames one at a time through a fixed algorithm – and they do not require a special GPU to run. Nvidia’s DLSS, on the other hand, is one temporally upscaling technology that compares multiple frames and takes into account how things move in a video game scene, and processes all of it using a neural network running exclusively on the Tensor cores that you can only find in an Nvidia RTX GPU.

Nvidia illustrates how temporal and spatial upscaling differ.

But before you say “it’s great Nvidia, but I can actually not buy your new GPUs!” you should probably know that Nvidia also has its own spatial upscaling, called Nvidia Image Scaling. It has apparently been buried in the Nvidia Control Panel for some time now, and today Nvidia is opening it on GitHub with its own SDK and support for all GPU brands. Developers can integrate it into their game if they want, just like AMD’s FSR.

And the company’s baking Nvidia Image Scaling in its GeForce Experience app, so you can turn it on for any game and adjust a slider for in-game sharpness (so you can see the difference) using Nvidia’s overlay.

Is it as good as Nvidia’s DLSS? Not even close, and Nvidia is the first to admit it. In a briefing with The edge, product manager Henry Lin hammered us with example after example of how DLSS (especially the new DLSS 2.3, which suffers from minor ghosts around moving objects) not only beats the pants of Nvidia’s own spatial upscaling, but sometimes performs better than an original picture. . Especially when you look at thin objects with many edges that tend to flicker:

The communication bowl is crispier and less aliased with DLSS.

The depicted communication bowl is one place DLSS has an advantage.
Image: Nvidia

Personally, I swore to native resolution all the way – but Nvidia’s ICAT made me curious enough to give it a try. So I fired up Deathloop and Back 4 Blood, two games offering both AMD FSR and Nvidia DLSS, on a 4K display in their highest quality.

IN Back 4 Blood, I felt confirmed. Each part of the image looked fuller, sharper and higher resolution at built-in 4K, up to and including the roof-mounted antenna:

You need to press to zoom and / or download, this is a 4K image.
Screenshot of Sean Hollister / The Verge

There were even parts of the scene where the AMD FSR clearly beat the Nvidia DLSS, like all these wood textures:

Perhaps Nvidia’s neural network needs more experience with wood.
Screenshot of Sean Hollister / The Verge

But when I tried Deathloop, it was the opposite story: almost everywhere I looked, resolved and preserved DLSS details that I could not even see at native resolution, and with far, far less flicker from distant alias objects.

In Deathloop, the VW bus in the background definitely looks less cartoonish with DLSS, and you can see far more details in distant textures.
Screenshot of Sean Hollister / The Verge


Native 4K, inflated to show details.
GIF by Sean Hollister / The Verge


DLSS Quality, in the same size. Less flicker, more details.
GIF by Sean Hollister / The Verge

That said, I still saw moments in Deathloop where DLSS also does weird things. Scroll up to the top of this post for an example: you can see how DLSS totally changed the glittery finish of the Strelak Verso guns and muddy the texture of the soil in the exact same scene as my other screenshots and GIFs.

I do not quite believe in DLSS yet, but I am fascinated. Nvidia’s promise is that with enough learning, games can actually look better and run faster than native with DLSS turned on. So far, this is something developers need to activate per game, and not all games have the same version to offer, but that may change along the way. Nvidia’s Lin tells me Cyberpunk 2077 already has over-the-air DLSS updates – when you open the game, it can check with the Nvidia driver to see if there is a new version, and change it quickly at launch.

“Wherever you see that DLSS does not match native, that’s what we want to work on,” says Lin.

We’ll add a link to Nvidia’s new ICAT tool when we see it live on the company’s website.

Leave a Comment