Menu Menu
[gtranslate]

NVIDIA’s DLSS 5 was never going to be well received

While aversion to it isn’t unanimous, NVIDIA’s DLSS 5 has received far more flack than praise. The AI graphics filter is being slammed as a slap in the face to game designers and a sad indictment of how generative tools are taking over creative spaces.

People are already thirsty for Leon. He doesn’t need a yassify filter.

If you missed NVIDIA’s recent demo, the company unveiled DLSS 5 as the next big leap in AI-assisted gaming performance – and the supposed ‘greatest’ advancement in gaming visuals since Ray Tracing.

Put simply, DLSS 55 takes frames a game has already rendered, reads underlying data like object movement and scene depth, then uses AI to generate extra lighting, material, and image detail to be overlaid in real time.

Unfortunately, the reality is less sophisticated than it sounds. When toggled on, intricately designed character models like Resident Evil Requiem’s Grace Ashcroft are transformed into the sort of sheen-heavy, over saturated (both literally and figuratively) figures we’ve become accustomed to seeing on Snapchat or dodgy ad banners online.

While the technology’s parameters ensure that some essence of the original scene remains, the finer creative details are replaced by overly polished assets. Not only do the characters appear illuminated by hero lighting, as though an LED ring light is plugged in out of view, but atmosphere is also traded for a more ‘photo realistic’ reimagining of the environment.

In the case of Resident Evil Requiem, immersion is not only broken by a yassified Grace and muted light sources that no longer bounce off the surrounding objects, but also by the AI erasing the natural effects of the gloomy, low visibility weather in the original game. In other words, intricate design choices are being overridden by an MO to make everything appear more ‘real’.

The Hogwarts Legacy example shows just how jarring this has the potential to be. The game’s character models are intentionally designed to have a semi-realistic yet animated feel, but the DLSS 55 clumsily plonks a photo-realistic face – complete with unnecessarily plump lips and shimmering pores – on the character’s neck. The result isn’t cinematic. Hell, uncanny valley wouldn’t even cut it as a description. It just looks utterly ridiculous.

There are examples on social media where the odd dull character model is given some life by the technology, but given the majority selected for the NVIDIA showcase look distinctly off, the signs aren’t exactly great. In trying to enhance realism, the DLSS 55 can quite clearly flatten style.

And besides, whether or not the graphics remain true to the original artistic vision, there will be plenty who take exception to the commercialisation of the technology purely out of principle.

When it comes to AI substituting creative works, this is hardly a knee-jark reaction. Gaming is an industry plagued by constant instability, with an estimated 6,000 to 9,000 workers laid off last year. Those who remain are being asked to do more with less, all while watching publishers and tech firms pour investment into tools that aim to automate the skills they’ve cultivated over years of graft.

As it stands, the announcement video has 46,000 dislikes to just 7,700 likes on YouTube. Will NVIDIA reconsider or double down in the coming weeks? Only time will tell.

Enjoyed this? Click here for more Gen Z focused gaming stories.

Accessibility