Nvidia DLSS 5 is "gen AI bullsh*t" that "looks like sh*t" says Dusk developer New Blood

In case it wasn't already clear, Nvidia DLSS 5 has been a bit divisive since Nvidia demoed it last week at GTC. The new AI-based real-time neural renderer that 'enhances' the look of your games through generative AI has caught the ire of many gamers, developers, and other industry folk, and the latest to add to the pile-on are David Szymanski and Dave Oshry, developer of Dusk and CEO of its publisher, New Blood Interactive, respectively.

Nvidia has been quick to try and temper criticism, with its CEO Jensen Huang saying much of the DLSS 5 criticism is completely wrong and insisting that suggestions DLSS 5 is just a 2D AI filter are misguided too.  However, clearly plenty of folk remain unconvinced.

Speaking to PC Gamer and posting the full response they gave to the publication on the New Blood X account, Dave Oshry kicked things off by pointing out how extra specially ridiculous it is that DLSS 5 is even part of the DLSS suite of tech. "Tell me what generative AI has to do with deep learning super sampling - which is what DLSS stands for […] they're hiding this generative AI bullshit behind the moniker because they think we're stupid."

"They know if they called it something like 'Nvidia Generative Upscaling' the public backlash would be immediate and intense. But, unfortunately for Nvidia, even if they call it DLSS 5, people have working eyes and can see that this is exactly what we think it is.

"The only thing we can do besides calling them out on it is vote with our wallets. Cripple their sales, tank their stock price. Stop collaborating with them as developers. […] We as developers and players need to push back against this bullshit just like we did with NFTs and crypto games and try in vain to do with predatory micro transactions, loot boxes, and battlepasses."

Pointing out the reality of New Blood's status in the whole situation, Oshry acknowledges that "I have not dog in this fight […] we make retro bloody indie games. […] We've only ever released one game that we put DLSS and RTX into (Amid Evil) and it was a huge pain in the ass. […] But it was a fun experiment and Nvidia sent us some free GPUs for our trouble."

Oshry closes by pointing out that thanks to the way the generative AI in DLSS 5 takes over the rendering, "why make game art at all? Why not just draw some shapes and colors and let AI generate what it thinks it should look like?" To which my thoughts are, "don't tempt fate."

Speaking after Dave Oshry, David Szymanski confirmed that he agrees with everything Oshry just said, adding that "The core problem here is that DLSS 5, per Nvidia's own showcase video, looks shit.

"Even if we set aside all (relevant and valid) concerns about artistic intent and generative AI itself, the lighting and contrast it adds (or removes, in some parts) make scenes less realistic and believable and more like one of those Breaking Bad YT shorts with the contrast and sharpening cranked up. So what exactly is the appeal here?"

Perhaps hitting the heart of the matter, given Nvidia's history of pushing many of the recent technologies we've seen in game graphics, Szymanski points out that "stuff like DLSS and TAA and raytracing are starting to pay off, but it's been at an immense cost to the clarity, accessibility, and playability of the games that use them for years." In other words, even if DLSS 5 image quality improves and is ultimately the future, we could face years of poor implementation and performance before the benefits are actually felt by gamers.