DLSS 5 received an immensely negative reaction from players when it was first revealed in early 2026, and it seems Nvidia has not taken those complaints to heart. Instead, it doubled down, and the latest work-in-progress leaked build is a mixed bag.
While DLSS 5 looks genuinely good in certain select scenarios, it’s not what this is about. Nvidia’s slop filter introduces unnecessary detail and “realism”, which drastically alters the game beyond its intended look, and could set a dangerous precedent.
DLSS 5 destroys the original artistic vision of the game
Chasing realism isn’t what games are always about
DLSS 5 is not an upscaler. In fact, it actively alters the output and enhances it further, using generative AI to “render” a more photorealistic image. This makes it more computationally expensive compared to traditional DLSS, which kind of defeats the whole point of the tech, but I’m getting ahead of myself here…
The current suite of DLSS 5 screenshots is from an unofficial build, though, which was leaked from an early access version of NBA 2K27 and subsequently ported over to other titles.
And the results have been quite varied, to say the least. DLSS 5 seems to work best under older titles, bringing them to a modern “RTX remaster” equivalent. That is, until you look into the finer details.
DLSS 5 is way too strong of a filter, and much of the remastering done here is full of flaws, as you’d expect from any kind of AI-generated frame. Not to mention it looks downright terrible on many titles, with Final Fantasy VII Remake’s Cloud being the stuff of nightmares.
If anything, though, DLSS 5 ruins a game’s graphical presentation. Games are designed with certain visual elements in mind, and forcibly applying a filter to change things so dramatically just doesn’t sit right with me.
Sure, you could argue that mods and reshade packs do the same thing, but Nvidia’s new shiny upscaler looks a lot worse in comparison. Not to mention, these community-driven shaders were created with intent — with each section of a game hand-crafted to suit the author’s tastes.
DLSS 5 in its current (beta) state is just a photorealism filter, and not all games aim for that. While Nvidia has confirmed that developers can fine-tune the DLSS 5 output to maintain the “original” artistic vision, it begs the question: why even ship games that need a visual filter to look as intended?
DLSS 5 has a massive impact on performance
And requires specific hardware to work
It’s not like DLSS 5 is a lightweight model either. Oh no, far from it — and the “upscaler” has been seen to slash frame rates by up to 50%, mandating the use of both traditional DLSS upscaling and frame generation to offset the difference, which works as well as you’d expect.
To make matters worse, it is also currently restricted to the RTX 5000 (Blackwell) Series, which means you’ll need a rather pricey GPU to even stand a chance at previewing this tech. You won’t be able to use this on something like an RTX 5060 well at all.
Nvidia plans to port it to the RTX 4000 Series later, although I wouldn’t be too optimistic about performance. Restricting the tech to current-gen cards that are mostly out of reach for most folks isn’t a great outlook.
The so-called improvements on offer simply aren’t worth the performance loss, and a broad majority will largely ignore them.
As the technology matures, it could slowly turn into a crutch
Another ray tracing-like situation could be on the horizon
I fear the worst with the eventual, full release of DLSS 5. The current crop of game tech, especially ray tracing and upscaling, has been a bane for gaming in recent years. While these were made with good intent, most games still overutilize them, to the point that they’ve become more of a crutch.
Running a modern AAA game at higher resolutions without upscaling often tanks frame rates below 60 FPS, making it mandatory no matter the card you use. The same can be said for ray tracing, which is being forced on every game these days.
DLSS 5 could pose a different kind of problem, though. By actively altering the game’s look, it could push developers to optimize for a DLSS 5 “exclusive” look, and because it would be gated behind specific GPUs, it would ultimately severely restrict the number of players who could achieve that “final” level of detail.
DLSS 5 is trying to solve a problem that does not exist
To address the elephant in the room, DLSS 5 is trying to solve a problem that doesn’t exist. Just play games the way they were meant to look in the first place. This kind of implementation has absolutely zero reason to exist and is a slap in the face to most designers.
If anything, it could set off a wave of DLSS 5 “enhanced” titles that force the tech to be used. It really makes you think, though: Why not just design the game with the original look and intent (without relying on an AI filter to enhance things over), instead of producing generic, photorealistic abominations?
Until then, there’s not much to do but hope that AMD doesn’t follow suit, along with the rest of the industry.