Is DLSS 5 Fixed? NVIDIA Doubles AI Speed Amid Ongoing Controversy

NVIDIA Unveils Major Performance Boost for DLSS 5, But Questions Remain Over AI-Driven Character Fidelity

NVIDIA has delivered a significant leap forward in its DLSS 5 (Deep Learning Super Sampling) technology, announcing that the latest update doubles the neural rendering model’s speed while slashing 4K video memory usage by 20%. The improvements come as part of an ongoing effort to refine AI-driven graphics enhancement without compromising on visual quality or developer intent. However, despite these technical advances, lingering concerns persist over how the AI affects character appearances—particularly following high-profile criticisms tied to facial alterations in earlier demos.

The October update reduces the model’s parameter count by 24%, bringing it down to approximately 118 million parameters. This reduction not only enhances efficiency but also significantly lowers GPU overhead—cutting it in half compared to the September version. According to NVIDIA, the updated model runs twice as fast as before and is now ten times faster than when DLSS 5 was first introduced in March. These gains are particularly impactful for high-resolution gaming, where performance and memory constraints often limit frame rates and graphical fidelity.

A New Era of Speed and Efficiency

The enhanced performance is already making waves across upcoming titles. NBA 2K27, set to launch on 16 October, will be among the first games to adopt the upgraded DLSS 5 model on GeForce RTX 50 Series GPUs, RTX Spark laptops, and GeForce NOW cloud streaming service. NVIDIA claims players can expect frame rate increases of up to 1.4x with the new implementation—though this figure is specific to the game and should not be interpreted as a universal doubling of performance across all titles.

DLSS 5 works by using artificial intelligence to intelligently upscale lower-resolution frames while enhancing lighting, material textures, and depth perception in real time. Rather than relying solely on traditional rendering pipelines, the system learns from vast datasets to predict and fill in missing details, resulting in sharper visuals with less computational strain. While this promises more immersive experiences, especially in graphically demanding environments, it has also sparked debate about whether the AI might inadvertently alter artistic direction.

Addressing the AI Face Controversy: Progress or Perceived Fix?

One of the most contentious moments came during the March reveal of DLSS 5, when NVIDIA showcased a demo featuring Grace Ashcroft from Resident Evil Requiem. The AI-enhanced version displayed smoother skin, brighter tones, and more defined facial features—leading many viewers to liken the transformation to a digital beauty filter. Critics argued that such changes risked undermining the original design choices made by developers and artists, potentially distorting characters’ identities.

In response, NVIDIA has emphasized that developers now have greater control over how DLSS 5 influences visual output. For instance, Avalanche Software has reportedly fine-tuned the implementation in Hogwarts Legacy to ensure the game’s distinct aesthetic remains intact. The updated version is scheduled to roll out on 5 November, offering a real-world test case for balancing AI enhancement with creative integrity.

Yet, the company has not explicitly confirmed whether the recent changes directly address the facial rendering concerns raised in March. The announcement highlights customization capabilities but stops short of stating that the core issue—AI-induced alterations to character appearance—has been resolved. As such, the true test lies in how the final implementation looks in practice once released.

Frame Generation Gets a Speed Boost Too

Separately, NVIDIA has introduced a beta Performance Mode for DLSS Frame Generation on GeForce RTX 40 and RTX 50 Series cards. This new feature includes a lighter algorithm called Preset C, designed to reduce processing demands when generating additional frames. While the trade-off involves minor reductions in image quality, NVIDIA reports performance gains of up to 35% on devices like the RTX 5060.

It's important to note that this improvement is independent of the neural rendering upgrades to DLSS 5. Frame Generation focuses on increasing frame rates by predicting intermediate frames between rendered ones, whereas DLSS 5 enhances visual quality through AI-based upscaling and detail reconstruction.

Future Expansions and Developer Tools

Looking ahead, NVIDIA plans to bring the updated DLSS 5 model to RTX Remix on 23 October, empowering modders with advanced tools to revitalize older games with modern visuals. Additionally, Where Winds Meet is expected to receive the improved neural rendering model in late November, further expanding the reach of the technology.

As the industry moves toward more AI-integrated rendering solutions, the balance between performance, realism, and artistic control becomes increasingly critical. The success of DLSS 5’s next wave of implementations will ultimately hinge on whether it can deliver stunning visuals without altering the soul of the games it enhances. The coming months will be pivotal in determining whether NVIDIA has truly addressed both the technical and ethical challenges surrounding AI-powered graphics.

Share :