Nvidia tones down DLSS 5 ahead of its Hogwarts Legacy debut in November
Nvidia is preparing a full DLSS 5 debut in Hogwarts Legacy, scheduled for next month. The technology's first showing more than six months ago drew a wave of criticism: gamers called it generative AI for games, and character faces looked distorted. Nvidia has now posted new comparison screenshots in which the face edits look much more restrained, roughly on par with NBA 2K27. - DLSS 5 is due to debut in Hogwarts Legacy in November. - Faces in the new screenshots change far less than in the first demo. - The final quality will only become clear once the game ships with the feature.
AI-processed from The Verge; edited by Hamidun News
More than six months ago, Nvidia showed off DLSS 5, and the gaming community's reaction was almost unanimous: people called the technology generative AI for video games and accused the company of losing its way. Now, as DLSS 5 prepares for a full debut in Hogwarts Legacy next month, Nvidia has posted new comparison screenshots. They show that the technology has been seriously dialed back.
How the controversy started
Faces were the main complaint. In the first demos, neural-network processing noticeably changed how characters looked. A 15-year-old schoolgirl in Hogwarts Legacy suddenly looked much older. Grace in Resident Evil Requiem got a smooth, "filtered" look that a writer at The Verge described as yassified, meaning beautified in the style of social media beauty filters.
For an audience used to DLSS raising frame rates and cleaning up the image, this came as a surprise. Previously, the technology upscaled the picture and generated extra frames. This time, players felt it was rewriting the artists' vision.
What changed in the new version
In Nvidia's latest screenshots, the face edits look far more modest. According to The Verge, they are now as subtle as in NBA 2K27, where DLSS 5 was rolled out earlier. The teenagers in Hogwarts Legacy no longer look like adults, so the most visible flaw of the first demo has been fixed.
Key facts so far:
- debut game: Hogwarts Legacy;
- timing: next month, i.e. November 2026;
- time since the first showing: more than six months;
- benchmark for how restrained the changes are: NBA 2K27;
- most criticized examples: the schoolgirl in Hogwarts Legacy and Grace in Resident Evil Requiem;
- material available for evaluation: comparison screenshots published by Nvidia.
What remains unclear
The official screenshots are still marketing material. Nvidia picks the frames, angles and scenes itself, so they say nothing about how the technology will behave in motion, in dark interiors and in close-up dialogue scenes. That is where neural-network face processing most often produces artifacts.
The available material contains no data on GPU requirements, on performance losses or gains, or on whether face processing can be turned off separately from the other features. These questions will determine whether DLSS 5 becomes a practical tool or whether players switch it off right after the first launch.
The story is also telling for Nvidia itself. The DLSS brand has long been associated with free performance, and the company risks that reputation when a neural network starts changing a game's art direction. The rollback to more restrained settings looks like an admission that players do not want AI to redraw characters in its own way.
How to approach the launch
If you have an Nvidia GPU and plan to play Hogwarts Legacy in November, don't enable DLSS 5 by default. First compare the image with the technology off and on in scenes you know well, especially in close-ups of faces, and only then decide. Watch for the first independent reviews and comparisons in motion: they will show whether the restraint seen in the screenshots really holds up in the game. For developers, the takeaway is simpler: if an AI layer changes how characters look, players should be able to turn that part off separately.
- Oct 24, 2026Fixer in Super Micro case pleads guilty to routing AI servers to China
- Oct 24, 2026OpenAI Reportedly Earns $20 Billion Less Than Previously Thought
- Oct 23, 2026Matrix Game 4.0 Hits 60 FPS on a Single Gaming GPU and Beats Nvidia's Rival Model
- Oct 23, 2026Moonshot releases Kimi Vendor Verifier: 6 tests for checking K2.6 providers
❓ Key Questions Answered
How the controversy started?
Faces were the main complaint. In the first demos, neural-network processing noticeably changed how characters looked. A 15-year-old schoolgirl in Hogwarts Legacy suddenly looked much older. Grace in Resident Evil Requiem got a smooth, "filtered" look that a writer at The Verge described as yassified, meaning beautified in the style of social media beauty filters. For an audience used to DLSS raising frame rates and cleaning up the image, this came as a surprise. Previously, the technology upscaled the picture and generated extra frames. This time, players felt it was rewri
What changed in the new version?
In Nvidia's latest screenshots, the face edits look far more modest. According to The Verge, they are now as subtle as in NBA 2K27, where DLSS 5 was rolled out earlier. The teenagers in Hogwarts Legacy no longer look like adults, so the most visible flaw of the first demo has been fixed. Key facts so far: debut game: Hogwarts Legacy; timing: next month, i.e. November 2026; time since the first showing: more than six months; benchmark for how restrained the changes are: NBA 2K27; most criticized examples: the schoolgirl in Hogwarts Legacy and Grace in Resident Evil Requiem;
What remains unclear?
The official screenshots are still marketing material. Nvidia picks the frames, angles and scenes itself, so they say nothing about how the technology will behave in motion, in dark interiors and in close-up dialogue scenes. That is where neural-network face processing most often produces artifacts. The available material contains no data on GPU requirements, on performance losses or gains, or on whether face processing can be turned off separately from the other features. These questions will determine whether DLSS 5 becomes a practical tool or whether players switch it of
Access Frontier LLMs & Multimodal AI in One Workspace
Discover NomadNet AI (nomadnet.ai) — unified access to frontier language models, research synthesis, and multimodal workflows.
The AI world, distilled — once a week
Seven stories that actually mattered, hand-picked. No noise, no reposts, no press releases.
Done! Check your inbox for a confirmation.