Guozhen AIGlobal AI field notes and model intelligence

Realtime AI News

Hogwarts Legacy gets DLSS 5 with 3D-Guided Neural Rendering on GeForce NOW

NVIDIA has pushed a DLSS 5 update to Hogwarts Legacy on GeForce NOW, adding 3D-Guided Neural Rendering that sharpens lighting and material detail while keeping the game's original geometry and textures. Avalanche Software tuned the integration through DLSS 5 developer controls to preserve the title's art direction, and the update is available to stream immediately.

Published
《霍格沃茨之遗》上线DLSS 5:3D引导神经渲染登陆GeForce NOW
Image source: blogs.nvidia.com

NVIDIA used its Thursday GeForce NOW update on November 5 to announce that Hogwarts Legacy now supports DLSS 5 with 3D-Guided Neural Rendering. The upgrade went live the same day, and members can stream the game rather than download it or upgrade their local GPU.

Hogwarts Legacy is developed by Avalanche Software and published by Warner Bros. Games under the Portkey Games label. It is an open-world action RPG set in the 1800s wizarding world, where players begin as a fifth-year student, explore familiar and unfamiliar locations, brew potions, learn spells and encounter magical creatures.

NVIDIA describes the change as adding lifelike lighting and material detail while retaining the underlying geometry and textures. In practice, neural rendering acts as an enhancement layer over the existing scene rather than a rebuild, and Avalanche Software tuned the integration through DLSS 5 developer controls to maintain the game's artistic intent and style.

The title also supports DLSS Multi Frame Generation and Ray Reconstruction, which accelerate and improve ray-traced graphics. On the cloud side, DLSS 4.5 controls let members fine-tune the quality and performance trade-off, with Ultimate members getting RTX 5080-class performance across PCs, Macs, handhelds, browsers, phones and compatible TVs.

The update shows neural rendering moving from tech demo to shipping game content, with cloud streaming acting as the fastest distribution channel for new graphics features. Players on older hardware see the new look without a GPU purchase, while NVIDIA gets a live showcase for its cloud GPUs and the pace of its DLSS cadence.

The same post laid out the November slate: 24 games joining GeForce NOW this month, eight of them this week. Blizzard's World of Warcraft: Forever launched on November 4 and streams on the service, with support planned for around 100 of the most popular CurseForge add-ons at launch and a list that will evolve based on player feedback.

To mark the release, GeForce NOW is offering a launch-week access reward across all membership tiers, redeemable through 11:59 p.m. PT on November 11 while supplies last. Rewards and release cadence are retention tools for the subscription, but the bigger signal is whether DLSS 5 reaches more AAA titles quickly, which would show how fast neural rendering becomes standard.

Why it matters

DLSS 5 arriving in an already-released blockbuster spreads neural rendering as a content update rather than a hardware selling point, and cloud streaming removes the local GPU barrier for seeing it.

NVIDIADLSS 5GeForce NOWNeural Rendering
Back to realtime news

Nearby Updates

All

11/05, 22:00

Box builds Nvidia's Nemotron embeddings into a new agent-first retrieval system

At its developer conference on November 5, Box unveiled Box Content Retrieval, which puts Nvidia's Nemotron embedding models at the core of its product so agents can answer queries from embedded metadata. Nvidia says the system delivers four-times-faster queries with 20% fewer errors and 90% fewer tokens, with a rollout to all Box Enterprise Advance customers early next year.

11/05, 21:40

ByteDance Seed publishes a sparse autoencoder for Qwen3-30B-A3B

ByteDance Seed has published a sparse autoencoder repository built on Qwen3-30B-A3B, tagging it for interpretability work on a mixture-of-experts model. The release carries no downloads or likes yet and discloses almost no training detail, so it reads as an early tooling drop rather than a finished research result.

11/05, 21:34

Ex-Huawei Pangu lead Wang Yunhe founds RSI and unveils a 125B agent model

According to Zhidongxi, former Huawei Pangu lead Wang Yunhe has founded a startup called RSI and released a 125-billion-parameter agent model. The report carries few technical details beyond the parameter count, leaving open whether the weights are open, how the model benchmarks, and how it will be sold.

11/05, 22:42

Aina launches a $399 AI ring with a screen, joystick and mic for talking to agents

Hardware startup Aina has released a $399 smart ring with a screen, joystick, microphone and speaker, positioned as a physical controller for AI agents and everyday apps. Pre-orders opened Thursday at early-bird pricing, with shipping planned for January 2027 and the price rising to $499 afterwards.