It appears that Roblox Corporation is currently working on a DLSS 5-like AI-based supersampler for Roblox. This supersampler will be part of Roblox Reality, which aims to make your creations in Roblox look and feel more realistic. So, let’s take a closer look at it.
According to the devs, Roblox Reality is a hybrid architecture blending the capabilities of the Roblox Cloud and Game Engine, with the photorealism of Video World Models.
“Core world state is durably and efficiently stored on the server to ensure consistency across clients and support consistency over time, sessions, and days using cost and space-efficient storage. Multiplayer gameplay is supported via strong server authority for fairness and consistency, alongside speculative client-side simulation to achieve low latency. For rendering, cloud-based level of detail (LOD) and compositing systems generate high-fidelity assets delivered via a content delivery network (CDN). The Roblox Video Model (Super Upsampler) leverages rendered video and rich data model context to produce stochastic visuals and striking realism, operating on the edge for every player with optimal performance powered by cloud-edge GPU infrastructure.”
To showcase Roblox Reality, the devs shared the following video. The video on the top left is Roblox content recorded using the Roblox rendering engine today. The video on the top right is a representation of the 3D data we can use to condition the video generation. Then we have the video on the bottom left that shows the current Roblox upsample video model running in their labs, which does not yet run in real time. Finally, the bottom right video shows a mockup of the product vision and what is possible in the future with this technology.
All in all, Roblox Reality will be a hybrid architecture that divides responsibilities between the Roblox Game Engine and the Roblox Video World Model.
The Roblox Game Engine will handle the structured and logical aspects of the world, providing stable long-term memory, symbolic logic, and repeatable simulation. It will also be responsible for fundamental physical operations like collision and behaviors. Primary movement of objects is managed in the engine. For example, the location and velocity of a car, its wheels, shocks, and steering, will all be handled by the game engine. The Video World Model will then layer on additional visual and generative components, like the beads of water streaming along the windshield and the fluttering of leaves as the car zooms by, delivering breathtaking visuals.
Roblox Reality will target 2K resolution at 60 Hz. The Super Upsampler will run in adjacent edge datacenters with optimal performance, powered by H200/B200-class GPUs (or equivalent accelerators). This means that your local GPU will not be used by the Upsampler. As such, you won’t need an RTX 5090 or an RTX 6090 to use it.
In short, this will do exactly what DLSS 5 aims to achieve, but in a cloud-based form. The Roblox Video World Model will be an AI-based post-process that can be applied to the final image. It will attempt to make the final image more realistic. Now, whether you like this approach or not is entirely up to you.
There is currently no ETA for when Roblox Reality will become available. Still, this shows that NVIDIA was, once again, ahead of the curve when it announced DLSS 5. So, it will be interesting to see the reaction of all Roblox fans.
Stay tuned for more!
John is the founder and Editor in Chief at DSOGaming. He is a PC gaming fan and highly supports the modding and indie communities. Before creating DSOGaming, John worked on numerous gaming websites. While he is a die-hard PC gamer, his gaming roots can be found on consoles. John loved – and still does – the 16-bit consoles, and considers SNES to be one of the best consoles. Still, the PC platform won him over consoles. That was mainly due to 3DFX and its iconic dedicated 3D accelerator graphics card, Voodoo 2. John has also written a higher degree thesis on the “The Evolution of PC graphics cards.”
Contact: Email