Sony is exploring a new AI-powered frame generation technology that could make its way to the PlayStation 6 (PS6), according to a recently published patent. Dubbed “Generation Super Sampling”, the technology aims to generate game frames before the console finishes rendering them, potentially reducing stuttering without introducing as much input lag as existing frame generation methods.
The patent application, numbered US 2026/0320740 A1, was recently published on 15 October 2026. While it does not mention the PS6 specifically, the technology would be a logical next step for Sony as it continues to develop AI-powered graphics features for its consoles.
How It Supposedly Works
To understand what makes Sony’s approach different, it helps to look at how conventional frame generation works. Technologies such as NVIDIA’s DLSS Frame Generation create artificial frames by analysing previously rendered images and inserting new ones between them. This can make a game appear to run at a higher frame rate, although the process can introduce additional latency because the system needs to wait for subsequent frames before generating the ones in between.
Sony’s proposed method instead attempts to predict what the next frame will look like using an image that has already been rendered. An AI model processes that image and generates a prediction, which can then be displayed while the graphics hardware continues rendering the next actual frame.
In other words, rather than waiting for two completed frames to generate one in between, Sony’s approach attempts to predict what comes next and display it ahead of time. This could help the console maintain smoother gameplay during demanding scenes, when the hardware might otherwise struggle to render frames consistently.
Motion Vectors And Controller Inputs
According to the patent, Sony’s system uses an encoder neural network to convert a rendered image into a compact representation of its visual information. A decoder neural network then uses this information to predict the next frame.
The system can also incorporate motion vectors, which describe the direction and speed at which objects move between frames. Recent controller inputs are another optional source of information, allowing the AI to account for player actions when predicting what should appear next.
For example, if a player presses the jump button, the system could use that input alongside the current image and movement data to predict the beginning of the jump animation. The generated frame could appear while the console is still rendering the actual image, potentially making the action look more immediate.
The patent also describes the possibility of generating multiple frames from a single rendered image. This could help maintain a more consistent frame rate when rendering demands fluctuate, although generating more frames does not necessarily mean the game would run faster.
Sony also outlines a non-AI alternative that uses motion vectors to move blocks of pixels and estimate their positions in subsequent frames. This provides another possible way to generate intermediate images without relying entirely on a neural network.
What This Could Mean For The PS6
If Sony eventually brings this technology to the PS6, it could complement the AI-powered upscaling offered by the PS5 Pro’s PlayStation Spectral Super Resolution (PSSR). While PSSR improves image quality by upscaling lower-resolution renders, frame generation aims to make gameplay appear smoother by increasing the number of frames displayed each second.
For example, a game that struggles to maintain 60 frames per second (FPS) during demanding scenes could use generated frames to deliver smoother motion. Sony’s approach could also reduce the latency associated with conventional frame generation by predicting the next frame before it is fully rendered, although the technology’s actual performance remains unknown.
There are still challenges to overcome. Rapid camera movements and sudden player actions could lead to inaccurate predictions and visual artefacts, while generated frames do not make the underlying game run faster. Multi-frame generation could also offer limited benefits on 120Hz televisions, which cannot display more than 120 frames per second.
Sony has yet to confirm whether Generation Super Sampling will feature in the PS6, and its patent provides no performance or image quality figures. Nevertheless, alongside the company’s Project Amethyst partnership with AMD, it offers a glimpse of how Sony could use AI to improve graphics performance on its next-generation console.
(Source: US Patent And Trademark Office, via Tweaktown)
The post New Sony Patent Hints At AI Frame Generation For PlayStation 6 appeared first on Lowyat.NET.