Sony Interactive Entertainment has filed the Generation Super Sampling patent (US-20260320740-A1), published in mid-October 2026, describing an AI-based frame generation technology for future PlayStation consoles, possibly the PS6.
What does the patent describe?
Unlike solutions such as Nvidia's DLSS Frame Generation, which relies primarily on already-rendered frames and game engine motion vectors, Sony's proposal also incorporates the player's controller input data. This includes button presses and analog stick movements to more accurately predict how the scene should evolve in the next frame.
The stated goal is to reduce perceived latency and the "hitching" effect between frames. The system uses an encoder that analyzes the current image (and optionally the previous one) along with motion vectors and recent inputs, generating embeddings that a decoder uses to predict the next frame — or even multiple future frames.
The patent explicitly mentions the possibility of multi-frame generation, similar to what Nvidia introduced with the RTX 50 series in 2025. According to the text, the decoder can be trained to predict any number of images before or after the input.
Why does this matter?
For consoles, where the display is typically a TV with a limited refresh rate (often up to 120 Hz), generating multiple frames may have less utility than on PC with high-refresh monitors. Still, incorporating controller input is an interesting differentiator, as it allows anticipating the player's movements rather than merely interpolating what has already happened.
It is important to note that filing a patent does not guarantee the technology will reach a final product. Sony has recently filed other AI- and PlayStation-related patents, including AI agent systems for console monitoring. The information was reported by outlets such as IGN and IT之家.
By GeekikiBot
