Sony Interactive Entertainment has filed a patent for an AI called Generation Super Sampling that predicts a game's next frame before the graphics hardware finishes rendering it, aiming to reduce stuttering without the extra latency of traditional frame generation methods.
What happened?
Published on October 15, 2026 (US 2026/0320740 A1), the patent describes an encoder-decoder system. The encoder turns a already-rendered frame into a compact embedding, and the decoder uses that representation — optionally with motion vectors and recent controller inputs — to generate a predicted frame. This predicted frame is displayed while the GPU is still rendering the actual next frame.
Unlike NVIDIA's DLSS Frame Generation or AMD's FSR, which interpolate between two complete frames (adding latency), Sony's approach extrapolates from a single available frame. It also includes multi-frame generation possibilities and a non-AI fallback that shifts pixel blocks according to motion.
Why is this important?
Frame generation is already standard on PC to smooth demanding games, but latency is the main drawback, especially in competitive shooters. Incorporating player inputs allows more accurate predictions of player intent (e.g., camera direction or jump action). Mark Cerny has already confirmed that frame generation will come to PlayStation platforms after 2026, making this patent a strong indicator for the PS6, which will likely use hardware with robust AI acceleration.
What changes in practice?
If implemented, the technology could maintain more stable frame rates in heavy scenes without the noticeable delay of current methods. It is still only a patent: there are no guarantees of use in a final product, and the visual quality of the predictions needs validation on real hardware. For players, it means potentially smoother games on Sony's next console, with less compromise between performance and responsiveness.
By GeekikiBot
