Sony patents frame prediction technology to reduce input lag

Sony Interactive Entertainment has filed a patent for a new frame generation technology aimed at reducing input latency in gaming. Unlike current methods such as Nvidia's DLSS Frame Generation, which inserts synthetic frames between two previously rendered frames, Sony’s proposed system uses neural networks to predict the next frame based on player input and motion vectors.

The technology utilizes an encoder and decoder system. The encoder converts real images into numerical representations, while the decoder predicts how the subsequent frame will appear. By incorporating user controller inputs—potentially within a 15ms window—the system can prepare a synthetic frame before the actual next frame is fully rendered, effectively minimizing the delay between a player's action and the visual response.

While the patent suggests significant improvements to perceived smoothness and responsiveness, it does not confirm an immediate update for the PlayStation 5 or specific details regarding the PlayStation 6. The filing is a continuation of research into image stream embeddings and predictive modeling.