Sony patent details AI frame prediction system
A Sony patent published on October 15 describes a system that predicts and displays a synthetic frame while the real next frame renders.
Sony has patented a frame-generation technique titled Generation Super Sampling, according to a filing published on October 15 and reported by Patentlyze. The patent describes a system that predicts and displays a synthetic image before the next real frame is complete.
The system uses an encoder to analyse previous gameplay images and passes image embeddings to a decoder that predicts the following image. A diagram in the filing shows the current image, motion vectors and the player's last input feeding into a neural network while the normal graphics pipeline renders the actual next frame.
The filing states that the frame generation system can incorporate motion vectors and optional player inputs such as button presses and joystick or mouse movements, and it discusses predicting multiple upcoming frames.
Sony has not announced PlayStation 6 hardware, a release date or pricing, and the patent does not confirm that the technique will appear in a product.
Existing upscaling work
Frame generation has been available in high-end PC gaming for several years through technologies such as Nvidia DLSS and AMD FSR. In October 2025, Sony introduced PSSR upscaling with the PS5 Pro, and its Project Amethyst partnership with AMD has produced Neural Arrays, Radiance Cores and Universal Compression.
Microsoft's Project Helix is confirmed to use a custom AMD SoC built around future DirectX and FSR technology, with hardware expected to reach developers beginning in 2027.
Quick answers
Has Sony announced the PS6?
No. Sony has not announced PlayStation 6 hardware, a release date or pricing.
What does the Generation Super Sampling patent describe?
A system that uses an encoder to analyse previous gameplay images and a decoder to predict the next image, displaying a synthetic frame while the real one renders.