Key information
Sony has filed a patent application titled "Generation Super Sampling" (US 2026/0320740 A1), published on October 15. While the document does not explicitly name a console, it is widely speculated that this technology is intended for the PlayStation 6.
The proposed system utilizes an encoder neural network to convert rendered frames into compact embeddings, which a decoder then uses to predict subsequent frames. The process incorporates motion vectors and controller inputs to anticipate player movement, allowing the system to display predicted frames while the next actual frame is still being processed.
Unlike traditional methods that insert frames between two real ones, Sony's approach predicts forward from a single frame to reduce latency. The technology includes a non-AI fallback for shifting pixel blocks and trains directly on video streams without human labeling. However, the patent lacks specific data regarding image quality or latency performance.
Industry observers note that this development follows the PS5 Pro's PSSR upscaler, aligning with rumors of a 4K at 120 FPS target for the next console. Despite the potential, concerns exist that developers might prioritize this technology over game optimization. As with other recent Sony filings, this remains an exploratory concept with no guarantee of implementation.
