What you need to know
Sony’s Generation Super Sampling patent describes a possible PS6 frame-generation system that uses motion vectors and controller input to predict a frame before rendering finishes, potentially reducing latency and supporting multi-frame generation. The patent provides no performance or image-quality results, and the technology’s arrival on PS6 is unconfirmed.
Voice: Kosta AndreadisSpeed
0:00 / 2:42
Use left and right arrow keys to seek audio.
Sony is all in on Neural Rendering, which began with the PlayStation 5 Pro and its use of PSSR (PlayStation Spectral Super Resolution), which brought DLSS-like, AI-powered upscaling to consoles for the first time. For the PlayStation 6, Sony has already confirmed that its Project Amethyst partnership with AMD will bring more advanced Neural Rendering technologies.
Powered by AMD’s upcoming RDNA 5 architecture, a new patent titled “Generation Super Sampling” (US-20260320740-A1) details how a potential new form of Frame Generation might work on the upcoming PlayStation 6. Like DLSS Frame Generation, it takes a traditionally rendered image alongside data like motion vectors to “predict” and generate a new frame.
However, unlike existing forms of Frame Generation that insert one or more generated frames between two completed frames, Sony’s patent describes a system that predicts and generates the next frame before the hardware (presumably the PS6) finishes rendering it. To achieve this, the patent includes an optional step that accounts for controller input. This means that this version of Frame Generation could potentially boost performance.
- Read more: Sony confirms PlayStation systems will be getting AI frame generation for ‘fake frames’
- Read more: AMD’s FSR 4.1 for RDNA 4 is based on the same tech as Sony’s new PSSR for the PS5 Pro
Although the patent doesn’t include any information on performance or the image fidelity of these generated frames, the diagrams do offer a quick overview of how it would work. When playing a game, the AI model and Frame Generation system run alongside the game. As the next frame renders, the AI model processes the previous frame and sends its “prediction,” or AI-generated frame, to the display engine/buffer. As a result, the generated frame can appear before the next rendered frame is ready, reducing the latency associated with existing Frame-Generation solutions.
How does Sony’s patent synchronize an AI-generated frame with the next traditionally rendered frame?
Which neural-rendering technologies has Sony discussed through its Project Amethyst partnership with AMD?
What role is AMD’s RDNA 5 architecture expected to play in the PS6’s neural-rendering pipeline?
Which PS5 Pro games support PSSR, and how is the upscaling feature enabled?
Want something else? Ask TweakBot.
For accuracy, the model also considers motion vectors and controller input in its predictions. The system also appears to support multi-frame generation, which could help minimize performance drops and maintain a solid 60 FPS without stutters. There’s no guarantee this technology will debut on PS6, but odds are Sony’s next-generation console is where it’s currently being prototyped and tested.
