Weekly Patent Digest

Sony's 10 filings this week cover a wide range of problems, from network interference to room-scale ambient effects. On the streaming side, 3 applications address cloud gaming specifically: one separates text draw calls from the main video stream so subtitles and UI remain legible when bandwidth drops, another lets players tag a recorded moment of gameplay and turn it into a shareable clip that others can watch or actually play, and a third ranks game objects by priority, giving characters and targets higher frame rate and resolution while UI panels like stat displays receive less, reducing overall bandwidth demand. A separate hardware-oriented filing describes using machine learning to read gameplay audio, video, and haptic output, then drive household devices like room lights, fans, and spare speakers in sync with the game, accounting for each device's response latency and skipping effects that would be redundant. Another application pulls 3D characters out of a running game and places them into a player's physical space via an AR headset, without modifying the underlying game program. Rounding out the Sony batch, 1 filing trains a model on high-viewership video content and then scores game scenes, trailers, and storylines for predicted popularity before release, and another describes scheduling downloads and adjusting frame rate or screen brightness based on real-time power grid data.

EA's 2 applications sit in noticeably different corners of game development. One describes an imitation learning agent that plays through a level, records its own state and action data, then uses that data to place assets and reshape terrain, replaying the level afterward to identify goals a player could not actually reach and correct them. The other addresses visual effects: a per-pixel color blending system for mesh particles where each pixel's distance to a specific point on a moving entity determines whether it renders as color A, color B, or a blend of the two, producing more dynamic-looking effects without manual per-frame authoring.

Samsung filed 1 application this week, focused on its gaming hub television. The described system compares the game screen state before and after a player pauses, then surfaces only the quick-panel overlay tools, things like a mini map, aim point, frame rate counter, or timer, that are relevant to whatever is currently displayed on screen.

Nintendo's 1 filing describes a world traversal mechanic built around fixed environmental objects that function as activation points. Reaching one triggers a cascade of effects across a surrounding dark region: the area lights up, map data for that zone becomes visible, player health restores, and the object doubles as a fast-travel destination.

Tencent's 1 application describes a pet interaction system for open-world games. When 2 pets with a defined species relationship encounter each other in the open world, the game transitions them into a turn-based combat mode that matches the nature of their interaction, and the player has the option to fight on behalf of either pet.

Century Games Pte. filed 1 application addressing loot recovery after a tower defense encounter. Items a player missed during the fight are not discarded. Instead, they follow the player back into the main city view, where a single tap on an icon claims them, with no instance switch and no change to the game interface.

Cross Game Play's 1 filing describes a portable identity system for players moving across different game engines. A player's avatar and its cosmetics are stored alongside a ledger of ownership, and an offline verifier handles credential checks by rejecting revocation data that has gone stale, allowing a LAN tournament client to authenticate player credentials without a live network connection.

Peloton Interactive's 1 filing describes a game interface designed for use on exercise machines. Menus appear as objects embedded inside the virtual environment rather than as overlaid screens, the camera holds a steady angle while the user exercises, and the player's avatar mirrors their physical movements and gestures in real time.