Data processing apparatus and method
By predicting higher quality mip levels using a machine learning model, the latency and 'popping' issues in texture streaming are addressed, ensuring seamless and realistic video game graphics.
GB2641071APending Publication Date: 2025-11-19SONY INTERACTIVE ENTERTAINMENT LLC
2 Cites 0 Cited by
Patent Information
- Application Number
- GB2024006867
- Authority / Receiving Office
- GB · GB
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2024-05-15
- Publication Date
- 2025-11-19
AI Technical Summary
Technical Problem
Texture streaming in video games often results in latency and 'popping' issues due to progressive loading of higher quality mip levels, which detracts from the realism and immersion of the game graphics.
Method used
Implementing a machine learning model to predict higher quality mip levels from lower quality mip levels, which are then output until the actual higher quality mip is obtained, thereby reducing latency and minimizing the appearance of 'popping'.
Benefits of technology
This approach reduces latency in loading higher quality textures to an imperceptible level (less than 100ms) and eliminates 'popping', enhancing the realism and immersion of video game graphics.
✦ Generated by Eureka AI based on patent content.
Abstract
A first version of a texture (e.g. 1st texture mipmap) having a first quality (e.g. detail, fineness) level is obtained (401) and used to generate (e.g. upsampling using a Machine Learning model) a prediction (303) of a second version of the texture (e.g. 2nd texture mipmap) having a second, higher, quality level. The predicted second version is displayed, e.g. as part of a video game. A (correct, not predicted) second version of the texture having the second quality level may be obtained (e.g. from memory) and (eventually) output for display in place of the predicted second version. The generation of the predicted second version may be performed when the generation time duration is less than a threshold, the threshold may be the time required to fetch the stored second (higher) version. The predicted texture may therefore eventually be replaced with the actual higher level texture when it arrives. Using the predicted texture instead of the lower quality texture while awaiting the stored higher texture reduces any ‘texture popping’ effect seen when a low quality texture is displayed until a higher quality texture arrives. Latency in displayed image texture improvement is therefore reduced.
Need to check novelty before this filing date? Find Prior Art