Client AI Enhance Module for Video Quality and Bandwidth
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Cloud-based online game services face challenges in maintaining high image quality while minimizing server loading and bandwidth consumption, as existing image enhancement technologies fail to effectively handle diverse graphical contents and often result in unnatural or degraded image enhancements.
Innovation Solution
A method using a pre-trained AI enhance module on the client device, which analyzes differences between decoded and raw images to apply customized algorithms and weighted parameters specific to different scene-modes, ensuring enhanced images are visually similar to the original raw images, thereby improving image quality and reducing server loading and bandwidth usage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If the resolution of raw images and transmission bitrate are increased to maintain high image quality, then the image quality is improved, but the server loading and bandwidth consumption are severely increased
Solution Approach 1:
The patent applies preliminary action by pre-training an AI model on the server to learn the mapping between compressed and high-quality images. This pre-processing of enhancement algorithms allows the server to send lower-resolution images while still achieving high-quality output at the client side, thereby reducing bandwidth consumption without sacrificing image quality.
Solution Approach 2:
The patent introduces an AI enhancement model as an intermediary between the transmitted image data and the final displayed image. This intermediary component processes the received images to reconstruct high-quality visuals from compressed inputs, effectively decoupling the transmission quality requirements from the display quality requirements.
2Manufacturing precision
If conventional image enhancement technologies are used to improve image quality, then the image quality is improved, but the enhancement results are unnatural and lose details due to excessive contrast enhancement
Solution Approach 1:
The patent implements feedback mechanisms where the AI model is trained using loss functions that compare enhanced images against reference high-quality images. This feedback loop continuously adjusts the enhancement parameters to minimize artifacts and preserve natural appearance, preventing excessive contrast enhancement and detail loss.
Solution Approach 2:
The patent dynamically adjusts enhancement parameters based on the specific characteristics of each image and scene mode. Rather than applying fixed enhancement rules, the system modifies parameters such as contrast, saturation, and sharpness adaptively to maintain natural appearance across diverse graphical contents.
3Device complexity
If a single set of enhancement algorithms is used for all images, then the device complexity is reduced, but the image enhancement quality deteriorates for diverse graphical contents
Solution Approach 1:
The patent introduces dynamic adaptability by enabling the AI enhancement model to automatically adjust its parameters and processing strategies based on the detected scene mode and image characteristics. This dynamic behavior allows a single model to effectively handle diverse graphical contents without requiring multiple specialized algorithms for each scenario.
Data Source
AI summary
A method for enhancing quality of media uses an AI enhancing model built-in the client device to enhance the quality of video streams. The AI enhance module is pre-trained by using a neural network in the server to analyze differences between the decoded images and the raw images that are generated by the server. Wherein, the AI enhance module enhances decoded images by using algorithms which are defined by analyzing differences between the decoded images and the raw images. Such that, the enhanced images are visually more similar to the raw images than the decoded images do.