Avatar Rendering for Virtual Clothing Fitting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional augmented reality technologies for virtual clothing fitting lack accuracy in user satisfaction, leading to reduced user experience due to increased frame operations and unnecessary power consumption during the application process.
Innovation Solution
A multimedia device and method that captures a person using a camera, recognizes specific gestures to alter virtual clothing, and displays a second avatar based on detected skeleton information, optimizing frame operations and power usage by enabling and disabling constituent elements step-by-step.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional AR technology is used for virtual clothing fitting, then the service can be provided, but the accuracy of user satisfaction indication is degraded
Solution Approach 1:
The patent segments the avatar rendering process into multiple stages: initial avatar display, gesture detection stage, and detailed clothing alteration stage. By dividing the process, the system can activate skeleton detection and detailed rendering only when needed (upon gesture detection), thereby improving measurement precision of user satisfaction without continuously consuming resources that would degrade reliability.
Solution Approach 2:
The system performs preliminary actions by capturing user video data and creating an initial avatar display before actual clothing alteration is requested. This preliminary avatar is displayed without full skeleton detection, and only when a specific gesture is detected does the system activate the full skeleton information processing. This preliminary action approach improves accuracy when needed while avoiding continuous resource consumption.
2Productivity
If the number of frame operations is increased in the application process, then more detailed processing can be performed, but unnecessary power consumption increases
Solution Approach 1:
The patent implements periodic action by activating intensive frame operations and skeleton detection only at specific intervals triggered by user gestures, rather than continuously. The system operates in low-power mode during normal avatar display, then periodically activates high-performance processing when a gesture is detected, improving productivity when needed while minimizing power consumption during idle periods.
Solution Approach 2:
The system dynamically adjusts its operational mode based on user interaction. When no gesture is detected, the system uses minimal frame operations to maintain the basic avatar display. Upon detecting a specific gesture, the system dynamically transitions to a high-performance mode with increased frame operations and skeleton detection, then returns to low-power mode afterward, optimizing the balance between productivity and power consumption.
3Measurement precision
If skeleton information detection is continuously performed, then avatar accuracy is improved, but device complexity and power consumption increase
Solution Approach 1:
The patent extracts the skeleton information detection function from continuous operation and activates it only when specifically needed. The system separates the basic avatar display function (which runs continuously with low complexity) from the skeleton detection function (which is extracted and activated only upon gesture detection). This extraction approach improves avatar accuracy when needed while reducing overall device complexity and power consumption.
Data Source
AI summary
A method for controlling a multimedia device can include capturing, by a camera, at least one person located around the multimedia device; displaying a first avatar that is created based on video data of the at least one captured person, by referring to first information stored in a memory; recognizing a specific gesture of the at least one person; entering a first mode to alter virtual clothing; detecting skeleton information of the at least one person when the specific gesture is recognized; and displaying a second avatar that is created based on video data of the at least one captured person and the detected skeleton information, by referring to second information stored in the memory, in which the first information for the first avatar is different from the second information for the second avatar, and the second avatar is altered by touching a body of the at least one person.