Neural Makeup Rendering via In Vitro Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing virtual makeup rendering technologies face challenges in accurately parameterizing and rendering complex makeup effects like pearl and metallic finishes, requiring expert knowledge and being inefficient in memory usage, especially in mobile applications.
Innovation Solution
Incorporating a neural renderer that uses tensors of neural descriptors derived from in vitro images of reference makeup, allowing for automatic parameterization and improved rendering of makeup attributes, including pearl and metallic finishes, using generative models like GANs and VAEs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional virtual makeup rendering technologies are used, then manual parameterization can be performed, but expert knowledge is required and the process is inefficient
Solution Approach 1:
The patent replaces manual mechanical parameterization processes with an automated neural rendering system. The neural network automatically extracts makeup attributes from reference images and applies them to virtual try-on scenarios, eliminating the need for expert manual intervention and significantly improving rendering efficiency
Solution Approach 2:
The neural rendering system performs self-parameterization by automatically learning and extracting makeup attributes from reference images. The system serves itself by autonomously completing the parameterization task without requiring external expert knowledge, thereby streamlining the workflow
2Measurement precision
If traditional rendering methods are used, then memory usage can be managed, but rendering accuracy for complex makeup effects is insufficient
Solution Approach 1:
The patent transforms the rendering approach by changing from manual parameter specification to neural network-generated parameters. This parameter transformation enables accurate rendering of complex makeup effects like pearl and metallic finishes while optimizing memory consumption through efficient tensor representations
3Ease of operation
If manual parameterization is used, then control over rendering parameters is maintained, but the process requires expert knowledge and time
Solution Approach 1:
The system performs preliminary action by pre-training the neural network on extensive makeup datasets. This preliminary training enables the model to automatically and accurately parameterize new makeup references without requiring expert intervention, significantly reducing the time needed for parameterization while maintaining ease of operation
Data Source
AI summary
In some embodiments, a computer-implemented method of rendering reference makeup on an input image is provided. A computing system obtains a tensor of neural descriptors that represent attributes of the reference makeup generated by an attribute extractor from a reference image showing the reference makeup. The computing system uses a renderer to generate at least one rendered image based on the input image and the tensor of neural descriptors. The computing system provides the at least one rendered image for display on a display device.