Hair Feature Synthesis via Orientation Map Matching

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing graphic manipulation applications face challenges in accurately synthesizing hair features, particularly facial hair, without multiple images of the subject, leading to time-consuming and laborious editing processes, and result in unrealistic or lacking details due to insufficient training data.

Innovation Solution

The system synthesizes image content by transforming user-provided guidance data into an input orientation map, matching it with a high-resolution exemplar orientation map, and applying color information to generate desired hair features, allowing for efficient and intuitive creation of high-quality hair features without relying on extensive training images.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If multiple images are used to generate a 3D reconstruction of facial hair, then the accuracy of hair modeling is improved, but the complexity of the process and time required increase significantly

Engineering Contradiction:
Improvehair modeling accuracyVSAvoidprocess complexity
Core Design Contradiction:
Manufacturing precisionVSDevice complexity

Solution Approach 1:

The patent extracts only the essential orientation information from images to create a simplified 2D orientation map, rather than performing full 3D reconstruction. This extraction approach maintains hair modeling accuracy while eliminating the complexity of 3D reconstruction processes

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent transitions from 3D reconstruction to 2D orientation map representation, changing the dimensional approach to hair modeling. This dimensionality reduction simplifies the process while preserving the essential orientation data needed for accurate hair feature synthesis

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

2Adaptability or versatility

If a deep neural network is trained to learn facial features into a feature space, then the ability to generate hair features is improved, but the intuitiveness of editing and realism of results deteriorate

Engineering Contradiction:
Improvehair feature generation capabilityVSAvoidediting intuitiveness
Core Design Contradiction:
Adaptability or versatilityVSEase of operation

Solution Approach 1:

The patent uses an exemplar orientation map as a template or copy that guides the synthesis process. Instead of directly editing complex feature space vectors, the system copies and adapts orientation patterns from the exemplar map, making the process more intuitive while maintaining generation capability

Inventive Principle:
Principle #26Copying

Solution Approach 2:

The orientation map serves as an intermediary between the neural network's feature understanding and the final hair feature generation. This intermediate representation makes editing more intuitive by providing a visual, spatial interface rather than abstract feature space manipulation

Inventive Principle:
Principle #24Intermediary (Mediator)

3Productivity

If a deep neural network is trained with insufficient data, then the training time is reduced, but the realism and detail of generated hair features worsen

Engineering Contradiction:
Improvetraining efficiencyVSAvoidhair feature realism
Core Design Contradiction:
ProductivityVSManufacturing precision

Solution Approach 1:

The patent performs preliminary extraction of orientation maps from training images before neural network processing. This pre-processing step creates structured orientation data that improves training efficiency and ensures that even with limited data, the essential orientation information is preserved for realistic hair generation

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent changes the training parameters from raw pixel data to extracted orientation map data. This parameter transformation allows the network to learn from fewer images while maintaining high realism, as the orientation maps provide concentrated, essential structural information

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentUS10515456B2Synthesizing hair features in image content based on orientation data from user guidance
Publication Date: 2019.12.24 ADOBE INC
  • US10515456B2 patent drawing
  • US10515456B2 patent drawing
  • US10515456B2 patent drawing

AI summary

Certain embodiments involve synthesizing image content depicting facial hair or other hair features based on orientation data obtained using guidance inputs or other user-provided guidance data. For instance, a graphic manipulation application accesses guidance data identifying a desired hair feature and an appearance exemplar having image data with color information for the desired hair feature. The graphic manipulation application transforms the guidance data into an input orientation map. The graphic manipulation application matches the input orientation map to an exemplar orientation map having a higher resolution than the input orientation map. The graphic manipulation application generates the desired hair feature by applying the color information from the appearance exemplar to the exemplar orientation map. The graphic manipulation application outputs the desired hair feature at a presentation device.