The Development of an Efficient Pipeline for Virtual Human Creation Using Accessible Modern Technologies Sunny Thapa Magar
- 주제(키워드) Digital Human , Virtual Human , 3D Character Production Pipeline , 3D Software , AI
- 주제(DDC) 302.23
- 발행기관 아주대학교 일반대학원
- 지도교수 Suk Haejung
- 발행년도 2025
- 학위수여년월 2025. 8
- 학위명 박사
- 학과 및 전공 일반대학원 라이프미디어협동과정
- 실제URI http://www.dcollection.net/handler/ajou/000000034891
- 본문언어 한국어
- 저작권 아주대학교 논문은 저작권에 의해 보호받습니다.
초록/요약
Virtual humans are developed as embodied agents to replicate human-like behaviors by employing social cues, including facial expressions, gestures, and body movements. These entities represent humans through computer-generated models in virtual spaces, capable of simulating motions, evaluating workloads, and interacting with virtual objects. Key features of virtual humans include a structure mimicking the human skeleton, adherence to human biomechanics, and the ability to integrate into virtual environments. Significant challenges persist despite advancements in 3D design software, game engines, and real-time rendering technologies enabling photorealistic virtual humans. Many tools available today require a high level of specialized expertise to operate and consist of various limitations, including limited appearance customization, platform dependency, licensing constraints, proprietary ownership, complex learning curve, and limited scalability. This thesis investigates the performance of current digital and virtual human creation pipelines across key factors such as Appearance, Appearance Customization, Visual Fidelity/Realism, Platform Dependency/Export Compatibility, Licensing Constraints, Time Efficiency, Scalability, Skill, and Learning Curve. This research proposes a virtual human creation pipeline specifically designed for academic research and small-scale production. The proposed pipeline was evaluated against an existing academic pipeline. It substantially improved development speed, platform adaptability/- Export Compatibility, customization, and interaction while reducing the required technical skill level. Built using accessible tools such as Reallusion, Blender, Unity, and AI-based technology, the pipeline shows that a single tool is insufficient for end-to-end virtual human creation, explaining that integration across platforms and automation are essential. The resulting pipeline offers a scalable and interactive solution suitable for researchers and small-scale production. While it relies on a combination of costly tools, the reduced need for manual effort and faster production time help offset labor costs, potentially lowering overall development expenses in terms of time and team size. Keywords: Digital Human, Virtual Human, 3D Character Production Pipeline, 3D Software, AI
more목차
1 Introduction 1
1.1 Background 3
1.2 Problem Statement 5
1.3 Study Objectives and Scope 6
1.3.1 Objectives 6
1.3.2 Scope 6
1.4 Structure of Dissertation 7
2 Literature Review 11
2.1 History and Evolution of Virtual Humans 11
2.2 Human Traits for Virtual Human Design 14
2.3 Digital Human Vs Virtual Human 16
2.4 Study on Existing Pipelines and Tools 18
2.4.1 3D Character Creation Tools 18
2.4.2 Virtual Human Creation Tools 23
2.4.3 Digital Human Creation Pipelines 25
2.5 Additional Tools and Technologies 31
2.6 Implementing Human-like Traits in Virtual Humans 34
3 Methodology 37
3.1 Literature Review and Tool Survey 37
3.2 Comparative Analysis 38
3.3 Pipeline Design and Practical Implementation 39
3.4 Comparative Evaluation of Pipelines 40
4 Comparative Analysis 41
4.1 Comparison of Virtual Human Creation Tools 41
5 Pipeline Design and Practical Implementation for Validation 49
5.1 Proposed Pipeline Design for Digital Human Creation 49
5.1.1 Concept and Design 50
5.1.2 3D Modeling 51
5.1.3 Texturing and Shaders 53
5.1.4 Grooming 54
5.1.5 Rigging 56
5.1.6 Animation 58
5.1.7 Implementation 60
5.2 Workflow of Virtual Human Generation 62
5.3 Practical Implementation for Validation 68
5.3.1 Verbally Interactive Real-Time Virtual Human 68
5.3.2 Implementation in other Real-world Projects 71
6 Comparative Evaluation 73
6.1 Comparison of Virtual Human Creation Pipelines 73
6.1.1 Pipeline Comparison Based on Workflow 74
6.1.2 Pipeline Comparison Based on Key Factors 89
7 Discussion 95
7.1 Discussion 95
7.1.1 Key Findings 95
7.1.2 Strengths of the Proposed Pipeline 96
7.1.3 Limitations 97
7.1.4 Revisiting the Research Questions 98
7.1.5 Implications 100
8 Conclusion 101
Bibliography 102
Appendix A List of Publications 119
A.1 SCI/SCIE Journal Papers 119
A.2 KCI Journal Papers 119
