산업융합연구 (Journal of Industrial Convergence)

대한산업경영학회 (DAEHAN Society of Industrial Management)
권호 표지
DOI QR코드

DOI QR Code

공간 컴퓨팅 적용을 위한 3D 생성 AI 플랫폼 비교 연구

Comparative Study of 3D Gen-AI Platform for Spatial Computing

  • 서동희 (남서울대학교 가상현실학과)
  • Donghee Suh (Dept. of Virtual Reality, Namsoeul University)
  • 투고 : 2024.08.19
  • 심사 : 2024.10.20
  • 발행 : 2024.10.28
PDF 다운로드
⟨ 이전 논문 다음 논문 ⟩

초록

본 연구는 3D 생성 AI 플랫폼의 기능과 효율성을 비교 분석하여 3D 콘텐츠 제작 공정에서의 실무 적용성을 평가하고 개선 방향을 제시하는 데 목적을 둔다. 9개의 플랫폼을 조사한 후, 최신 기술 활용 여부, 호환성, 사용자 접근성을 기준으로 4개 플랫폼을 선정하였다. 각 플랫폼에 동일한 프롬프트를 적용해 3D 오브젝트를 생성하고 결과를 살펴보았다, 사용자 지정이 가능한지, 실감 콘텐츠 제작에 이점이 있는지, 제작에서의 효율성을 높일 수 있는 것인지, 무료 테스트가 가능하거나 가성비가 좋은지 등을 중심으로 분석하였다. 연구 결과, 'Meshy'와 'Tripo'는 빠른 생성 속도와 효율적인 폴리곤 최적화로 우수한 성능을 보였으며, 'Spline'은 다양한 미디어 적용 기능을 제공하지만 품질에 제한이 있었다. 이를 통해 3D 생성 AI 플랫폼이 각기 다른 제작 파이프라인과 사용자 요구에 따라 적합성을 달리한다는 것을 확인했다. 본 연구는 3D 콘텐츠 제작에 관심있는 실무자들에게 플랫폼 선택을 위한 실질적인 가이드를 제공하고, 3D 생성 AI 기술의 발전 방향에 대한 통찰을 제시하여 향후 연구와 산업 적용에 기여할 것으로 사료된다.

This study aims to compare and analyze the functionality and efficiency of 3D generation AI platforms to evaluate their practical applicability in the 3D content creation process and suggest improvement directions. A total of nine platforms were researched using search, and four platforms were selected based on their utilization of the latest technology, compatibility, and user accessibility. We used the same prompts to create 3D objects on each platform and analyzed the results, focusing on whether they were customizable, beneficial for creating immersive content, efficient in production, free to test, or good value for money. The results showed that Meshy and Tripo performed well with fast generation speeds and efficient polygon optimization, while Spline offered a wide range of media application capabilities but was limited in quality. We found that different 3D generation AI platforms are suitable for different production pipelines and user needs. This study provides practitioners interested in 3D content creation with a practical guide for platform selection and provides insights into the future direction of 3D generative AI technology, which will contribute to future research and industrial applications.

키워드

과제정보

Funding for this paper was provided by Namseoul University year 2023.

참고문헌

  1. RODIN Diffusion. (2022). A Generative Model for Sculpting 3D Digital Avatars. Microsoft RODIN Diffusion from https://3d-avatar-diffusion.microsoft.com/
  2. METAHUMAN. (2021). High-fidelity digital humans made easy. EPIC GAMES METAHUMAN from, https://www.unrealengine.com/en-US/metahuman
  3. Ben Poole, Ajay Jain, Jonathan T. Barron, & Ben Mildenhall. (2022). DreamFusion: Text-to-3D using 2D Diffusion. arXiv preprint arXiv: 2209.14988, https://arxiv.org/abs/2209.14988
  4. Jiaxiang Tang, Jiawei Ren, Hang Zhou, Ziwei Liu, & Gang Zeng. (2023). DreamGaussian: Generative Gaussian Splatting for Efficient 3D Content Creation, arXiv preprint arXiv: 2309. 16653. The 12th International Conference on Learning Representations. https://dreamgaussian.github.io/
  5. Xie, K., Lorraine, J., Cao, T., Gao, J., Lucas, J., Torralba, A., Fidler, S., & Zeng, X. (2024). LATTE3D: Large-scale Amortized Text-To-Enhanced3D Synthesis. The 18th European Conference on Computer Vision. https://research.nvidia.com/labs/toronto-ai/LATTE3D/
  6. Raphael Bensadoun, Tom Monnier, Yanir Kleiman, Filippos Kokkinos, Yawar Siddiqui, Mahendra Kariya, Omri Harosh, Roman Shapovalov, Benjamin Graham, Emilien Garreau, Animesh Karnewar, Ang Cao, Idan Azuri, Iurii Makarov, Eric-Tuan Le, Antoine Toisoul, David Novotny, Oran Gafni, Natalia Neverova, & Andrea Vedaldi. (2024). Meta 3D Gen, https://ai.meta.com/research/publications/meta-3d-gen/
  7. Dwivedi, Y. K., Hughes, L., Baabdullah, A. M., Ribeiro-Navarrete, S., Giannakis, M., Al-Debei, M. M., ... & Wamba, S. F. (2022). Metaverse beyond the hype: Multidisciplinary perspectives on emerging challenges, opportunities, and agenda for research, practice and policy. International journal of information management, 66, 102542. https://doi.org/10.1016/j.ijinfomgt.2022.102542.
  8. Yongwoon Shim, (2023), 테크 트렌드 애플, 메타버스를 공간 컴퓨팅으로 재정의하다. 한경 BUSINESS,(1438), 88-89.
  9. Liu, Jian, et al. (2024). "A comprehensive survey on 3D content generation." arXiv preprint arXiv:2402.01166 (2024).
  10. AI Report. (2024.07.03.). 'Meta 3D Zen' creates 3D models in a minute, Digital Today, https://www.digitaltoday.co.kr/news/articlePrint.html?idxno=523763
  11. Lawton George. (2022). "Spatial computing." TechTarget. https://www.techtarget.com/searchcio/definition/spatial-computing
  12. Karan Wankhede and Rashmi Gourkar. (2024). Futuristic Rise of Spatial Computing, International Journal Of Advance Research And Innovative Ideas In Education, 10.
  13. Frontways Larry. (2024). Battle Talent... but in Mixed Reality, Youtube. https://youtu.be/7wFdUHYr-bs?si=cM1LmghoAub8f9b0
  14. Alex Heath. (2023). This is Meta's AR/VR hardware roadmap through 2027, The Verge, https://www.theverge.com/2023/2/28/23619730/meta-vr-oculus-ar-glasses-smartwatch-plans
  15. Digital Insight, (2023). NIA, How spatial computing will change the world, https://www.nia.or.kr/site/nia_kor/ex/bbs/View.do?cbIdx=82618&bcIdx=26112&parentSeq=26112
  16. Neo Jun Rong Jeffrey, Won Andrea Stevenson and Shepley Mardelle McCuskey. (2021). Designing Immersive Virtual Environments for Human Behavior Research., Frontiers in Virtual Reality(2673-4192), 2. DOI: 10.3389/frvir.2021.603750
  17. Autodesk. (2024). Retopology: Unlocking new horizons in 3D artistry. https://www.autodesk.com/solutions/retopology
  18. Hennig-Thurau, T., Aliman, D.N., Herting, A.M. et al. (2023). Social interactions in the metaverse: Framework, initial evidence, and research roadmap. J. of the Acad. Mark. Sci. 51, 889-913. https://doi.org/10.1007/s11747-022-00908-0
  19. Wu, Sixue, Le Xu, Zhaoyang Dai, & Younghwan Pan. (2023). Factors Affecting Avatar Customization Behavior in Virtual Environments. Electronics 12, no. 10: 2286. https://doi.org/10.3390/electronics12102286
  20. Jung-Hun Lee. (2014). Facebook, Facebook acquires 'virtual reality device' Oculus for $2.2bn, E-Daily. https://m.edaily.co.kr/news/read?newsId=01190646606027256&mediaCodeNo=257