EUVIMEDEUVIMED entdecken
Crossref

Towards High-Quality and Disentangled Face Editing in a 3D GAN

Kaiwen Jiang, Shu-Yu Chen, Feng-Lin Liu, Hongbo Fu, Lin Gao

IEEE Transactions on Pattern Analysis and Machine Intelligence · 2025-04 · 47 (4) · 2533-2544

DOI: 10.1109/tpami.2024.3523422 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Transactions on Pattern Analysis and Machine Intelligence, 2025-04. Verfasst von Kaiwen Jiang und 4 weiteren Autor:innen. Die Arbeit wurde bisher 8-mal zitiert (Crossref) und nennt 55 Literaturangaben. Gefördert von: National Natural Science Foundation of China, Innovation Funding of ICT, CAS, Beijing Municipal Natural Science Foundation for Distinguished Young Scholars.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Recent methods for synthesizing 3D-aware face images have achieved rapid development thanks to neural radiance fields, allowing for high quality and fast inference speed. However, existing solutions for editing facial geometry and appearance independently usually require retraining and are not optimized for the recent work of generation, thus tending to lag behind the generation process. To address these issues, we introduce NeRFFaceEditing, which enables editing and decoupling geometry and appearance in the pretrained tri-plane-based neural radiance field while retaining its high quality and fast inference speed. Our key idea for disentanglement is to use the statistics of the tri-plane to represent the high-level appearance of its corresponding facial volume. Moreover, we leverage a generated 3D-continuous semantic mask as an intermediary for geometry editing. We devise a geometry decoder (whose output is unchanged when the appearance changes) and an appearance decoder. The geometry decoder aligns the original facial volume with the semantic mask volume. We also enhance the disentanglement by explicitly regularizing rendered images with the same appearance but different geometry to be similar in terms of color distribution for each facial component separately. Our method allows users to edit via semantic masks with decoupled control of geometry and appearance. Both qualitative and quantitative evaluations show the superior geometry and appearance control abilities of our method compared to existing and alternative solutions.

Förderung

  • National Natural Science Foundation of China
  • Innovation Funding of ICT, CAS
  • Beijing Municipal Natural Science Foundation for Distinguished Young Scholars
  • Beijing Municipal Science and Technology Commission

Zitieren

APA

Jiang, K., Chen, S. Y., Liu, F. L., Fu, H., & Gao, L. (2025). Towards High-Quality and Disentangled Face Editing in a 3D GAN. IEEE Transactions on Pattern Analysis and Machine Intelligence, 47(4), 2533-2544. https://doi.org/10.1109/tpami.2024.3523422

Vancouver

Jiang K, Chen SY, Liu FL, Fu H, Gao L. Towards High-Quality and Disentangled Face Editing in a 3D GAN. IEEE Transactions on Pattern Analysis and Machine Intelligence. 2025;47(4):2533-2544. doi:10.1109/tpami.2024.3523422

BibTeX
@article{Jiang2025,
  author = {Jiang, Kaiwen and Chen, Shu-Yu and Liu, Feng-Lin and Fu, Hongbo and Gao, Lin},
  title = {Towards High-Quality and Disentangled Face Editing in a 3D GAN},
  journal = {IEEE Transactions on Pattern Analysis and Machine Intelligence},
  year = {2025},
  volume = {47},
  number = {4},
  pages = {2533--2544},
  doi = {10.1109/tpami.2024.3523422}
}
Kostenlos starten →