BROWSE
Detail
Conference Paper
Enhanced Adaptive Subdivision for Dynamic Mesh Compression to Improve Visual Quality in VR/AR Rendering
Cited 0 time in 
Share 
Share
- Authors
- Dayun Nam, Seong Yong Lim, Hyon-Gon Choo
- Issue Date
- 2025-10
- Citation
- International Symposium on Mixed and Augmented Reality (ISMAR) 2025, pp.1450-1460
- Language
- English
- Type
- Conference Paper
- Abstract
- Volumetric display technologies continue to advance, accompanied by increasing complexity in volumetric media services, particularly for XR (eXtended Reality) mobile devices. This trend underscores the need for efficient compression and transmission of volumetric video. In response, ISO/IEC JTC1/SC29 WG7 is developing a standard for mesh-based volumetric media, known as Video-based Dynamic Mesh Coding (V-DMC). V-DMC reduces geometry bitstream size through lossy mesh decimation and compensates for over-simplified geometry using mesh subdivision and displacement generation, which includes an adaptive face-and-edge update process. However, the existing subdivision framework often struggles to recover fine surface details due to rigid subdivision criteria and unintended mesh separations in geometrically irregular regions. This leads to visible artifacts-especially under dynamic lighting and viewpoint changes common in XR environments. To address these issues, this paper analyzes key performance bottlenecks and introduces several implementation strategies to improve both compression efficiency and visual fidelity through an enhanced adaptive subdivision framework. The proposed method introduces Level-of-Detail (LoD)-aware subdivision to improve the flexibility and effectiveness of the adaptive face-and-edge update. It also incorporates regional feature-aware adaptivity to mitigate geometric distortions, particularly in narrow or topologically complex regions. Experimental results using MPEG-I test content demonstrate a 10-20% reduction in displacement bitstream size and BD-rate gains of 2-5% compared to the V-DMC Test Model (TMM v12.0). These results confirm that the proposed method shows promise in enhancing both visual fidelity and compression efficiency, potentially contributing to the fulfillment of the stringent requirements of modern 3D rendering.
- KSP Keywords
- 3D Rendering, Adaptive subdivision, Display technology, Dynamic Mesh, Geometric distortions, Key Performance, Level of Detail, MPEG-I, Mesh subdivision, Mesh-based, Mobile devices