EXPLORING 3D MESH COMPRESSION: METHODS, TRADE-OFFS, AND APPLICATIONS

Abstract

Three-dimensional (3D) mesh compression has become increasingly important across a range of fields—from virtual and augmented reality to computer-aided design, gaming, and large-scale scientific visualization—where both storage constraints and real-time rendering demands continue to escalate. As 3D models grow in complexity and resolution, efficient compression becomes critical for enabling smooth interaction, fast transmission, and scalable visualization on modern hardware. This paper provides a structured overview of the primary categories of 3D mesh compression techniques: single-rate compression, progressive compression, random-accessible compression, and hybrid approaches that combine progressive refinement with selective decoding. Single-rate methods offer compact, fixed-level encodings with simple and fast decoding pipelines, though they lack adaptability in dynamic scenarios. Progressive methods construct a coarse-to-fine hierarchy, allowing for level-of-detail management and gradual refinement during streaming or transmission. Random accessible techniques emphasize localized decoding, enabling targeted access to specific mesh regions without decompressing the entire model—useful in interactive or memory-limited settings. Hybrid approaches merge these advantages, supporting both progressive reconstruction and partial access for highly responsive performance. The paper also briefly considers recent advances in neural-based compression, which offer adaptive, data-driven encoding schemes that could further enhance compression efficiency. Based on this survey, we conclude that progressive and random accessible mesh compression methods offer the most practical benefits for real-world applications that require real-time responsiveness and efficient resource usage. These approaches show strong potential and should be the focus of future research and optimization.