Mesh Optimization as a Service

Motivation

High-resolution polygon meshes, arising from manual modeling or from 3D scanning, often consist of up to several million polygons. For use with real-time visualization applications, they must be simplified to a more compact representation. Such a task usually involves a lot of manual interaction, strongly influences the visual quality and can become time-critical, especially if a large number of 3D assets needs to be optimized. This is where our fully automatic Mesh Optimization as a Service (Mesh OaaS) infrastructure comes in. Using instant3Dhub, our 3D mesh optimization pipeline can be easily integrated into external applications.

The benefits of using our instantUV / Mesh Optimization and Processing Service (MOPS) component can be summarized as follows:

  • Fully automatic optimization of 3D mesh data, resulting in compact, high-quality 3D models
  • Various mesh processing tools at your hand: Simplification, Parameterization / Generation of Texture Coordinates, Mesh Smoothing, and more
  • Perfect integration with our Web-based delivery and rendering tools:
    Just input your unoptimized 3D mesh to receive a high-performance Web site application
  • Integration possible via instant3Dhub transcoding infrastructure, providing features such as load balancing, caching, and a RESTful API
  • Flexible Integration: As a service, as standalone command line tool, or as C++ library

More information about the mesh processing features available through instantUV / MOPS can be found on the official homepage:

Flyer

download-flyer

Click the above thumbnail to download our flyer about automatic 3D mesh optimization

Mesh Reduction: Optimal Quality at Minimum Mesh Complexity

cruciform image

Even for highly complex input data, our 3D mesh optimization pipeline creates compact 3D representations in a fully automatic fashion. Several variants of the 3D asset, tailored into a maximum of a few thousands of polygons towards the specific needs of different device classes, can be easily generated – without any manual interaction.

Our mesh processing and optimization service automatically reduces input meshes to compact, yet still high-quality representations. To achieve this, we employ established approaches, as well as latest research results, from the fields of mesh processing and computer graphics:

  • First, input data is geometrically simplified (Mesh Simplification), with the aim of preserving the overall shape of the input mesh as closely as possible.
  • Next, a mapping from each object’s 3D surface to a 2D image domain is computed (Mesh Parameterization). This, finally, enables the preservation of the high-resolution surface details within texture images (Texture Synthesis).
  • At the end of the optimization process, a compact, textured 3D polygon mesh has been created. This optimized mesh is visually nearly identical when compared to the original 3D asset, but it consumes much less bandwidth and storage space.

As can be seen in the above figure, reducing the polygon count by two or more orders of magnitude is easily possible in many cases – event without notable difference in the final visual appearance.