Speaker: Federico Rocha

Viewing extremely large 2D and 3D models in the browser in real time is a difficult problem because of limited client-side computational resources, limited bandwidth, and the challenges of using JavaScript for high-performance computing. By introducing many heavily optimized techniques and strategies, Autodesk Forge Viewer is able to address these difficulties and render millions of meshes with gigabytes of memory at interactive rates.  

In this session, we will explore in detail some of the most interesting of these techniques, including progressive rendering to reduce time to first pixel, geometry consolidation to reduce draw overhead, the use of flat arrays instead of objects to reduce memory-allocation and copying overhead, data alignment for fast memory access, the use of a cheaper non-photorealistic graphics style for certain effects, and CPU profiling to identify further optimization opportunities.  

Real-time rendering of large models is a complex problem with many interesting facets. We have been developing Forge Viewer for five years and believe that we have found many interesting solutions from which the community can benefit.