Towards a 3D transmission format for the Web

Kristian Sons and Philipp Slussalek


Recent advances in the internet technology landscape raised the demand for efficient transmission of large amounts of binary data. This includes HTML5 video and audio streaming, data to do audio processing using the upcoming Web Audio API, and 3D related data to used by the WebGL API. However, unlike video, audio or images, 3D data is very inhomogeneous and can be highly structured. Thus it’s more challenging to establish a common 3D transmission format for the Web.

Beside binary XHR based on ArrayBuffers, which is commonly used but insufficient for many use-cases, developers found creative solutions to transmit 3D data into the browser. This includes exploitation of fast native parsing of JSON or UTF-8, and use of 2D images to transfer 3D data. However, all approaches that can be found in scientific literature or “in the wild” address only a subset of the challenges need to be solved for a general, common, and efficient transmission of 3D data. Additionally, few serious comparisons can be found in literature: lossless and lossy approaches are compared without mentioning, or approaches are tested against very specific data sets only.

In this position paper we want to recall the factors that are decisive for an efficient transmission in the web context, derive some requirements on a 3D transmission format, review some approaches with respect to our requirements, and sketch a possible solution including current issues.