This application demonstrates an efficient way to transmit 3D scans consisting of hundreds of thousands, or even millions of vertices. Such models can provide detailed reconstructions of objects and locations for both laypeople and experts to observe, manipulate and study remotely, over the web. While 3D streaming is a very active research direction, with multiple proposals competing over transmission speed and quality, most such approaches are essentially based in a pre-processing step where the X3D model is converted into an appropriate streamable (and usually binary) format. The display of such objects requires specialized players directly accessing WebG capabilities. In our approach, we opted for a face-by-face approach, where a given X3D model is not pre-processed, but instead separated into subsets on-the-fly. The model is split into "chunks" of customizable size, which are progressively transmitted to the client, and displayed as an X3DOM scene. Technologies such as X3DOM, WebSockets, Spring, and Stomp are used for the implementation, to achieve a fast, scalable streaming approach that directly streams text-based declarative data to a client with no binarization or "black-box" processing steps.