This is the old SGI Teapot, with a Shader that takes light intensity, and assigns colours based on ranges of intensity. The model is available here for your enjoyment.

To see what the model looks like without the shader, you can get the model here.

Two Cylinders, rotating around each other, one with per-fragment lighting, the other lit on a per-vertex basis.

Fragment lighting is more time consuming, but sometimes the results are worth it. (run the test to really see the differences in action)

The Android FreeX3D App allows you to set "best" or "fast" rendering; feel free to play around with the settings when viewing your own models so you see for yourself what the trade-offs are for each style of rendering. here is this model.

A flat IndexedFaceSet, 20x20 vertices per side, is deformed in the Vertex shader. A TimeSensor feeds in a floating point number, and the Shader moves the vertices according to an algorithm running on the GPU.

This example is in two formats, a "ProgramShader" format here, or written as a "ComposedShader" here.

An industry-standard image ("Lenna", or "Lena", depending on the reference) used as input to a Sobel Operator for edge finding. The image size (512x512) is hard-coded into the Shader, but it could be passed in as a field.

Here is the file for you to try.

(if you can not click on the link, the image is here; save it and modify the source code to point to it)

These examples contain GLSL code to show the effects of calculating lighting on either fragments or on vertices.

In this series of tests, two identical black squares are made up of triangles; a SpotLight is programmed to sweep across both black squares.

In all sequences, Vertex lighting is performed on the left side, Fragment shading on the right. The higher the triangle count (thus, the number of vertices) the "better" the SpotLight when Vertex shading. 16 Vertices is horrible, but in the 4 vertex example, the SpotLight never hits a vertex, thus is never shown.

In all sequences, Fragment lighting is (or, should be) identical.

Better is the high vertex count model for you to try.

Poor is the 16 vertex count model for you to try.

Abysmal is the 4 vertex count model for you to try.

Take a normal generated 3D terrain, manipulate vertices to change from a normal view.

Original View is the normal terrain view.

Fish Eye view with movement programmed in is the one with the GLSL shader to show the fisheye view. There is a PositionInterpolator in there to animate the Viewpoint.

As of 2016, IOS and OSX builds are not supported. At one time, FreeWRL was distributed by Apple for their desktop systems, but maintenance support lapsed when The Communications Research Centre stopped support for research into Virtual Reality and 3D Visualization, as these topics did not directly support the then-current views on how to manage wide wireless spectrum, of which CRC and its parent organization were tasked.

Of course, as of the 2016 timeframe, the new crop of researchers and managers at CRC realize the importance of visualization in their tasks, and have adopted the technology whole-heartedly, much to their credit, and are well positioned for the coming decade. (John Stewart's opinion)

The status bar shows, in this view, the current Viewpoint; it will show other text depending on FreeWRL run status.

This is test 16, available at http://freewrl.sf.net/tests/16.wrl, and in the distribution.

The iPhone app was demonstrated at SIGGRAPH 2011 in Vancouver, the successful port of FreeWRL for the iPhone required moving from fixed-function OpenGL to OpenGL-ES; Shaders required to be built at load-time. (reference and more pictures at this mail archive)

A 200x200 IndexedFaceSet with z=0 on all vertices is manipulated by a Vertex Shader. The X3D program with integral shader code is here.

CRC Canada was contracted to the U.S. Department of Defense to provide High Resolution Geospatialized 3D model rendering additions to FreeWRL. Contract No. N00244-08-P-1769, awarded 27 May 2008, issued by FISC San Diego, Code 200, 1 University Circle, Monterey CA.

The original panoramic shot is here and the X3D file is here .

You know what is neat about this model? In 2008 this 3D model was shown, which started (in 2009) a sub-project documenting areas around Ottawa, Canada, creating 360 degree images by stitching a large sequence of photographs taken by a robotic camera platform. The Canada Parliament buildings were photographed September 23, 2009. The camera platform also was used in the province of Alberta.

Fast forward to 2016, and the "in thing" are 360 degree models, VR viewing apparatus, 8 years after our teams' first efforts went on line. Of course, now we have much better hardware; 360 degree movies (instead of one still image) and viewing platforms that make it seem almost as if you are there.

It is interesting to see the fantastic progress happening.

This model was significant as it showed (circa 2007) that real-time navigation through a virtualized landscape was possible.

The FreeWRL team coded Avatar-Object collision in OpenCL, which increased rendering speed by approximately 10 fold on devices that can/could support OpenCL/OpenGL sharing of data.

The X3D file is here