The Geopgraphic Data Support nodes are partly based on GeoVRML (http://www.geovrml.org/), and partly based on our own UTM-nodes which can be found in the SmallChange library.
The Geographic Data Support nodes makes it possible to create e.g. globe-spanning models that you can navigate around in without getting into float precision problems ('the shakes'). Some assistance from the application code is needed for this though; it is not a fully automatic solution.
In addition to new nodes, SoCamera and its derived nodes have been updated to support the SoGeoOrigin node.
The coordinate system support in the above nodes are currently a bit lacking (only a few systems/zones are supported), but this will improve over time. API-changes should not be needed to extend this, so patchlevel releases would be enough to add more coordinate systems.
The VRML2 counterparts of the above nodes, which would be a subpart of implementing the GeoVRML extension, are not yet in development, but they might show up for the next major release since a big part of the implementation will be shared.
Note: These above SoGeo*-nodes make the UTM*-nodes in the SmallChange library obsolete (for anyone who uses those), so the support for those nodes will eventually be deprecated and removed. They won't be removed over night though, but for new users: avoid starting to use the UTM*-nodes at all cost to avoid having to waste time on a port down the line.
Shader support is implemented through a set of new shader nodes, more or less compatible with the corresponding TGS extensions. Beyond GLSL, Coin also supports ARB shaders and Cg shaders (if the Cg library is installed). However, we will only recommend the use of the shader language GLSL and put our focus on that one.
See also the 'Shaders in Coin' page for more information.
Support for rendering shadows has been added to Coin through a set of shader-based nodes. The technique used is the Variance Shadow Maps technique (http://www.punkuser.net/vsm/).
A lot of nodes have had their rendering code optimized for VBO usage, giving great performance increases. However, this is not universally enabled - it is necessary to organize the scene graphs in specific ways for VBOs to be used.
See the VBO Rendering in Coin page for a detailed description.
The set of Sb* classes in Coin has been expanded greatly, with most notable a whole skew of SbVec and SbBox classes. The reasoning behind this move was that historically these classes were just added as needed, one by one, causing the stable branch of Coin and the development branch of Coin to diverge and become more and more incompatible as time went by. Also, extension nodes were created in peripheral libraries that used these new classes from the Coin development branch, and consequently couldn't be built gainst the stable branch of Coin. With this preemptive class set expansion, we hope to cause that the Coin stable branch and the Coin development branch stay closer in sync over a longer period, and that the incompatibility divergence will happen around major new feature sets instead of around minor details.
New Sb*-classes: SbBox2i32, SbBox3i32, SbBox3d, SbVec2b, SbVec2ub, SbVec2us, SbVec2i32, SbVec2ui32, SbVec3b, SbVec3ub, SbVec3us, SbVec3i32, SbVec3ui32, SbVec4b, SbVec4ub, SbVec4s, SbVec4us, SbVec4i32, SbVec4ui32, SbXfBox3d
Note: You can also find a good deal of these classes in TGS Inventor, so this was also a TGS compatibility move.
The exact same reasoning as for the Sb*-classes goes for the field classes.
New field classes: SoSFBox2s, SoSFBox2i32, SoSFBox2d, SoSFBox2d, SoSFBox3i32, SoSFBox3d, SoSFColorRGBA, SoSFDouble, SoSFVec2b, SoSFVec2i32, SoSFVec2d, SoSFVec3b, SoSFVec3i32, SoSFVec4b, SoSFVec4ub, SoSFVec4s, SoSFVec4us, SoSFVec4i32, SoSFVec4ui32, SoSFVec4d, SoMFColorRGBA, SoMFDouble, SoMFVec2b, SoMFVec2s, SoMFVec2i32, SoMFVec2d, SoMFVec3b, SoMFVec3s, SoMFVec3i32, SoMFVec4b, SoMFVec4ub, SoMFVec4s, SoMFVec4us, SoMFVec4i32, SoMFVec4ui32, SoMFVec4d
This node is added to make it possible to set up a custom camera with full control over the view volume. SoFrustumCamera is basically a mapping of the OpenGL glFrustum() method.
The typical use of SoFrustumCamera is in a CAVE or MultiPipe environment where it's often necessary to set up cameras based on the configuration of the different displays.
Below follows a non-exhaustive list of other, more minor, tweaks and tidbits.
New features for Coin v2.4.0 are here.