Today, the development branch v1.1 has been made available, with version 1.1.0.0 being the first incarnation. There might be further refinements, corrections and additions released in subsequent versions before this eventually results in the stable branch v1.2, which will be recommended for the average user.
The development branch is primarily intended for developers to study or experiment with, but users might also find it useful to get a preview of the new features that are offered. However, it should be pointed out that the new set of features might be incomplete, be subject to some problems or yet undetermined bugs. Also, features might change at any time without prior notice until the new features are considered complete.
What is new
Well, basically, support for 3D cabs has been introduced with the panel.animated file. As a demonstration, I have made the 113-1000atccab-3D train, which is a non-realistic add-on to PM's 113-1000atccab train with the sole purpose of showing off how to make a 3D cab. You can find this demo train on the Developing for openBVE pages.

While the animated object format itself has not undergone any changes, the set of supported variables has been greatly expanded to meet the needs of creating a working animated panel. All changes are specifically highlighted as such in the documentation on the animated object format.
Along with the 3D cabs, there is now support for rotating the view with the mouse. Just press the left mouse button at any time to enable mouse grab, then move the mouse to rotate the view. Clicking again
disables the mouse grab. Enabling and disabling mouse grab has been specifically done for use in window mode, while it wouldn't be technically necessary in fullscreen mode.
Potential
Creating a 3D cab with the current textual file formats is a challenge, and is likely more difficult to do than the bitmap-based 2D panel formats. However, the feeling of a 3D cab is vastly different from its 2D counterpart, and thus I would expect that at least some developers will be highly interested in started to create 3D cabs.
Just as panels for BVE Trainsim started with hand-drawn ones of low quality and have only eventually emerged in highly photo-realistic ones, I also suspect the first set of 3D cabs that might be produced in the upcoming months or in the upcoming year to be of also relatively low quality. 3D cabs from other train simulators tend to use very detailed geometry at the expense of good textures, resulting in a highly CGI look, while 2D panels can offer extremely detailed textures at the expense of geometry, or actually with no geometry at all. Therefore, developers will have to learn and experiment with the trade-off of using textures and geometry, and it might take a while before everyone finds their personal preference.
Ways for developers to create 3D cabs
The CGI way
Build the cab in truly 3D geometry. If you have photographs of the cab, you will likely to have cut out individual pieces so you can apply them to individual faces. Most of the work would be involved in building the geometry, while photorealism is hard to achieve.
Cylindrical projection
A photorealistic way of doing things. Shoot many pictures of the cab from a perfectly fixed location so you can cover a 360° field of view. Then stitch the individual photographs together to form a cylindrical mapping. You will need a fairly decent imaging program for this, such as Photoshop. The texture would look somewhat like this, but when mapped to a cylinder, it will look perspectively correct. Thus, for the geometry, you only need to create a cylinder and map the texture onto it. This is basically how the background image works in openBVE. The bottleneck is that this approach looks only good from a fixed location, and only in the horizontal. If you pitch the view too much up/down, you will see the cylinder. Also, animated parts might be harder to integrate.
Spherical mapping
Basically the same as cylindrical projection, only that you map to a sphere instead of a cylinder. This gives a perfectly photorealistic view from a fixed location, but would be tremendously difficult to produce. This is basically how Google Street View works.

