1 Problem
With articulated railcars, it can happen that the entire driver's compartment becomes completely invisible in tight curves, as if one were suddenly sitting outside. As soon as you look around or change to the outside view, the vehicle is completely and normally visible again.
2 Background
This misrepresentation occurs because, for performance reasons, before the actual rendering, an attempt is made to "sort out" objects that are not visible anyway. This applies especially to objects which are outside the field of view.
In order to avoid having to check the complete, complex object, the ContentTool places a so-called bounding box around the object, which is selected so that all individual parts lie within this box. This box is then used for the collision and graphics check. The box does not take any animations into account in automatic mode. This is normally not a problem; however, in the case of articulated railcars (or certain scenery objects, such as animated tower cranes) the animation causes parts of the object to leave the bounding box considerably. The result is that the check occasionally produces incorrect results: Although parts of the object are in the field of view, they remain invisible because the bounding box is not and the check has come to the result "not visible".
The same effect occurs with the collision calculation - here, too, the bounding box first decides whether a collision could occur at all, and only then checks whether there is actually a collision based on the exact shape and arrangement of the individual parts.
3 Solution
In the object properties in the Object & Vehicle Tool there is a section called "Collision Box". There the box must be enlarged so that the train is inside the box at all times. For articulated railcars, it must be enlarged along the X-axis (in the example: Minimum-X to -8m, Maximum-X to 8m). It is also important that the checkbox "Automatic" is switched off, because otherwise the bounding box will be calculated again at the next mesh import and thus set back to the previous dimension.
The following picture should summarize the problem, solution and approach:
