- 8,715
It comes down to two aspects, for me:
1. Whether pre-set mechanically damaged states are sufficient.
2. Whether a physically realistic mechanical damage system lends itself to extrapolation to a visual representation.
On the first point, taking GT6 as the example, you get a damaged state for the engine and the four wheels. This appears to work on a threshold, according to how hard you hit something. There were two types of damaged state for the wheels: a wobble; a misalignment (steering).
The problem with this approach is that it becomes predictable (if I just nudge this, I'll get away with it), and the severity of the consequence is fixed, no matter the severity of the impact.
Now, there are easy ways to extend that approach, but it comes down to juggling special cases and just feels fake. On that basis it is not, in my opinion, appropriate for a sim.
For the second point, it first comes down to what is important, in respect of the driving physics, in order to inform the overall approach that should be taken.
In my opinion, you should start with suspension damage, because it is the most likely direct result of rubbing paint or clipping trackside objects (as opposed to the indirect result of leaving the track and hitting something else). Modeling suspension misalignment based on its geometry and the forces involved in the impact is relatively easy using "formulas" (and lots of data, or generic guesses). The visual representation of that is trivial.
From there, you should move to chassis damage, which would include displacement of the wheels caused by damage to the suspension's supporting structure and, potentially, changes to its stiffness (if modeled). At that point, you can quite easily morph the body to match the underlying changes (iRacing does this).
Substantial deformation of the chassis in key areas, such as the suspension, engine and drivetrain, can be used to create various "damage effects" on those parts, with as much detail as you can stomach, because the deformation is there to interpret how you please. Deformation to the supporting structure of glass-work or doors etc. can inform whether to "break" them or not, and direct impacts on or around certain objects such as headlights, wings, mirrors and other nubbins are easy to implement (which PD already do one way or another).
A more complicated example would be bodywork damage, where it is separate from what might be classed as the rigid "chassis" structure. The most important aspects here might be aerodynamics and cooling. You could run separate deformation / detachment processes that are purely visual in nature on any non-rigid bodywork parts (of the kind that PD have already produced, just with more finesse), but informed by the more physical chassis modeling via some form of witchcraft - i.e. "formulas".
Anything that looks convincing will be pretty close in respect of the effect on aero (maybe just measure total deformation in key areas), such that the visual information can be used to inform the actual "damage" in this case.
So on that basis, I believe it is not helpful to separate the concepts of "visual" and "mechanical" damage, instead we should concentrate on the best way to produce believable consequences for activities that might induce "damage". I believe that it will not take an excessive amount of effort to create a pleasing and useful source of "visual" feedback to go with the physics-oriented consequences of "mechanical" damage.
Last edited:
