Show Posts
1
Early Dev / Re: Armoured Commander Alpha 1: WW2 Tank Commander Roguelike
« on: March 24, 2015, 07:47:09 PM »2) I noticed in the Python source, you were using integer values for armor that abstract 'protection'. Personally, I'd be really interested in seeing a more sophisticated penetration modeling based on round type and velocity, angle of incidence and armor thickness and steel type. I expect that would be a lot of work to code, but the complexity can be hidden from the player, and the high fidelity of penetration ballistics could in-turn enable high fidelity damage modeling. But... I'm not coding this,.
This would be a mistake. It is a common misconception in WWII wargames for some odd reason - Steel Panthers had one derivative work that went down this path. Combat Mission also made a similar mistake. There are two problems, both showstoppers.
The first is fundamental game design, you have an abstract system that resolves a large number of approximations into a historically statistically realistic result. Now you decide to take one factor out of dozens, examine only one small cross section of that factor, and resolve it to oh lets say +/- 0.0001. Now every thing else is still integers, by how much have you reduced any error? If you had two factors, both integers and made one a float calculated to the nth degree, at best you've only cut the error by less than half. The error sources are additive, not multiplicative and in most game systems there are far more than two factors to consider. Worse, if you assume you've gotten all the errors out of one case, you'll actually screw up the formulas behind the abstraction.
The second is, well the Aberdeen proving ground tests are only averages for one thing, and that's about your best source of information on WWII ordinance. Of course those tests don't agree with the Russian tests, or the UK tests. Then consider logistics, some guns had bad rounds but the lots went to different theatres, mostly. Factor those in where necessary. Same goes for armor, some factories produced armor that was better than others, some supposedly face hardened wasn't, some was brittle. Now consider the various flaws in different vehicle designs. Consider maintenance. Oh and angle of incidence - highly dependent upon terrain, speed and suspension, relative altitude, even Combat Mission's vaunted 3d model was flawed in those regards. Factor in the effects of that tree branch that deflected the round by 0.01mm.
It sounds good, but it is flawed reasoning and can actually hurt the game.
I'm not suggesting a higher fidelity model makes for inherently better game play. But it certainly might do so, and while there's some variation (as you noted) between armor penetration testing results, that largely comes down to differences in how the measurements were taken. Obviously this can't be a 'true' simulation without modeling 3D space, vector, drag blah blah. A lot of the variation you suggest (bad rounds, precise AOI calculations, small variations in armor thickness or weak spots such as view ports) can be handled appropriately by RNG.
I'm only suggesting that being able to model more factors, and with greater fidelity, might in fact make for a more interesting game even if the overall outcomes are often similar.