Definitive projectile accuracy formula?

[Rayguncourtesan]

Has anyone figured this out yet? While we have the accuracy modifiers for projectiles for both the projectile itself and the unit shooting but they're essentially n-values. Absent a baseline for measuring what '10' in accuracy translates to its essentially a meaningless number and might as well read 'Booblefruit'

So is there a definitive or 'close enough to it-working' formula for whether a given shot connects?

[Larynja]

It's a percent value. So an accuracy of 10 from the unit and a marksmanship bonus of 10 from the projectile gives a 20% chance per entity shot to exactly hit. If the projectile misses, it lands somewhere within its calibration area, so it theoretically can still hit. A unit gains 1,3 * n^0,3 accuracy per level up btw.

[Goumindong]

It's a percent value. So an accuracy of 10 from the unit and a marksmanship bonus of 10 from the projectile gives a 20% chance per entity shot to exactly hit. If the projectile misses, it lands somewhere within its calibration area, so it theoretically can still hit. A unit gains 1,3 * n^0,3 accuracy per level up btw.

What is n? I can see the 1.3 and .3 in the DB but no structure for what n would be

[Larynja]

What is n? I can see the 1.3 and .3 in the DB but no structure for what n would be

n is the base accuracy of the unit.

[Rayguncourtesan]

Are we sure those're the only numbers used in the calculation? For example with melee there's a base probability - now admittedly you're adding AND subtracting there (we assume...it'd be...odd, if it wasn't working that way) so you need something to avoid wildly disparate input values or very low probabilities.

[Rhavanna]

Very interesting. Do we have any idea what melee accuracy does also?

[Rayguncourtesan]

Melee Attack? Pretty sure its a straight +% chance to hit, and counted together with the target's melee defense.

So the formula goes: base (40% by vanilla kv_rules) + melee attack - melee defense = % chance to hit, with anything below 10% or over 90% being simply counted as 10 or 90 respectively

It could be doing things differently (multiple rolls etc) but it'd be computationally expensive and a lot more complex than it needs to be. So my money's on simplicity for a game that already hammers the CPU quite hard.

[Rayguncourtesan]

Is it possible for the missed shot to still hit the same target - lets say for argument's sake that the calibration area is 0? Does this mean a miss is always a true miss, or does this mean the shot deviates absolutely not at all and hits the original target?