Pavonis Interactive Forums

As the title says, i get confused when i decided to fire a free aim missile. What does the "X.X scatter" probability mean? (Example, 4.1 scatter)

The only thing i know is... that tthe lesser the number is, there is a HIGH chance of hitting the right spot, but if there is a number like... 3.9, it means that it's more likely to miss and shoot everywhere.

I get that part very well, but i dunno how that X.X number work. I tried getting into Ufopaedia but the website is broken xS

It's the number of tiles (Only one tile can be occupied by a unit so that is the size of a tile) the rocket will scatter by on average(?). There is a 65% for it to be this number or less and a 98% (or similar i think) chance for it to be twice this number or less.

Hi!

From what I've learned from ufopaedia.org website, it follows a gaussian distribution (or normal distribution).
https://en.wikipedia.org/wiki/Normal_distribution

It indicates the deviation, so, You can guess the how accurate the shot will be. And yes, the smaller number means less scatter.
To have an image what (should be) going on, check below page:
https://en.wikipedia.org/wiki/Normal_di ... l_Rule.PNG

So, around 2/3rd of shots will be within 1 deviation from the aimed center, further 25-30% will be between 1 and 2 deviation some 4-5% will be between 2-3 deviation from aimed centrum, anything above that is negligable. I guess it is even 'boxed' in the program (no scatter above 3 deviation), in theory You could have 10 deviation scatter with unthinkable low probabilty. (should be 1,5×10^-22, so 0.000...000015, 21 zero after 0, that is a probabilty of flipping 72-73 times the same side of a coin in a row, or winning the Euro-lottery 3 times in row with 300 tickets).
Though I do not know how it determines scatter way (angle). And rockets do not scatter in height, only in 2 dimension (left/right/behind/before center).

So what I do is generate two independent, gaussian distributed variables.

Vanilla EU/EW only scattered the rockets in "cross range". That is, it back-computed from the target point to a range-azimuth (polar) format, then randomized the azimuth and forward-computed back to a Cartesian x-y hit point.

In Long War I adjust both range and cross-range (using similar transformations). Also the random distribution is Gaussian (normal) in each dimension. I probably did over-indulge my math-ey algorithm developer side a bit on this, since this mechanic has been difficult to explain to players since its introduction.

It does seem odd just how inaccurate missiles can be, given the "I hit the gnat in his left eye." accuracy of grenades. But that's just my taste. I can follow the math although I suspect I would have generated a single distance from aim point with a normal distribution and then a direction of drift with a uniform distribution.

It is a pity the new engine doesn't support the old games ray-tracing method of generating misses. That lead to some hilarious "Miss the alien, go through the broken window, down the hall, miss the muton, out the destroyed wall, across the map, hit commander in the left eye." misses.

We'd have make grenades less accurate if we could, but there are significant technical issues with grenades.

Grenades work using the pathing system, to trying to randomize them is essentially the same as trying to randomize a soldier movement. In both cases a path is computed based on cursor position. When the ability ("throw grenade" or "move") the pawn moves along the predefined path.

However, in the bigger picture, the discrepancy between grenades and rockets was definitely a problem. The original motivation for scatter on rockets was to try and keep aim a relevant stat on Rocketeers, which sort of worked, but it didn't work for Engineers with grenades.

It worked to differentiate the classes really. I value aim on my rocketeers to the same extent as I do for snipers. Engineers are where I will stick soldiers with crap aim. It actually highly useful to have a class which isn't aim dependent. (Other than medics and suppression gunners.)

In any case it is hilarious when you attempt to fire a rocket which gets a relatively high scatter value, the soldier will literally spin in a completely unrelated direction and blast the scenery giving the feedback of "rocket went off course" - YOU DON'T SAY!?

And I love the new system. Great idea and great job, Amineri. But, well, math is my great love and statistics is a piece of cake to me, so that's probably why As soon as I saw "there is 68%..." I got that 'oh, that's Gaussian distribution'.