Do we know how bloom works?

[breeeze 446]

I'm kinda curious about this one, I know the accuracy is taken as a baseline and in some way modified by the bloom values depending on speed/traverse. But do we know how this is calculated? Is there a way to calculate actual bloom depending on speed and how it stacks with tank & turret traverse?

[It's_Matra 866]

Well you have a max bloom* and a min bloom*

This is modified by

• turret traverse*
• vehicle traverse (turning)*
• vehicle movement (backwards/forwards)*
• aiming time* - the base value for the time taken to go from max to min if the above = 0
• Crew skills and perks**
• Equipment

*differs for each tank based on base values and soft stats

**also a hard variable to deal with as crew skill (0-100% affects it, well commander and gunner anyway with perks on top of that; smooth driving to help negate when driving back/forth, snap shot for turret rotation etc.)

I suppose you could calculate the bloom if you had an algorithm that copes with all of this (like WG use) but you'd have to know/create that first!

DON'T FORGET on top of that you also have the RNG factor when actually firing too which you cannot account for, which does give some validity to the 'accuracy is a meaningless stat' argument.

[EndlessAgony 129]

https://www.wired.com/2011/03/flower-bloom-physics/

On a more serious note, some time ago, I had a discussion with some people from RU server and they told me it works in a pretty retarded way, basically it's based on full accuracy and the bloom works like a modifier/multiplier of sorts. You multiply the base acc by the bloom and add the value to the base accuraccy and get the actual current  size of aiming circle.

For example, if your full speed movement bloom multiplier is 1 and your base accuracy is 0.25, then your acc on the move at full speed should be 0.5, if you are also turning your turret for 1 bloom, they add, so your bloom multiplier/modifier is 2, and with 0.25 base acc, that results in 0.75 acc.

I can't, however, back this up - it's just a theory I talked about with some random russians. But it seems quite possible it works this way.

[Veo 67]

I'm thinking the bloom is completely dependent on the velocity of your tank, and other components of course as well. You can tell it's not acceleration because there's 3 autopilot speeds to choose, and all 3 have different blooms. Of course if you could turn your hull at different speeds, that would probably act the same way, but the angular velocity of the hull is constant, not accelerating. The turret is different, I don't think there's acceleration on the turret, but if your careful enough with your mouse, you could get any angular velocity you want. (As in kv2 turning turret slow when aiming conserves having to aim in again, or moving mouse slowly when playing arty will conserve the bloom)

The bloom would probably be a multiplier function of these instantaneous velocities by the hidden dispersion variables that you can find on tanks.gg and such. I dunno, but from my experience, thinking of it this way helps me play long aim time tanks better.

[breeeze 446]

@EndlessAgony's hypothesis seems very interesting. I've been playing around with it and the results seem to be (from what I've seen so far which isn't very much) quite plausible when just having a single modifier e.g. moving acting on the total bloom at once, but then just adding the other ones on top seems to result in way more bloom than is actually seen ingame. Interesting stuff, guess I'll have to take some actual data.

[Wanderjar 298]

On 3/9/2017 at 3:11 PM, breeeze said:

I'm kinda curious about this one, I know the accuracy is taken as a baseline and in some way modified by the bloom values depending on speed/traverse. But do we know how this is calculated? Is there a way to calculate actual bloom depending on speed and how it stacks with tank & turret traverse?

yes we do:

Quote

Quote

ComputedDispersion = GunAccuracy * SQRT (1 + DispMovingTraverse^2 + DispHullTraverse^2 + DispTurretTraverse^2)

where:

GunAccuracy = ReportedGunDispersion

DispMovingTraverse = DispersionMoving*HullSpeed (kmh)

DispHullTraverse = DispersionHull*HullTraverseSpeed (Deg/s independent of turret movement)

DispTurretTraverse = DispersionTurret*TurretTraverseSpeed (deg/s independent of hull movement)

Then Smooth Ride seems to be better than Snap Shot in many cases and the benefits of both are minor at best and misleading/useless at worst.

now with formatting for easy viewing!
disp = Gun accy*sqrt(1+forward moving^2 + hull turning^2 + turret turn^2)
                   ^                        ^                                ^                         ^
              base value          SR skill                        nada                Snap skill    <-- skills that help. BIA helps all.

and the files you can verify it in is: 
res/scripts/avatar.pyc
it's unencrypted but you do need a tool reader to view it. Also the lang is python if its not obvious.

File path as relayed to me from RichardNixon and personally verified.

My Blog article with more info if you would like further info (and the source of this quote) and some spreadsheets to view how it shakes out

[breeeze 446]

Nice stuff, thanks!

[DogSqueezer 0]

informative thanks