Great find, you're definitely onto something here. I've done a bit of testing and I see the same discrepancy between the in-game ROF and theoretical values.
| Name | theoretical | in-game |
| Pershing | 6.750 | 7.000 |
| M8 Greyhound | 4.000 | 4.250 |
| Pack Howi AF | 9.750 | 10.125 |
| Brumm | 8.500 | 8.613 |
| Zis-3 | 5.863 | 6.245 |
| KV-8 45mm | 5.000 | 5.112 |
| HM-38 120mm | 8.238 | 8.542 |
| PM-41 82mm | 6.950 | 7.280 |
I've tried correcting the theoretical TBS by your method, but I always get 0.25 s more than in-game. Pretty strange, but I agree this can be pretty important and deserves a closer look!
The more you know... |
Vipper just wrote me that the cool down only applied for weapons with magazines, so tanks are irrelevant for this.
That's correct. However, if you want to include the Ostwind (or PzIII, Centaur or the AAHTs) in your simulation cooldown may still be important. Usually I use a "mean" shot delay for these according to the formula I posted earlier, but it's also possible to calculate this more accurately (e.g. 10 shots fired would be 7 x cooldown + 1 reload + 2 x cooldown).
|
I have to crrect myself on the original post, tanks have a reload frequency of 0 (from the couple I looked at at least). This would mean that cooldown and fire_aim are eliminated from the formula, but see below.
Are you sure about your formula?
I just modded a T34 (to all values = 1 and then I started changing them to 2 to see if and how often it is applied and affects the time between shots) and from what I see cooldown does not matter at all.
Now I'm not sure if there is a quirk with the attribute editor, but I went to the cooldown->duration and added 3 seconds for both min and max. In the game however it did nothing at all.
Or did I do something wrong?
From what I see for tanks it goes like this:
ready_aim_time: is applied only ONCE the first shot (probably for every new target)
wind_up and wind_down are applied once per cylce
fire_aim_time is applied once per cylce
reload_duration is applied once per cylce
HOWEVER I still get half a second I cannot explain. No idea where it comes from. With all the values above set to 1, my modded T34 should have 4 seconds between shots. But what I actually measured were 4,5 seconds. Regardless of which factor I subsequently changed, there was always half a second added to what I expected.
Actually I'm not 100% sure if the formula is correct, at least not on the part if it actually works correctly in-game. I used to assume the NearShots/Sec from the Weapon Report is accurate and never did any in-game testing to confirm.
Cooldown duration will have no effect if reload frequency is still set to 0 (e.g. no salvo is fired, just a single shot. Did you try setting this to >0 already?
Other than that I'm not sure where the mysterious 0.5 sec comes from. Maybe anyone else has an idea? |
A quick question on the ROF calculation:
In Vippers DPS guide, he shows a formula that was developed by Cruzz to calculate the rate of fire of a given unit. The formula goes like this:
(wind up+fire aim(range)+burst duration(range, moving)+wind down+cooldown(range, moving))*(1+reload frequency)
- cooldown(range, moving) - fire aim(range) + ready aim(range) + reload duration(range)
There are two parts. The first line is mltiplied with (1+reload frequency), afterwards there are some corrections by subtracting cooldown and fire aim time, apparently because they do not exist after the last burst.
My question concerns tank combat. Tanks have a reload frequency of 1. According to the formula above, wind up, wind down and burst duration would be factored in twice. But I find thia counter-intuitive.
What is the correct formula to calculate the time between shota for a tank?
Since I'm quite lazy, I usually use the reciprocal of the NearShots/Sec from the Weapon Report .csv file to get the total time between shots. This is usually enough for most cases, but if I need to factor in buffs to reload from abilities or vet I split this into a fixed 'delay' part and the actual 'reload' component (by simply subtracting the reload).
I've tried the formula above but got slightly off results from what I'd expect (i.e. 1 / NearShots/Sec). Removing the fire_aim and ready_aim from the formula you posted seems to work, though... at least it gives the correct delay between shots:
((windup + aim_time + burst_length + winddown + cooldown) * (1 + reload_frequency) - cooldown + reload) / (1 + reload_frequency)
|
While I’m not going to say you’re wrong, the model and tests don’t incorporate elevation or debris. The expectation of perfect tests is clearly unattainable, but I think it’s probable that tests can show the pinnacle of performance given they’re done in vacuum environments a lot of times.
You're of course right, and probably all these testing and modelling results should come with a disclaimer that they represent only part of the overall picture. There are simply too many additional variables (collision w/ terrain or obstacles, squad spacing and movement, etc...) involved to be accurately modeled without making things overly complicated.
Still, while absolute numbers such as the S2K or T2K presented above shouldn't necessarily be taken face value, I'm confident the proportions or relative trends these results outline are pretty robust. After all, things like elevation and collision hits affect all tanks more or less in the same way.
But only time and plugging a lot of in-game stats will tell how accurate all the predictions and in vacuum test really are. Until then I guess this is the best we have - or at least better than a bits of anecdotal evidence. |
This is exactly what we found in our testing which, admittedly, was much less rigorous/thorough than your own! It seemed to have close to zero impact on the average time to kill. The intention was to reduce the unavoidable bleed that heavies inflicted across the board while still maintaining their strong anti-infantry firepower. They still should and do kill infantry or support weapons when they stick around, but shouldn't be quite as punishing in the first shot out of the fog of war.
There seems to be a placebo effect where people have decided that heavies are bad now and so they see heavies performing badly.
I think what adds to this problem is that people seem to grossly underestimate the huge variation in possible outcomes, especially in Tank vs Inf fights, due to the several layers of RNG in effect here. This, paired with confirmation bias as you pointed out, inevitably leads to wrong conclusions being drawn - which is understandable since especially those very unlucky, highly improbable events tend to stick in mind. KT overshoots four times in a row? Tank is total garbage!
The problem is that, sadly, this kind of preformed opinion is very hard to change, even in light of overwhelming evidence to the contrary.
|
Yeah please try to find out if there's something that triggers it and report back. I never play vs AI so I don't know how the AI is really supposed to work or if their behaviour has changed since the last patch. When I launched a custom game vs 4x UKF AI (all levels) to see if they'd spam medics or not, they sent all their units to some spots and just idled there, but then again I wasn't doing anything. But I've run some test games just now and the AI seemed to be running around and attacking like it should.
Though not related to Brits, I remember people complained about the OKW AI not purchasing the T4 Panzer Authorization upgrade (or very late). As a result, it keeps spamming light vehicles instead of tanks until very late in the game, which of course makes it quite easy to counter. One of the modders back then suggested this could be easily fixed by increasing the 'priority threshold' to purchase the upgrade.
Not sure if this has been considered already, but I thought I'll throw it in here. |
You mean, on top of the
- scatter nerf
- vet 2 scatter nerf (Something different than the scatter nerf, since the bugged version of tiger was nerfed without taking into consideration that the values it meant to have, never saw gameplay
- cp increase to cp 12.
Yeah, literally just a small nerf, which doesnt effect its peformance at all.
(I am aware, that some changes had to be done; but this solution is not the way to go imo.
As Katitof already pointed out: There was no scatter nerf to the Tiger whatsoever. In fact it still has by far the lowest scatter of all heavies (with the exception of the KV2 which is superior below 22 m range), and that is before the vet2 scatter buff it used to have. Now you can argue that losing the vet2 buff does indeed impact the Tiger's AI-performance, but I doubt it needed it in the first place.
There was also no "bugged" version with regard to scatter that could have influenced balancing as you claim (please feel free to provide evidence to the contrary). The only difference between the 'live' pre-WBP Tiger and the one advertised in the respective patchnotes is the reduced near AOE (1 m instead of the intended 0.25 m), which had an almost negligible effect on its overall AI-performance.
As a matter of fact all the AOE-nerf in the WBP achieved is to reduce the effective OHK radius a bit further than what was already planned for the Tiger before, as you can see in the graph below (the area under the dotted line doesn't matter as almost all infantry models have 80 HP).
While this reduces the chances for one-hit wipes of individual models quite significantly, the impact on the average time to nuke a whole squad is tiny.
To put this into perspective, and since you've claimed before the PzIV now performs better than the Tiger in the AI department, here a comparison of how long it takes the PzIV, Pre-WBP and live Tiger to kill a 6-man squad in different formations on average (n = 5,000 tries).
The AI of the Tiger was and still is more than fine, despite the anecdotal evidence and conspiracy theories circling around here obviously.
|
|
though i don't share your opinion on the direction and intention of the balance changes at all, i do agree that unintended consequences, such as the AI essentially becoming defunct as you mentioned, should be looked at. many users here, myself included, probably don't care much about compstomps, but i understand that for a large part of the non-competitive community these issues are absolutely game breaking.
i doubt it would do much harm if a rollback as proposed was introduced, still the better solution would of course be to have the voices of this part of the playerbase not to be completely ignored. |
Stop this nonsense thread. Heavies are still more than good, all of them, so just stop.
/thread |
Relic Entertainment
cblanco ★ 
보드카 중대 
VonManteuffel 
Heartless Jäger 
