#numbers and stuff
1 messages ยท Page 1 of 1 (latest)
Your point of failure is "present or not" so you don't care about the number of books, so it is exactly 3x.
for a three-item box, .35% works out to almost 1 box in 100 having the desired item
it's a little less than 1 in 100, but that's close
Where did we get the idea that the box has three items?
someone said it, i picked it out and used it as an arbitrary example
Someone said it
i don't know of any loot containers that spawn with just a single item
the math gets even more worse when you consider that the number of items is itself usually random (to varying degrees for different containers)
I think the issue is that Graphko calculate with Decimals
While I'm calculating with percentages and divided by 100 per
thats why Im getting alot lower
So what's you calculation like?
You agree then that when you assume there's 1 item in the box the chance of that item being present is .00615?
I don't =].
You're essentially calculating with decimals still.
yeah
just so i know i'm on the right page... we're talking about finding a specific book in a savage country crate, right?
Yes
yes Impyre
I thought the last operator was .35?
It's not, when you take it being a decimal into account.
0,006125 is 0,6125%.
๐
How come?
1 in 28 books is a chance of 3.57% and that's factored in as 0.035
Okay, I cannot a test this for sure
First is a Roll of yes or no
50%, you get the Loot or Not
That's not 100% sure either.
the second is either, one item only, or multiple items with their own %
again, I haven't found much Savage Country Crates, the max I found in a single crate is 3 Items
That is the max indeed.
okay, so Graphko is correct
you add up all the chances and take the specific one
So it is about half the efficiency
0.35% is not something to write home about
Depending on if the same method is used with force_prob, the probability could also be 0,20%.
Force Prob is already set to true
Yes
We can do math or we can debate whether the information it's based on accurate. We can't do both at the same time. I'm only working with numbers provided. ^^
if false, yes, 0.20%
Hmm?
Sorry mac, its both of them
Me and Graphko was looking into the code
And then trying to calculate the chances
It is my error that I didn't take one specific thing into account
But really, 0.35%
The groupSavageCountryCrate has a count=all, which normally means you get all of the items in the crate.
Whats the Electronic Crate name?
mopowerCrate
Well the important thing is that the math was correct. It's easier to fix incorrect numbers in correct equations than fixing the equation.
I'm still voting for 0,20%.
On the practical side, I vote that drop rates under 1% can be filed under "oh lord" and forgotten about. ^^
Mo Power Crate have 2 Different Loot Group that contains the Tech Junkie Book
@modern ingot here, let me give the fact
I'm starting to worry about my 6/7 needle and thread now. Do you have a number at which loot stage the books stop dropping outside the crate?
60
groupMoPower02 have 100% To activate, the child loot group have 50% chance to give tech junkies
groupMoPower03 follows the same rule
oof. Time to take off the glasses and raid T1 suburbs. I'm 53 right now.
What are the odds of getting one tech Junkie book of choice?
So 2 Loot Groups with 50% Chance to get a book
(Since the parent loot group is 100%, we can throw that number)
What are the odds of getting one specific Tech Junkie book? 1/7
i calculate around a 3% chance of getting said book in savage country crate
Whats the Equation?
though, i'd like to offer that I know that number's actually a bit low because i used some simplifications that add some error
@modern ingot So 2 Loot Groups with 50% Chance to get a book. What are the odds of getting one specific Tech Junkie book? 1/7
here's the formula:
P(not the book)=.5+.5*.65+.5*.35*.96428^5
p(book wanted)=1-p(not the book)
Huh
2 loot groups out of how many?
It doesn't matter since the parent loot group is 100% to trigger
Yeah but how many in that parent group?
so 2 Loot Group, each having 50% chance to drop a book. and 1/7 is a book of choice
4
I don't quite understand where these numbers come from.
Same Graphko
.5 is approximately the chance of rolling the other crate type
my instinct told me its 50% x 50%. then subtract 100% with that number
So the first multiplier is 1, because always, second is .5 because you need to hit 2 out 4 groups, last is .167
.65 is the chance (with lootstage 0 and no loot bonuses) of not rolling the books
.96428 is the chance the book chosen is one of the other 27
it's ^5 because there are an average of 5 chances to roll said book
Its .194 for me
(that last one's the simplification btw)
Wait, let me ask Mac this question and I'll get back to you imp
@modern ingot Parent Loot group have 4 Different Loot Groups, the 2 Is Guaranteed, the one we are using
The Child have 50% Chance to get a book, 2 Child Loot Groups
.167 for the Specific Book
So its (1 - (50% x 50%)) x .167
Am I correct?
Sorry, we're using child to designate different stages in the tree multiple times. When you're not looking at the underlying data that's hard to follow.
Okay sec
First, ignore all static events.
Second, how many end points to you have, at each stage in the tree, count the events possible leading to an end point vs the total number of choices.
There you have all the loot table math you will ever need.
It's also how it works when you have to explain it without redoing 7th grade stochastics.
not feasible in the given context
unless you wanna be here until you're 200
what if the crate rolls 4 items? 5?... what if the desired item is the first item rolled? the fifth?... what if you get two of them? is that even possible or is there code preventing it? do you have source code?
without knowing precisely how the code handles loot rolls, any answer is going to be a best-guess answer, and the math for finding a good guess isn't all that difficult
here's a question
according to prob templates
do all items on all templates get weighted evenly?
as long as they meet the threshold of the roll?
the book group is on the MedLow template which is .35, but the sewing resource is on Low, which is .2
Yes, those are flat numbers.
so given a roll of .37, do the sewing items have the same chance as the books?
or is it weighted by .2/.35?
A roll of .37?
yes
What does that mean?
it means that every random event in the game is worded in the form of 0-1 probability
and that random number generators usually roll a number between 0-1
the loot prob templates are obviously used to delineate percentages
ye ye
the easy way of doing that
Yes
so, .37 is high enough that both .2 and .35 both qualify
Wrong.
The values are not straight converted to %.
whether they're on a curve or not is besides the point
the point is, you cannot exclude lower groups
because then the number doesn't reflect "rarity" as such
my question is, if a roll includes a rare group, and a common one, how do the probabilities get distributed?
i'd assume it'd be weighted to preserve rarity, but i don't know that for a fact
Official 7 Days to Die forum links:
Home Page
Developer's Diary
Suggestions
Bug Reports
Support
Mods
We figured it out.
Odds to get a sewing book: Really bad
Odds to get a tech junkie book: Really good.
Those calculations are accurate
i also wanna know if one loot selection that contains multiple items takes up multiple rolls, i don't think it does but i'm not sure
I don't know whether we need to know.
Whether we can just take a step higher from the fundamental level of the random number generation.
I think we can assume that the basis of the probabilities is made so that the calculations based on the numbers we have reflect the truth.
As we really have no way of seeing beyond that.
@vague sierra One more question
The Quest Reward says "Prob = 4"
what does that mean?
No
The probabilities are summed up, thus the actual percentage is still between 0 and 1.
Okay, it still wont make a difference
Since someone suggest to do Low Tier Quest for Perk Books
But it still under perkBooks loot group, which uses the old Probt0
so I still wont get ranger books, sad
I'll do some testing.
Okay so.
force_prob is the flat % the template has.
So with a probability of 1 and force_prob the item is acquired always, no matter the other probabilities.
Thus I revert to the chance of ~0,35%.
@mellow swan Here again.
okay so
tech 2, the purple one
it says 0.5, So a 0.5 Chance of getting a Tier 6 Item
Then there is Loot Stage that starts at 68 to 159
First question, what happens if you go over that loot stage
No? Then what are the numbers for?
That is the loot stage, since there is so many levels on that
If you look at the Base Model, it has the same format.
There Gamestage is on the top.
so its a fluke, then, this is the important bit
Here the quest tier is on the top.
Thats not the quest tier, its the quest reward value
Which is equal to the quest tier.
Yes and no
How come?
My bad
Okay, this is the important bit
Notice that the arrangement of the Colored Numbers
Yes
Notice that Tech 2 and Tech 3, it can give Tier 3 Items
But from my testing, it will never give Tier 3 Items
minimum is 4, at my testing
There should be a small chance of getting Q3 items as well.
Well I never encountered that
Looking here, you can only get Q3 items from T5 quests.
Tech 2 are always Q4 or higher in tier 4 and 5 quests.
Oh, hold on.
Yeah
I did use the templates that are used in the game as well.
have you ever get T4 or T3 Items from T5 Quest?
I haven't done T5 quests in a while.
see, there is a bit of contradicting values I find
The values in the xmls take priority.
Or the values used in templates that are used.
There is Quest.XML and Loot.XML
Yes
Both have contradicting values
How come?
Well I wanted to go over the passives, but let me explain this
First, The Quest.XML, There is Tier 6 Quest, but Im not sure if its used or not, for the sake of consistency, lets use Tier 5 Quest, it have Quest Reward Value of 5, Which means "the reward will be this tier but can be randomly 1 tier lower"
Not quite.
The value="5" is like lootstage when the game checks which reward you get.
Then it goes to loot.xml and checks both probability templates and quality templates.
In the quality templates the value 5 is the same as level 5.
So based on that you get certain possibilities to different qualiities.
okay, that makes sense what doesnt is
@vague sierra I found a Ranger Book at Tier 4 Quest
Which technically shouldn't happened
Which quest type?
Clear
There is perkbooks in the quest rewards so it is possible.
but as we established back then
perkBooks use the ProbT0 to T2
and it scales based on your loot stage
And you had lootstage higher than 60?
thats why I think it contradicts
But it doesn't?
Well, idk
It doesn't use the lootstage, but the value given by the quest.
so, what does the value brings us?
Also, I find it weird that Im getting 15000 xp from the quest but 2448 (20% Extra from 2040) for the duke
That's just a balancing thing.
Balancing thing?
wdym?
wdym?
I mean as in, the Tier 4 Quest stated I should get 7500 Xp
Not 15000xp
It's not a game mechanic, just values that can be balanced.
Yeah, what alters that values?
That's defined in the quests.xml.
im lost
Tier 4 clear quest rewards, for example:
<reward type="Exp" value="7500"/>
<reward type="Item" id="casinoCoin" value="2040"/>
<reward type="LootItem" id="groupQuestWeapons" ischosen="true" isfixed="true" value="4"/>
<reward type="LootItem" id="groupQuestArmor,groupQuestMedical" ischosen="true" isfixed="true" value="4"/>
<reward type="LootItem" id="groupQuestAmmo" ischosen="true" value="4"/>
<reward type="LootItem" id="groupQuestTools" ischosen="true" value="4"/>
<reward type="LootItem" id="groupQuestSchematics" ischosen="true" value="4"/>
<reward type="LootItem" id="groupQuestMods" ischosen="true" value="4"/>
yeah yeah, but why Im getting 15000 xp?
Does the Server XP gain have anything to do with it?
If xp is at 200%, yes.
groupQuestSchematics.
mhm, and how come it gives ranger books?
It has perkBooks in it.
okay, lets jump ahead and say, what the <reward type="LootItem" id="groupQuestSchematics" ischosen="true" value="4"/> does
Chooses an item from that group with "lootstage" of 4.
Yes, seems to be the case.
Since it never go past Lootstage of 5, and there is nothing defining the probability of groupQuestSchematics
so it takes that Quest Loot Stage and Put it as a General Loot Stage
I assume so.
so Both Quest Loot Stage and Normal Loot Stage uses the same code
however, some items have Quality and Tier advancements
Farming T1 Quest it is
@vague sierra One more question unrelated to this quest thing. How does Timed Charge damage calculated?
it supposed to give 110 x 38 damage to Safes and Doors
But its inconsistent
Im assuming the .65 Blast Radius have something to do with it?
At the centre of the explosion it's that, but I think it scales linearly to the range.
So yes, that short range does have something to do with it.
but it sticks to the safes and doors
Yes
idk
@mellow swan We have this one.
okay so
private float GetDamageFraction(float _damage) { return _damage / zombie.GetMaxHealth() ;}
GetDamageOverlimit = Max(0, GetDamageFraction(bodyDamage.RightLowerLeg + _damage)- ec.LowerLegDismemberThreshold)
baseChance = ec.LowerLegDismemberBaseChance + (GetDamageOverlimit * ec.LowerLegDismemberBonusChance);
totalChance = canDismember ? (baseChance + _weaponBase + (_damage * _weaponBonus)) : 0f;
This is the pseudocode for how Dismember Chance work
I dont understand half of these, if you can make sense of them that would be great
All I can say is that the damage also corelates to the dismember chance
but if the dismember chance is 100%, shouldn't low damage would work 100% of the time?
Yes
Then why its not working.. hmm
Because it doesn't seem to be that easily calculated.
It only works when I deal about 40-50% damage in a single hit towards an enemy limb
okay, could you make sense of the formula I just sent
From the top
private float GetDamageFraction(float _damage) { return _damage / zombie.GetMaxHealth() ;}
Give me some time
will do
Somewhat difficult to say exactly what's happening not knowing what all of those values refer to.
Exactly, its just a random gibberish for me
How I understood it was it first checks the % damage you do to a limb, then remove a certain treshold value from it, multiply with bonus chance and add the base chance, add weapon's base chance and any bonus multiplied with the damage done.
but then if the base chance is 100% and no Subtraction or Division
Then even low damage as 1 Should Dismember the enemy?
Yes
but it doesn't
It especially rare for Irradiated Zombies to get dismember cause they have another 0.4 Dismember Multiplier
at that point, this is why Clubs reigns supreme, cause the Clubs doesn't do anything about Dismember, it just brute force with alot of damage multiplier
But base chance depends on the damage done.
But, it does seem that there is another variable at play
I just changed the entityclasses.xml on the playerMale entity
<passive_effect name="DismemberChance" operation="base_set" value=".95"><!--Attribute "value" replaced by: "Dismember Test"--></passive_effect>
So All I did is put one point on the skill tree for specified weapon and it goes to 100%
Yea
but it doesn't one shots enemy heads
Tried with the Level 1 Hunting Knife
Light attack on the head did nothing, 18 Damage
3 Zombies tested, no dismember
Heavy Attack deal 24 x 2 (On Head) so thats 48 damage, which is triple, 3 zombies, everything dies from Dismember heads
Hmm, interesting.
my biggest fear is that the code is hardcoded
Which means its fruitless to change the dismember chance, as all tied to damage
@vague sierra do you have a decompiler for dll files?
No
Rip, I think we may never know how this works
@vague sierra where do you think the Dismember Calculation located?
No idea.
