Tree of Savior Forum

About drop system in TOS

General Discussion
#1

Hello guys!

So my question here is about drop system in TOS
My guess is that it is the pattern used in another games (basic binomial distribution) which makes much sense if you pay attention while you’re playing, someone get a rare item killing a some monsters, and others killing 8.000
I know that binomial distribution is used in another games, but I see some guys saying here that exist some hours and situation of the days that the drop rate is easier and they are creating entire theories in his minds that exist mobs with the drop marked on them, and they not believe that this is basic binomial distribuition
This doesn’t have a decent logic for me since I’m a programmer and already made RO servers before, but I still need to discuss this with the community to clear the doubts of this guys

PS: I sent a ticket and obviously the IMC guys can’t confirm this information

#2

When the mob spawns, their loot are already predetermined.
This is why Oracles can see what the mob is carrying.

5 Likes
#3

IMC will never-ever comment on this.

There are several things without a decent logic behind them. See net code (TCP).

That said, the incremental loot table theories seem VERY believable. At least for characteristic drops (everything but golden anvil etc)…

But in the end there is no way to 100% confirm until we get either a response (won’t happen) or a look into the source code (leaks are unlikely).

2 Likes
#4

@tikzfuu this could make sense for common monsters, but how about the cubes?

#5

@fira_wolf42 what you mean with the incremental loot table?

But in the end there is no way to 100% confirm until we get either a response (won’t happen) or a look into the source code (leaks are unlikely).

Yep, I’m thinking about this too, that’s how people confirmed the drop system in RO and other games

#6

The drop is generated to any monster based on its loot table, the chances to get is what change, it’s just a guess, but i think it would go from 1%-100% (or maybe 75% max) chance to drop something at random times.

The rate it is probably another variable, now as for how the calculation goes is as @fira_wolf42 said.

#7

Maybe when rolling cubes, the loots are sorted on different numbers at first then the roll system starts and chooses a number, if the number doesn’t have anything special you get a talt instead.

Example:
1 - Beetleback Recipe
2
3
4
5 - Arde Dagger
6 - Boss Card
7
8
9
10

When you open the cube, the system sorts the items randomly, then starts choosing a number randomly.

If you get a jackpot roll, you get the item.
If not, you get talts.

Also, maybe the items are already sorted and the system just rolls a random number and gives you its result.

#8

yep there are many ways to sort / rand drops
But that thing about oracle is the only stuff that can give us some certain about a part of TOS drop system, about mob drops it could be predetermined when the monster spawn because he can check if monster has an item, it sounds ridiculous to me but it’s an evidence as @tikzfuu said

In the end, I think there’s no way to prove nothing and we can just observe and play hahaha XD

#9

I actually think the cube system works so that it checks every single drop in order, if a drop succeeds, it drops, if not, check next item and if all fails you get a talt.

I also think that it always checks for a card first at 10% chance. Beetleback has a 10% chance itself, but because card is checked first it’s only 10% of the remaining 90% so 9%. And Arde is checked after that but only 5% based on the remaining 81%, so it’s more like 4%.

At least this fits better to the average drop results (noticably more people that need to open more than 20 cubes per arde).

And yes, drops are determined on spawn, so it can’t be kill related. Either RNG at spawn or more like “Every 1000th monster that spawns has this drop”.

#10

So this way sounds really unfair, you haven’t any chance of drop per monster kill D:
How you know this?

#11

No it is not like this, i can tell that, since i got once 3x glass brac while questing in Downtown while my friend only got 1. And we didn’t killed more than 1000 monsters and where the only ones atm at that map if memory don’t trick me now.
Anoher example is, last week while leveling my Sadhu me and the very same friend where at Forest of Prayer, she was helping me on quests, and while we where killing Stoulet Archers i got 3 thick chains, while she got none, and that map is filled with bots.

I got one or 2 more later on Grand Corridor while she got none, and it was 4 people questing there, me, she and 2 more players.

As for the cubes, it would be like @tikzfuu said, of course more calculations are involved.

1 Like
#12

Wouldn’t RNG at spawn be more taxing to the server? If each monster needs to roll a drop rate%* 10000* Math.rand() on spawn then match it to the corresponding monster loot table it will take up more computing power when the server refreshes the monsters every now and then?

Personally I think that an incremental drop table is more likely. Like a table with predetermined drops on each row with a counter to increment the pointer everytime a monster pertaining to that table is killed. The server can then refresh the table once it reaches the end of the rows. This saves more computing power since lesser rand() calculations are needed?

#13

Please correct me if I’m wrong - wouldn’t that qualify as DPK then?

#14

I really need to refresh my knowledge on acronyms. What is a DPK?

#15

Drop Per Kill e.g. after a certain number of kills/spawns per map/channel a monster type drops X item.

#16

It would be more taxing if it was running while the mob is alive, if it only does when they spawn and keep on the waiting until someone go there and kill them, i think it is less taxing.

It is the concept i think.

#17

You know this because the Oracle class has a skill to see the drops which a monster drops when killed. This is proof that the drops are already determined before the monster is killed.

#18

computing power (CPU cycles) are actually less demanding on the system than hard drive I/O. Storing everything in databases and reading/writing from the hard drive slows the system down more than calculating.

@Nirimetus

It would actually calculate on monster death, which is the same as doing it on spawn.

#19

That said, in the current server state, not optimized, and if the host is working with SSD it is less problematic, but SSD on a host that need lots of data writing is not worth, unless there is some sort of technology i am not aware about.

Indeed.

#20

Its not, the drop may be set on kill AND on oracle sight instead of on spawn.
+[quote=“Yuuseki, post:18, topic:305637”]
computing power (CPU cycles) are actually less demanding on the system than hard drive I/O. Storing everything in databases and reading/writing from the hard drive slows the system down more than calculating.
[/quote]

makes more sense than predetermining drops on spawn.
Oracle chance doesnt work on shining mobs though, they have a fixed loottable as someone found out with change skill.

Oh, and there is this guy here on the boards who is sure to have proven DKP system. He had a friend of his to kill a mob with a 100% drop chance on an item cause he kill counted until then -> that only works with droptable on kill (which is also an explanation for delayed droppings, otherwise there would be needed a function to create drop delay)