[April 12] Initial Development Manifesto Feedback

I have a proposition for the loot allocation options.

Have the player select their prefered loot allocation mode somewhere in the options menu.
Then, have the game automatically select the loot allocation mode according to the preferences of all the party memebers.

For example, in a party of 6, 4 chose "FFA", 2 chose "longer", the game automatically sets the looting to FFA. (I'll let you guys decide what would be chosen automatically in situations where it's tied, like 3 vs 3)


I'm just trying to streamline the system, so that people dont have to choose (and argue over) which loot system to choose, each time they form a party. With my system, everybody has already voted (by chosing their prefered system in the options menu) and the winning loot mode is automatically chosen as you form the party.

With respect to the desync problem, have you considered the idea of making directional velocity adjustments rather than discontinuous rubberbanding to try to correct smaller variations in client/server position? i.e. very slightly adjust the (pre-collision) client-side velocity of moving entities in the opposite direction of the detected error. This would hopefully bring small errors back in line as things move around without requiring sharp discontinuities, and would be barely perceptible if the velocity adjustment were not too large.

There's a free parameter here (how much client-side velocity adjustment to perform) which trades effectiveness of position correction vs. how uniform the movement looks. I have a feeling that you could get away with quite a bit of correction here without the character looking too drunk, especially given that characters turn instantly, so it won't look like they're crab-walking, but it may seem like they pathfind a bit strangely. Say, a correction of 10-25% of the normal movement velocity wouldn't cause the direction to vary too far, and could pull small variations back in line before they have a chance to turn into large ones when the server side character hits an obstacle which the client side one doesn't.

Another thing which seems like it would be worth doing on the large scale is to limit the extent to which pathfinding can disagree. Perhaps consider making the sorts of actions which cause long paths to be computed to synchronously agree on an overall path before switching into a pathfinding mode, rather than "dumb" movement in the straight-line direction of the clicked-on object.

That is, do major pathfinding *only* on the server-side, and send the long paths to the client.

The only case where a player should really be able to notice the synchronisation in this case would be the small delay before pathfinding kicks in when you're right up against an obstacle and you try to click something on the opposite side of it (and your character runs toward the obstacle for the lag time before starting to head along the server-computed path).

It would hopefully mean that the client never heads the wrong way around a large obstacle, which seems to frequently be the cause of bad desync.

Another thing which might be nice to have is the ability to press a key to turn on a "ghost mode" which shows, say, outlines or circles in the position that the server thinks things are. I'm sure I could provide better feedback as a beta tester on cases where desync becomes an issue if I could more easily visualise that kind of information.

One thing that I *suspect* is happening, but I can't really say for certain, is in cases where you're firing a ranged skill, like a bow, through a small constriction like a doorway trying to hit a desynched monster which on the client appears to be in line-of-sight. On the server, I suspect the character will sometimes move through the doorway into the room to obtain line-of-sight and be able to attack (and if you're lucky, you will rubberband forward into the room at this point, rather than simply instadying to the pack of lightning skeletons you just derped into). I know that Kripparrian swears by using hold position constantly to limit the issues with desync for this kind of reason, and since I started using it more consistently, I've had fewer problems of this sort as well.

If this indeed is the case as I suspect, something perhaps needs to be done to treat it a bit specially. One option, which might actually be really nice even if desync were not a problem, is for the pathfinding algorithm used to get in range for a ranged attack to compute a path not to the mob being attacked but to the region in which there is a line of sight to the target. This might mean that the character moves in a direction perpendicular to the target when trying to shoot through a doorway, which is a smarter thing to do anyway.

It may be somewhat more computationally expensive than whatever you're presently doing to find a path -- A* can treat multiple goals like this pretty straightforwardly though, once you know which set of points are visible by the monster being targetted and in range for the attack.

A simpler approach would be just to refuse to pathfind as far in the general case of trying to make an approach for a ranged attack (i.e. if a few steps don't cut it, just stop), or treat constrictions like doorways as obstacles for this particular sort of pathfinding with a ranged skill.

The only significant issue I dislike in POE is the party loot system--I understand many people prefer more item tension/free for all looting, but I prefer to concentrate on killing and take my time and savor my item pick ups. So, I am very happy the Devs are considering/planning to provide loot options for party play--this was exactly the solution I was hoping for--everybody can play the way they want. Perfect. And, thanks for posting this manifesto, as well.

Great read and I am very glad that you're on that topic (I know you were but confirmation feels good).
I find this topic really fascinating and look forward for more news about it.

"

artemis42 wrote:

This manifesto worries me deeply. These points are particularly worrying:

1. adding more AoE attacks to monsters in order to challenge groups. This is BAD design. Not only will this exacerbate melee builds' survivability issues (I'll explain why shortly), but it further removes from the player the power to actively avoid damage. I'm reminded of Torchlight 2 which, even solo, has so much magic and shit flying all over the screen that one quickly gives up trying to dodge any of it and instead soaks it all up with potions (boring).

What you need to do is add dangerous attacks that have the potential to hit multiple players (I'm thinking of Piety's lightning ball), but can be actively avoided. You do NOT want to, say, have trash mobs casting Fireballs with a large AoE, or Lightning Strike because 1) melee can get hit by the splash of a spell aimed at a ranged player, 2) the game's difficulty becomes even more married to one's passive defenses, amplifying balance problems and forcing players down the same, most effective path, and 3) higher chance of multiple AoE spells hitting the same player with their splash damage (sure, you can minimize this by spreading out, but what about tight maps such as Spider Lair?), which leads us into a particular case of 2), that of path of life nodes.

Eh? Melee getting hit by the splash targetting a ranged player? That's either (a) too large of a splash or (b) bad player positioning (players clumped, ranged in on melee, etc).

The mechanics of this game heavily encourage and favor player skill, particularly in regards to positioning. For example, as readily evident if you do race parties, watch players who successfully spread out around, say, Brutus, to mitigate his ground slam damage. Try not being a face masher and actually pay attention to your positioning and I'm sure you'll not worry about getting hit with AoEs, even if melee heavy parties. Just because you can stack on top of other players doesn't mean you should.

I really enjoyed reading all of those topics. Thank you! It sounds like a lot of really cool stuff on the horizon. I also feel like I could b.s. my way through a lecture on client/server synchronization now. :P (something I formerly knew nothing about)

PoE's concept of how to make ranged and melee, just throws the whole concept of risk/reward tradeoff into the sewers. XD

"

SunPrice wrote:

well let's say i get a critical which deal 15000 damage total , he got 30000 armor ==> 14,3% damage reduction .
if i land a critical which deal 10000 damage ==> 20% damage reduction ...
So i have to deal extra 5000 damage ( which is 50% extra ) to reduce 7,7% enemy's damage reduction from armor ...

See my point now ?

What i want is something like '10-20% Armor piecing per critical hit' , it makes sense .

And critical hit could not be affected by those endurance charges also , lol

You are using values where armor piercing wouldn´t make much of a difference anyways.
Just look at your calculations. If the enemy armor only amounts to 20% damage reduction, piercing that armor would at most make a damage difference of 20%, which is negligible since it is not much better than going from 20% to 14%.

Lets look at some better values.
Lets say the enemy has 30k armor and you deal 1000 physical damage per single hit.
30k / ((1k x 12) + 30k)) = 71% damage reduction
The enemy would take 290 damage.

Now lets say you crit for 2000 damage.
30k / ((2k x 12) + 30k)) = 55% damage reduction
The enemy would take 900 damage.

As you can see, instead of critting for double the amount of damage, you crit for triple the amount.
At values where armor piercing would make a difference, crits already reduce armor innately, and thus crits already work like armor piercing attacks.
The values you brought up are pointless because if you hit for such a huge single hit, you dont need any armor piercing in the first place since you already negate most of the enemy´s armor.
Only against enemies that reduce your damage greatly (70% or more damage reduction) is armor piercing something that would benefit you greatly, and crits already are doing a great job in these scenarios.
If you encounter enemies that only reduce your damage by 50% or less, then you can easily kill them without critical hits just fine.
If you encounter enemies with more than 70%, or even 75%, then things can get very hard to kill.
Critical hits pierce armor greatly in situations like these with the current formula.

Here is another scenario where piercing would be of great help.
The enemy has 50k armor and you deal 1000 damage with a single hit.
The enemy would have 80% damage reduction, which makes him very hard to kill.
Instead of 1000 damage, you only deal 200 damage.

Now you crit for 2000 damage.
Now the damage reduction is down to 67%.
Instead of 2000 damage, you deal 660 damage.
660 damage is more than triple the amount of 200 damage.
Instead of critting for double or triple, you are critting for even more than triple the amount.

Now lets say you crit for 3000 damage instead of 2000 (since crit multiplier is easy to come by in this game).
The damage reduction is down to 58%.
Instead of 3000 damage, you deal 1260 damage.
Instead of triple the amount on crit, you deal more than 6 times the amount.

Critical strikes are already armor piercing HEAVILY in this game, except in scenarios where the armor of the enemy would make no difference anyways (very low damage reduction values).
If the enemy has no damage reduction to begin with, piercing does not help with that too much.
If the enemy has a lot of damage reduction, critial hits already pierce that armor to great effect.

Basically what this post proves, is that if you go physical damage, you should definitely favor critical damage and critical strike chance.
Thats also why my 110k physical dps duelist in closed beta was doing just fine.
He had 100% critchance 100% of the time, and thus pierced armor on every hit.
Diamond flasks were removed, and thus physical builds nerfed to oblivion.

I think what physical crits should apply (as a status ailment) is "stun weakness".
The duration of this ailment would always last a certain amount of time (lets say 5 seconds), but the strength of the ailment would be determined by the amount of damage you did with a single hit.
This stun weakness basically reduces the enemy stun treshold, and the more life you took with a single hit, the further the stun treshold is reduced.
If you hit for a large enough portion of the enemies health, the stun treshold should even be lowered to zero, meaning that you can stunlock the enemy for 5 seconds.
This would make physical much more viable.

I'd like to see melee crits function like Puncture. I also believe the calculated relationship to armor and size of damaging hits needs to be evaluated as well.

"

mustafa2812 wrote:

"

In addition to the new loot rules, please add a way so maps drop with an indefinite timer for the maker. People will NEVER run public maps until this is done. And i want to!

Second. Maybe make this an option for people who create public games.

Bullshit. Just as others will take creator player maps so can that player go to other public map groups and take their maps