tdphys

I'd really like to back up Sensuki's critical hit resolution idea here... 1. One of the major things that normalizing the ciritical hit chance would do is detach perception from getting crits, which would allow room to put modifiying the crit range on another stat for balancing purposes ( int I'd propose) 2. Actually having a capped critical hit chance, rather then allowing high-per characters to always crit, will give crit focused characters a reason to pick dexterity ( agility? ) to increase the frequency , imho, make it a valid stat choice. 3. It would allow higher crit bonuses, giving that kind of sneaky, high damage, high risk play style that some (perhaps me?) like to play. Using mechanics to get a better crit chance (attack from behind? other character is engaged) requires more tactical thought. 4. It would give more strategic decisions in character building and tactical decisions in combat since there would be two paths to damage, crits and regular hits. you'd need to rely on high crit against soloish monsters with high DT, and good hitting for mobs of low DT creeps. 5. Ultimately, balancing would be easier, since you can effectively tighten damage range per minute based on the difference between character levels/ differences in Per/ Def etc..

I'd love to see the code, but first steps I think would be to get a mock up in unity running with some toy nav meshes and pathfinding AI, then maybe we could see if we could match (ahehmm... maybe exactly) the POE implementation. I'm going to see If I can get Unity Editor running in wine on linux or in a VM. Can somebody verify for me that POE is using Unity 4? This will probably take a weekish , but hey, we've got months now Maybe some of you unity gurus can put a test case together. Even if we don't help fix things, maybe we can inform the devs a little bit more, and perhaps learn something in the process. I'm more interested in the pathfinding algorithms themselves, but I don't think some Unity chops wouldn't be a bad thing to acquire.

So I'm not a backer, expecting to Pre-Order and hoping to contribute to the *Profit* argument for POE However, I'd like to contribute and given that we now have 6 ish months, I wonder if anybody could fill me in on what's wrong with pathfinding. Does anybody know what algorithms/data structures OE is using, and maybe we could set up some mock tests/frameworks and do some crowd source research on how to help fix it. Does Unity come with its own pathfinding libraries? Edit: I guess this is what they're using, and why Sensuki keeps referring to Nav-meshes http://unity3d.com/unity/quality/ai So what's the problem with the unity based finding in the beta's? Is it a quick fix or something more substantial. Okay... I assume it's A* path finding using nav-meshes as detailed here. http://unitygems.com/astar-1-journeys-start-single-step/

I guess I am unable to rise to the level of childishly calling people mentally ill for disagreeing with me. Truly high quality rhetoric on your part. If that's your impression of what I said, and you think it's childish - then why not rise above it and prove me wrong properly? In any case, the problem is that some argue that numbers that have no meaning or relation to anything, as long as they become bigger and bigger over time -- is the ultimate form of entertainment. And they insist that there is no other possible point of view in the entire world. End of discussion. I don't think that this is a reasonable point of view. (You can also tick off the two top indicators for Aspergers disease from that - the point being that the game as it's headed is going to be obsessively engaging for people with Aspergers on one of the bad days. And like I said in the other thread - if that's who Obsidian wants to appease with my pledge money, there are more effective ways to do it that will actually help. Last chance before the money goes to a charity of my choosing.) I don't believe you actually got the money back (I don't think you should get a return anyhow), but If you want to prove it, you can send it to me.

Is there any info on when pre-orders will be available?

Yeah, but the explosion of splat books to rescue trap builds isn't really available to POE devs who are trying to release a playable game. That and the feat bloat of certain DnD editions is a holy war design decision itself. In terms of computer games, I'd almost be happier if attributes were separated into narrative ones and combat oriented ones. You could min-max or balance combat ones and then have the narrative ones be driven by story-telling. For example all the narrative options are available, choosing one gives a dice roll and then pumps your narrative stat in that area. You could run with narrative attributes like Intimidation, Intuition, Kindness, Deceit, Arrogance, Logic etc... kind of like the reputation system. That way you'd be harnessing the computational power of your machine to enable better storytelling. Physical challenges could stay related to combat attributes. DnD has to dump all of combat/physical/storytelling challenges into 6 stats, primarily for accounting purposes, with the rest being a roleplay-background mechanic which relies on imaginative game play. Computer games are limited in computing imaginative game-play ( that's dependent on the player ) , but not in the accounting department.

I'm pretty sure that making a high -int dump everything class in DnD except wizard makes an absolute useless Char. In the modern DnD's (4e more then 5e) if you don't pick the attributes designed and stated in the manual that your class needs, you can pretty much kiss your class features good bye. Sure, you get some skill points you might not have been able to use before, but you'd be absolutely neglecting the reason for picking the class in the first place. I've seen people do this, and they spend months not hitting anything and being pretty much useless, even in a pen-and-paper sense. It's pretty painful, they participate in Role-play and then zone-out in combat. In a game... there's no role-play (there's narrative) and a dead-end character like that is IMHO, it is something it's really important to avoid. Also, If obsidian can be successful, by making attributes provide different , but successful game play avenues for all characters, the re-play value of the game benefits immensely. I think it looks like they'll succeed, along with help of beta testers who buy into this idea. I don't even think the narrative will suffer. I would like to remind you that the same will happen in POE. Just try and make a high int dump everything else in PoE and report back. Ahh, my hyperbole catches me. I should have said, intelligence is a useless attribute to anybody but the wizard, rather then propping up a sole-stat character. And your arguments probably correct, the design process currently trying to optimize towards no stat is useless, doesn't necessarily allow single pumped stat characters being useful. I think by trying to not make attributes interdependent for mechanics might actually reach this, which would be fantastic, but it's a harder design. If the designers are truly successful, It would be fun to see single pumped-stat runs for different classes, requiring different type of game-play for each. Half of most people's problem seems to be affecting the immersion/simulation desires, ie the game in their head rather than the one on the computer. I'd offer that once the game starts to play, that'll emerge, it's just people are married to their old imaginations, and resistant to the new

I'm pretty sure that making a high -int dump everything class in DnD except wizard makes an absolute useless Char. In the modern DnD's (4e more then 5e) if you don't pick the attributes designed and stated in the manual that your class needs, you can pretty much kiss your class features good bye. Sure, you get some skill points you might not have been able to use before, but you'd be absolutely neglecting the reason for picking the class in the first place. I've seen people do this, and they spend months not hitting anything and being pretty much useless, even in a pen-and-paper sense. It's pretty painful, they participate in Role-play and then zone-out in combat. In a game... there's no role-play (there's narrative) and a dead-end character like that is IMHO, it is something it's really important to avoid. Also, If obsidian can be successful, by making attributes provide different , but successful game play avenues for all characters, the re-play value of the game benefits immensely. I think it looks like they'll succeed, along with help of beta testers who buy into this idea. I don't even think the narrative will suffer.

MATLAB is a horrible programming language... I kind of loathe it actually. Numpy and Scipy for me > But all the gotos in that code freaks me right out. All I can say is "who does case 5 work for?' I wonder if it's and enum and the compiler killed the name for the goto point. For most programming, yes. xD For math and engineering though - it is wonderful. You won't hear me speaking badly of the program that kept me from evaluating 36 integrals and derivatives by hand last night. ;P Numerical or Symbolic integration? you'd be the first person I'd met praising matlab's symbolic's capabilities, usually you'd go to maple or mathematica for that. Yes, I know lots of mathematicians and engineers who like matlab, and I've used it a fair bit and It's really powerful for scripting up and evaluating short problems. Once you get a system with over a couple m files, it quickly becomes unreadable and the scope rules are horrendous and the Object oriented part is just tagged on... Numpy and Scipy ( and sympy for that matter, if you need to do symbolic stuff) give you all the goodness of python and close to the mathematical scripting power of matlab, and it's easier to rewrite slow code in Fortran and C++ if you need to. However, no need to descend into a holy war, there's enough of that on these boards, unless you want to argue emacs of vim. I agree that matlab is a powerful tool, just not my preferred one Oh , and are you using matlab to do homework ?? (shame shame ) edit: I should add, the UI and Matlab debugger is nicely put together. It's hard to get something similar for numpy/scipy unless you use something like sage.

MATLAB is a horrible programming language... I kind of loathe it actually. Numpy and Scipy for me > But all the gotos in that code freaks me right out. All I can say is "who does case 5 work for?' I wonder if it's and enum and the compiler killed the name for the goto point.

Hmm, doesn't really state all the modifiers. I'd ask you to post more... but I'm seeing visions of the copyright swat team showing up at your door

I used a program to reverse engineer the Assembly-CSharp.dll to get the source code. I can see pretty much everything. Hah.. that's hard core... did you make up the function names? If it was c- c++ I'd expect they'd be mangled unless they're shipping debug builds. Regardless, is it possible that you can post the function where the actual damage is calculated? I assume that's way more complicated and I'd love to see it. That's not quite true. The increased damage from Accuracy comes from the increased chances of better attack resolutions and the decreased chances of worse ones. When Accuracy is greater than Defense, that is indeed an increased chance of crit - but the increase in DPS isn't only dependent on what the "good" attack resolution is that is being maximized, but also the "bad" attack resolution that is being minimized. There are distinct 5 regimes of "Accuracy - Defense" - and the effectiveness of one additional point in Accuracy is different based on which regime you are in. In the "-5 < ACC-DEF < 5" regime, each point of Accuracy increases the chance of a crit by 1% and decreases the chance of a miss by 1%. This is the most powerful regime of ACC-DEF for Accuracy, because you're maximizing the best attack resolution and minimizing the worst one. Raw 1.5% DPS increase (before factoring in Might, Speed, and other multipliers to DPS) "-50 < ACC-DEF < -5" regime: 1 point in Accuracy increases the chance of a hit by 1% and decreases the chance of a miss by 1%. Raw 1% DPS increase. "5 < ACC-DEF < 50" regime: 1 point in Accuracy increases the chance of a crit by 1% and decreases the chance of a graze by 1%. Raw 1% DPS increase. "-95 < ACC-DEF < -50" regime: 1 point in Accuracy increases the chance of a graze by 1% and decreases the chance of a miss by 1%. Raw 0.5% DPS increase. "50 < ACC-DEF < 95" regime: 1 point in Accuracy increases the chance of a graze by 1% and decreases the chance of a hit by 1%. Raw 0.5% DPS increase. So as you can see - the benefits of Accuracy are a little more complicated than just "increased crit chance". It very much depends on which regime of the Accuracy - Defense axis you're in. Do note that these "Raw % DPS" increases can't be directly compared to, say, the 2% increase from Might because they actually depend on each other. Low Might means your % DPS increases from Accuracy are worth less, and low ACC-DEF means your % DPS increases from Might are worth less. This means that there is actually a crossover point where either Accuracy or Might is better for increasing DPS. In the last build, below ACC-DEF = -20, Accuracy was generally better, and above ACC-DEF = 5, Might was generally better. The region in between wasn't one or the other. I haven't redone the graphs for the new Might multipliers (centered at 10), but I expect they'll be very similar. EDIT: Granted, the last 2 regimes I showed may or may not see much use in game as those are pretty big differences. Even so, thought I'd include them. Also, I can still see your point about crits maybe being better if they weren't super common, and if Accuracy only increased hit chance. I'm... undecided. I think that argument has merit, but I'm not sure how well that'd work in this system. One thing is for sure - that change would nerf Accuracy quite a bit when compared to Might. YMMV if that's a good idea or not. Could probably be counterbalanced somehow, maybe by moving the attribute boni to +3% and +2 respectively, which gives a little more weight to the Accuracy attribute (currently Perception). Yes, it's the third case which I was referring to , where dps comes strictly from higher crit chance. I wasn't counting the reduced graze chance, just to not complicate things. My bias towards crits being an exciting thing is obviously my own D&D based prejudice, so I don't want to get all huffy about things. I'll just call it amplified damage That being said, I think it would be fun to put INT as an added damage multiplier to the amplified ( err crit ) damage along with might. I think that goes along with the int based def narrative, and would be more effective in buffing int then AOE effects.

Sensuki, how do you have access to the C-code? ( not complaining, just curious, and how much code do you have access to, could you post, for example, damage calculations, etc?) Thanks, the wiki article made good sense. The dependency of crits probability on accuracy is nonlinear, since above a certain threshold, there is zero crit chance while below it there's a linear dependency. This seems to me to make accuracy a very critical part of DPS by enabling crits, but from what you've said, PER doesn't do that much to affect it compared to class features, and therefore isn't necessarily the buff stat it was hypothesized to be in the beginning of the thread. I actually think that INT as a critical hit modifier (def) is narrative, as the genius/knowledge it takes to exploit a persons anatomy/situation or avoid being exploited. So if Int is still the nerf stat I think it could be buffed by moving crit ranges rather then, or including the AOE mods. Well, it's fully linear within the regime in which crits are possible. 1 point of Accuracy is always +1% crit chance if crits are possible. And accuracy is indeed a very critical part of DPS - but its not more important than Might. 1 point of Accuracy gives, at most +1.5% effective DPS (+1% of 150% damage and -1% of 0% damage in the -5 to 5 range of Accuracy - Defense). 1 point of Might always gives +2% effective DPS. Now, the true value of a point in either one of these is actually dependent on the value of the other (you can't compare the flat percentages) - but in general, Might gives a lot more effective DPS if your Accuracy is higher than their Defense and Accuracy gives a bit more DPS if your Accuracy is lower than their Defense. Might and Accuracy are pretty well balanced as far as DPS goes (assuming there isn't an overwhelming trend of enemies with more or less Defense than Accuracy values, that is). The problem with the attack resolution for me is that all of your expected damage increase from accuracy comes from higher crit percentages. It seems to me that it excludes mechanisms that could meaningfully increase hit chance without increasing crit chance. I realize that some modifiers do this by reducing the baseline crit percentage, but once that's set, higher accuracy only gives higher crit chance, the to-hit chance without critting is always the same. I'm not knocking the decision, I just think crits should be that once in a while awesome thing that happens, not something that happens 50 % of the time because you're against a weak enemy and your accuracy is way better then their def.

Thanks, the wiki article made good sense. The dependency of crits probability on accuracy is nonlinear, since above a certain threshold, there is zero crit chance while below it there's a linear dependency. This seems to me to make accuracy a very critical part of DPS by enabling crits, but from what you've said, PER doesn't do that much to affect it compared to class features, and therefore isn't necessarily the buff stat it was hypothesized to be in the beginning of the thread. I actually think that INT as a critical hit modifier (def) is narrative, as the genius/knowledge it takes to exploit a persons anatomy/situation or avoid being exploited. So if Int is still the nerf stat I think it could be buffed by moving crit ranges rather then, or including the AOE mods.

It seems odd that accuracy would affect crit chance. To-hit bonuses don't in DnD.

Here's how it was supposed to work: "Yeah, the way we display the non-verbose final roll is always relative to the standard ranges: Miss on <=5, Graze on 6-50, Hit on 51-95, Crit on >=96. Defense is subtracted from Accuracy and then applied as a modifier to the roll itself. E.g. the attacker has 52 Accuracy and the defender has 30 Defense. The difference is 22. Three attacks happen in sequence. The actual rolls are 65, 43, and 84. Those are modified to 85 (Hit), 65 (Hit), and 106 (Crit). There are two exceptions to this: a natural roll of <=5 can never be better than a Graze and a natural roll of >=96 can never be worse than a Hit, no matter how much the table gets skewed." So says Josh. If this is the case, then high accuracy gives inordinate higher chances of crits and I can see how that could be overpowered. The fact that crits depends non-uniformly on accuracy /deflection / DS etc also probably lends itself to feelings like this: For simplicity's sake, how about: Accuracy and Defense operate as mentioned for hits and grazes. A hit always has 5% chance of being a crit, a graze has 5% chance of being a miss. ( I think a miss should be a critical miss, like .5 seconds of stun or something) Intellect: anything above 10 gives +1.5% to critical hit chance and -.5% to critical miss chance anything below 10 gives -.5% to critical hit chance and +1.5% to critical miss chance. I think this could balance per and int. And now, to weigh in on the acrimonious simulation vs gamey definition, though I hope it doesn't swamp the design debate, which is more interesting to me. Role-Play Simulation is one of the greatest parts of playing pen and paper DnD. It also requires the flexibility and brain power of the DM and the players. You can't simulate that on a computer. Role-play on the computer emerges from game-play. Having a robust stat system that enables emergent game play will increase the role-playability (with emphasis on playability). You can't ask obsidian to make a simulationist RPG, only ask that it balance narrative and combat game-play in its stat build. I think the narrative part will be primarily supported by the background, writing,reputation and narrative system that will be POE. It therefore makes sense to weight the stats to apply more in combat then the narrative. I agree that trying to make the stats make sense in narrative and combat is a worthy goal, but It's much easier to write in *intellect* into a quest dialogue then make a balanced combat game. It's been fun to see the game design and I'm really looking forwards to playing this.

Would somebody kindly point out the to-hit/damage dependency for stats for a poor forum lurker, as in, is there a wiki entry somewhere? It seems that Accuracy (per) gives better chance to hit and to crit. I wonder if you could fix the balance of stats by leaving Accuracy(per) as more of a better hit but not crit chance and assign intellect as giving a wider crit range. It seems to me the AOE/range thing is a little too class specific, switching to crit range should benefit all classes, you could motivate it as critical hits being the genius ability to exploit a situation to cause maximum damage. (or healing? can you crit with healing spells?)

Making it work would just be a question of UI. Some people have suggested mouse wheel. Another way would be to down-click to place the center and drag left to shrink, right to grow. I'm generally spam pause when I play IE games, combat doesn't seem fun to me unless it's scary enough to require frequent pausing

If the goal is to get INT to be valuable for area buffs/debuffs and ranged spells, then I think the most simple thing would be to allow AOE size to range based on intelligence. So have a set size on an AOE (large for fireball, small for buff/debuffs) and allow int based characters to resize it up and down based on intellect. While decreasing radii might be meaningless to buff that only target allies, it still allows int to be more effective for casters etc who can now surgically cast mini-fireballs. Thus, both use cases get use out of int stats.

It seems to me that a lot of complaints about the attribute system is simply a communication problem, where people expect an attribute to mean something based on decades of experience and habit rather then on the designer description. I'm really excited about the opportunity to have 6 different meaningful ways to customize every class. My hypothesis is that if the attributes were renamed to not collide with past mechanics and with perhaps more obscure english, they might actually encourage people to read and understand what they do. Here we go. Might -> Potency the raw damage/healing effect of your abilities Constitution -> Resiliency the overall ability of you character to sustain long-term debilitating damage Resolve -> Resolve Perception-> Perception Intelligence -> Acumen The ability of your character to employ theoretical and applied knowledge to increase their effectiveness in life and combat Dexterity -> Reflex The innate trained capacity of your character to accurately and swiftly apply their abilities ( muscle memory, mind training etc) And last of all... to balance Potency with everything else, I think it should cause a negative bonus to your lowest stat if maxed to far over the average of your stats, or perhaps just constitution or Resolve; Thus a truly potent character recklessly trades damage for some other important aspect. Now, I'm just going off of lurked understanding of the mechanics at this point, So I can't guarantee that my attribute words correspond correctly. I still think renaming them should be done, at least so I don't get sucked into another monster attributes thread and miss more sleep

Hah, Here I thought I'd just lurk PoE until it was released and buy it on steam... But this forum stuff has just sucked me in. I really like the priest faith mechanics, especially bonus/penalties for following a creed I think the way to really make this an interesting mechanic, however, is to make the penalized priest a ultimately viable path in it's own right, ie... at some low *piety* or whatever, allow an option to renounce the faith, which removes a lot of the class powers/spells, but gives some new ones , in more of an anti-support role. or else they could pick some other non-based faith class to restart in with some added heretic bonuses. the anti-priest / heretic could have an aura than deadens or sucks up healing/stamina (maybe both ally and enemy) buffs into some kind of rage/heresy bonus that gets applied as random damage on attacks, with a bonus against faith based classes I like the randomness, as a result of instability of losing faith. Possibly the higher the level, the greater the bonus ( and the lesser time to specialize in another class). I'm thinking of some poor heretic now( rogue ) being pushed to distraction by a buch of allies priests healing and erupting in massive damage... doesn't that sound like fun? standing bonus to attacks against faith based classes particular bonus against its own order/deity To me this just adds a great avenue to throw more antagonism / conflict into PoE, which seems to be the intent of the storyline, and of which I approve