Tuesday 31 December 2013

Converting a storm call totem build to crit

I got an interesting message in two parts today. This particular friend of mine comes up with weirder builds than I do, but I have more that reach endgame. I suspect this has to do with what we consider finished - he just drops characters when he's bored of the concept, and I push it through to either success at lower maps or failure somewhere along the line. Please note that I am not saying either style is better or worse: I'm pretty sure he's also played more different builds than I have, and I have this whole backlog of things I want to do but can never find the time to...

This is the second part of the message. Commentary on the first part in another post, but there's more data for this one.

Current Storm Call Tree
Storm Call Crit Tree I've come up so far

Level 80 scion, EB dual totems using storm call, considering respeccing into crit.

Relevant stats from first tree: 168% HP, 80 dex 220 str 140 int, 140% mana regen, 48% mana, 52% aoe radius, total about 130% spell damage. 
Relevant stats from second tree: 184% HP, 40 dex 170 str 210 int, 100% mana regen, 40% aoe radius, total about 90% spell damage, total 405% spell crit and 50% spell crit multiplier.

Some points, in no particular order of importance.
- good luck with your dex requirements mate
- totem range node wasn't worth much anyway
- yeah 140% mana regen is probably overkill
- consider getting a Divinarius
- also consider using ice spear pcoc during downtime between casting totems
- is unwavering stance really necessary? that's a lot of points.
- also what's up with taking out that HP cluster near CI/EB?
- small 15% crit nodes are really, really small and I'd think twice about getting those immediately, how badly do you need them?

94 points build with US removed. Also added the Coordination node near shadow for +20 dex and 3%cspd, and removed the small cluster of spellcrit stuff near templar (55 crit, 30 multiplier) because it took a whole 4 travel points to get there. For 2 travel points below RT you can get 40% multiplier in 2 points, and for 3 points you can get 45% crit just on small nodes, and I believe I've already made my stance on small nodes clear. In short, the reason I removed it is that that was the most inefficient cluster and I needed to cut points.

Relevant stats: 160% HP (suggest keeping that HP/str cluster below CI, by the way), 60 dex 100 str 180 int (did you know that HP/str cluster is a whopping 50 str?), total 350% spell crit and 20% multiplier. You can probably tell what the 101 point build I would go for is like.

Spell crit has some really spectacular numbers available from wands/daggers and shields, and this is a pretty decent position to run ice spear pcoc from. Multiplier is perhaps a shade harder to get from gear than passives, relative to spell crit (the position, incidentally, is reversed for weapon crit and multiplier) so I might consider taking out the small nodes in favour of multiplier ones too.

I hope this helps.

Thanks a bunch for the aura changes, GGG

So I was respeccing my ranger to an Aegis Rain build. Pretty standard, I know. But the point was to make a reliable build that could get her to level 90, and there's nothing more reliable than Aegis Rain.

I plugged in (level 19) RM - Grace - Determination and turned them all on. Oops, I have 85 mana left. Big whoop, now I can't run Hatred or Tempest. No reduced reservation nodes on this tree, after all.

Current status: 33 points unassigned, level 86. Cleared a3x in merc, though because said points were unassigned I was using an Astra. It's a really good cheat during testing. And as a result, no Lazhwar = zerg Dominus. He died eventually, but I had to stay the @#$% away from the lightning and when I didn't I died. Using BM Cyclone in Rain, yes, I know, how standard. The DPS was actually pretty significant, though my 320dps claw skewed that a little for sure. Comparable to the doublestrike build at least.

Resolution, what resolution? The solution is simple, spend a couple of passives on shield block and give up the idea of running Hatred. That one's a fairly common aura to find in parties anyway, I understand, and it's not like I lack DPS, see above. Hatred would be nice but not actually necessary.

I still don't know what my main skill will end up being. BM cyclone with a claw was just plain convenient to do the testing with. Considering the mana thing, I'll probably end up staying with BM. Hm.

Next up, an aura bot build. Because I think it would be funny to run Ephemeral Edge.

Monday 30 December 2013

How to make money in this game

What would you do with an ex drop? If you're like most people, you'll just spend it on something you want, and that'd be the end of that. Or maybe you'll lament how the only people who earn currency are the no life market flippers, and then spend it on something you want. But you know, there's more to it than that. 

I'm going to sort this by methods, and of course you can do each at different levels of intensity. Pretty well everyone I know who's got significant ingame currency uses at least one of these. With a side of larceny on one memorable occasion*, but that was in addition to this, not instead of it.

1. Playing the game

At its most basic level, this involves selling magic/rare items for chaos and alterations, or even transmutes. Slow but steady. Alternatively, getting a couple of Perandus rings and selling id scrolls to people. One level up, and it involves selling drops for currency. The majority of saleable drops are from maps in the 72+ range, and most of the better drops in that set are from i75+ items. Which is just as well, because at that level, it takes a lot of currency to feed a mapping habit. One other method, though, involves running all maps from 72-75 and selling anything that drops both above and below that range. Between that and good drops, it's possible to break even on solo mapping and then some. If running 76+ maps consistently, the sale of ilvl whites for crafting is a good side income.

But wait, what about MFers, you might ask. And sure there are some people who farm Piety/Dominus with MF and play with the chaos recipe. But a dedicated MF support character isn't in the business of alterations - a typical map run with one of those involves a stack or two of id scrolls per person, rares dropped on the ground as you run once it's established that they aren't good rares, and multiple uniques per map. The good ones get sold, the bad ones get dropped, and yes, people leave uniques behind.

2. Playing the market

Buy low, sell high. The best way to do this is on currency and uniques, because the demand for those is guaranteed. You can do this the regular way, which is to know the market well enough to establish a range of prices - and do note that prices change by time zone: ex are 26-27 chaos at US prime time on Nemesis, but 28-29 at EU prime time. Then buy low and sell high. You could start small, with chaos/alt/fusing flipping, and move on to bigger game with Andvarius flipping, and at the end of this line are mirror and shav flipping.

Or you could find and exploit a hole in the market. In early 2013, 20q FP gems sold for 2 ex during the FP craze, but below that (in the 13-18 range) for several gcps. People made exalts per day upgrading and reselling gems. During the entire time Onslaught was active, death rush rings were in the range of 2-4 chaos on OS...and 1-2 ex on SC. I made exalts per day on that and the cross-trading forums (1 OS = 4 SC). I will not post any current exploits.

This is where I'm going to classify "running Dominus for people" - there's a market, you fill it.

3. Playing the slots

No kidding, crafting is a really big earner, once several conditions are fulfilled. One, you have sufficient currency to start and to ride out streaks of bad luck. Two, it's done on large enough scale that item sales take place relatively often. Three, effort has to be made to actually sell items. Once these conditions are met, however, one realises that others will pay them to bet against the RNG. The chance of spending 2-3 times someone's asking price on crafting a usable item for yourself is too high if you're going to just craft the one item, so it makes sense for you to just buy said item. The guy crafting said item, on the other hand, is rolling enough items that his luck averages out. For every bad luck streak, he's getting a good one - and his asking price is rather more than the average amount he spends. And ta-da, instant profit, both sides satisfied. (Yes, good luck streaks exist. I once spent 6 chaos rolling two pairs of ES gloves for about 3ex value each. I've also spent 4ex of chaos on a helm and gotten absolutely nothing worth selling. Go figure.)

This has gotten long enough, so I'll analyse the effects of this...and why I don't much favour SFL...in another post another day. But you can probably already tell from here.

*Don't ever trust Hauntworld with anything you can't afford to lose. He's a pretty cool guy to talk to, and really good at PvP, but don't trust him with your stuff. I've also heard some things about his guild, and from observation it seems to be more than plausible, but have no personal experience there.

Earning currency in solo self-found

Cross-post from this thread. Will post actual opinion when I wake up tomorrow, this one's for the calculations.

My total net-worth in Dominion was 4 chaos and 2 stacks of alts plus a couple of miscellaneous orbs, at the time I lent them to a friend. In addition, I had bought about 4 stacks of transmutes (1alc per stack), leveling weapons (3chaos + 2chaos), Sunblast (8chaos), Karui Ward (1chaos) and Deerstalker (1chaos). Total of about 23 chaos. Two characters, level 47 and level 30. Can't use /played because there's a lot of dead-time where I AFK for lunch or just chat, but you can draw your own conclusions from the character levels. Ratios are assumed to be 20 alts to a chaos, 20 chromatics to a chaos, and everything following from it. I sold no items but I did freely convert currency in trade chat.

This is including picking up every blue item so I can id and vendor it - the average return, buying 160 ids to 1 chaos, is that I end up with about double the amount of currency I bought the id scrolls with when I run out of scrolls. I make close to zero if I don't pick blues, simply because rares and orbs drop so rarely. With a Perandus ring and no significant MF, I don't need to buy id scrolls even when picking up blues.

I can only do back-of-envelope calculations, but bear with me here. Clearing normal difficulty is about 3 hours on a fresh start (check the most recent 3 hour race). We'll say my level 30 cleared normal difficulty, since he pretty much did, so 3 hours there.

I'll now take some numbers from someone else's thread: level 65 in 10 hours at the highest clearspeed with a leveling build and all best gear available. I have used said leveling build, as well as others, and I can reasonably say the normal difficulty clear time is 2 hours, give or take 0.5. On the level 47 character, for all intents and purposes I started using a leveling build at level 28 (Geofri's Baptism) and have cleared the entirety of cruel with it. More or less. That's 3 hours, plus we'll just say half the time between a1c and a2m, so 4 hours. Would have been longer if I'd leveled with my final build.

So that comes out to about 3 chaos per hour, rounded up, playing solo self-found without MF gear and identifying every item. I'm not including the time taken going over vendors for chromatics, or going back to town to sell items, which are both fairly significant time sinks. It's likely to work out more around 2 chaos per playing hour.

I can make the figures per hour mentioned by someone else (80 alts, plus a few chaos from recipe, plus drops, adds up to about 8 chaos per hour) if I run a high-clearspeed MF build in endgame Piety/Dominus or maps. Goes through id scrolls at a fantastic rate too.

Thus, 2-8 chaos per hour without trading. There is an alternative - buy a pair of Perandus Signets and sell id scrolls for 200 to the chaos - I'll probably test it some other time, but this method is severely curtailed if you also want to find usable blue items. Blue items are really significant when you don't have much else. And, y'know, it kind of assumes trading.

Added a new page to the site

I keep on promising to do things. Like build concepts. Also I have the skilltrees for some of them in my phone already, give or take a couple of points. But the list is growing, so I've added a page where I keep track of them. It's called accountability.

Someday I'll also clean up the rest of the pages. Someday.

A collection of semi-random points

I'm tired today, and it occurs to me that I haven't written one of those opinion pieces for a while, so here's a compromise and expect to see some of these written in more detail somewhere down the line. 

- Playing HC when distracted is as great a way to die as playing while half asleep
- Someone I bought something from is of the opinion that races should give 3x more points but occur 1/3 as frequently; it'd be as hard to reach 1k as ever, but the time investment is less. Also, I metagame races for points.
- NotRegret's quest to beat the game on Nemesis without using rares is commendable. He also has a bunch of essays written on assorted different points, some of which I disagree with, but definitely worth reading even so.
- I'm not particularly in favour of SFL. This from the perspective of end game mapping. This thread is also a good read on the subject.
- Not every build is equal in clearing all content. Case in point, I absolutely hate first form Dominus, but second is a cakewalk; my dual totem buddy reports exactly the opposite.
- I really need my previous forum posts in list format.

Oh, and Munchkin is a great card game. 

Saturday 28 December 2013

Back to Searing Bond

Cyclone + CMK + SB + culling?

Or CMK anything really, using culling to get around the dual totems no-self-damage thing, but it'll only really work with casting a totem because otherwise the spell still deals no damage.

To no one's surprise, self casted RF deals no damage on a dual totem build. Burn animation still plays though. 

Aegis, Lightbane Raiment, and Viper Strike

got an idea from this thread

The question is whether desecrated ground is affected by DoT/chaos nodes? My guess is no. Pity. I mean, I've got a Rain and everything, but Rain + Aegis builds are so overdone, and X-on-block is a cool mechanic.

I'm thinking Abyssus, to offset using a 1h weapon for damage. Aegis should be good for the damage taken thing, not to mention it has a reasonable amount of armour. In addition, crit multiplier, which means I should go crit for even more damage.

I'll post a complete concept once I've shaved it down to 100 or so points and figured out Lightbane's ground effect.

Sunday 22 December 2013

Traps and totems, really?

So there's this friend of mine, he wanted to see if he could get around the "less" modifier on totems. And of course, he hit on the idea of trapping them.

I'd always figured that the summon time would be a bit too much, that throwing a trap which then threw up a totem only if it was hit and then totem cast speed...yeah, not going there. But then he mentioned Sunblast.

Oh snap.

Imagine the possibilities.

I asked, and he wasn't double dipping Reduced Duration (it was storm call totems) but I imagine that's definitely something one could do.

It does make me want to throw lightning warp totem traps now, too. 

Friday 20 December 2013

Light at the end of the tunnel

moving away from you at speed c.

It gets a lot better after lightning trap! (play tested as witch trapper in 2h solo race)

I fail at descent champions


Previous post

So I put together a build plan for a shadow (I thought it'd be easiest to rank) and decided to go for it. First try, got killed at Kuduku. Go figure. Second try, ditto. Third try, I actually got past there, went into obelisks of faith, and got killed by the goatmen. By this point, I'm thinking something's a bit wrong, so maybe let's rerun the old build and see what I'm missing?

The old build was Scion, 2h ST with fire trap support, except the way I played it was actually a great deal more like fire trap with 2h ST support, at least in the early part. Every skill point went toward ST, so it made sense that it moved away eventually, but I thought why not go for firetrap? But anyway, I rolled a scion, it was easy, and I learnt something!

- I learnt that the limit of 3 traps at once is pretty annoying. I probably also spam them a little too much. And I'm pretty sure now that trap positioning is important in order not to waste traps.
- Traps of all kinds work on Kuduku, but it's a lot easier to have his minions trigger the fire trap and burn him with it.
- Freezing halls is annoying as @#$% and it's also not worth going out of my way to find the thing if I find Kuduku first. 20 movespeed avian slippers are fun though.
- Trap builds tend to leave items behind because, well, trap build, throw it down and move on. This is very annoying too.
- I really, really hate galvanic ribbons. 
- And I probably shouldn't be testing in races when I'm still a fair few points from the reward marker I want. I have slightly messed up priorities when it comes to gaming, apparently.

Most importantly, I learnt that it's good to have a cleanup aoe besides the traps, at least in early game. This should also solve the problem of leaving stuff behind. However, shadows...don't exactly have a safe and fast cleanup aoe like ST. Closest I can think of is reave, maybe that'll work? Here's hoping I can get enough points in the regular races to use D:C as testing grounds, rather than having to push hard with an old build.

Monday 16 December 2013

Going to make searing bond work, just watch me

The idea post

The basic idea is to run totems on CDT so I get a free source of casting totems. HP return will be done by life-on-hit, since I can't exactly leech on ancestral bond. Since the bond connects to me, I should be right up in the middle of everything, therefore cyclone.

Using a searing touch and soul mantle, with the requisite flasks of warding. 

Cyclone - life on hit - curse on hit - ele weakness
Searing bond - CDT - increased burning - empower - (does fire pen affect this?)
(soul mantle spelltotem) - q20 shock nova - conc.e - proliferation - CDT - enhance - ??? (probably running a level 19 CDT in order not to proc it at the identical time with searing bond, maybe)
Fire trap - proliferation - conc.e - increased burning (as long as we're doing burning damage, why not)
Lightning trap - proliferation - ??? (ditto the above)

Build link

Yeah, that's got like no dex on it. More's the pity. Anyway, notables: RT (no way I want to miss anything that returns life on hit), static blows (well, duh), ancestral bond (OP), US (running this on CDT, right? not sure this is essential though), etc. I'm not entirely happy about how the entire templar section links up, but luckily that's all non-essential, so meh I'll think about it later.

Turns out not a lot of supports affect searing bond, maybe I'll keep it 4-linked in that case, which would allow me to use a 1h and shield rather than searing touch. Which in turn would mean skip the staffblock cluster, reroute templar (I did say I wasn't happy with that) and get shieldblock.

RF - empower - aoe/conc.e, in a +2/+1 (does RF get negated by dual totems?) and using ruby flask(s) of warding.

Might need to pick up some armour clusters. Take the hybrid ES ones and go Aegis maybe. That all depends on how much I can afford to not be hit and still keep the totems up for a reasonable amount of time. Lose the fire and burning damage from ST though, not to mention the additional +2 to searing bond, and that's actually really significant.

Update: dumped the templar nodes, and in my previous experience with cyclone as HP I don't need US either. That way I can work on managing damage taken through ev as well as ar. build link the last nodes to be taken are Arsonist and the DoT/totem clusters below IR. Non-essentials, the whole bunch of them.

Thursday 12 December 2013

Descent: Champions

I got my first Mirror yesterday. Missed 9 points by a couple minutes, started a couple minutes late, have a really bad habit of pausing to do inventory rather than doing it on the go. The last few Descent: Champions races I've been following the same strategy, getting a feel for the zones. Like I prefer Croaking to Chattering Halls because chimerals are easier to deal with than avians...both suck, mind you, but one sort's easier. Similarly, I'd much rather face Hatestorm than Avatar. Anyway, this makes it about time to plot strategy. 

Current: Scion, 2h ST with fire trap support. Everything after Berserking depends on the best weapon found/crafted, so some days it won't go to the axe node. Depending on weapon ias relative to dps, and how much of it is physical, either Iron Grip or the ias nodes at start are taken.

I pulled the CSV of that race (big surprise) and checked it across classes (bigger surprise). Rank #20 of each class, overall rank and level. Pattern holds across multiple races.

Scion: #40, level 22
Duelist: #200, level 20
Marauder: #100, level 21
Ranger: #180, level 20
Shadow: #500, level 17
Templar: #350, level 18
Witch: #300, level 19

Looking at the skill tree, it's no wonder shadow does terribly, they have a crappy starting build and items. Relatively speaking. Not to mention that shadows always run into mana issues. Templar doesn't do too hot either after the early good nodes are taken, and witch has no survivability at all. Yeah.

I'm in a hurry and I have a few more days out of town, so concepting a different build will come another day. I have some ideas in mind already.

Wednesday 11 December 2013

What would you do with 6 green links?

There's a mirrored chest floating around SC, it's got HP tri-res and something like 2k evasion...and 6 green sockets. And it's mirrored. Regretfully, at the time I had the opportunity to buy it, I turned it down. More's the pity. Still, it's been in my mind for a while that I should really try to make a build which works with 6 green links.

List of green active skills that can realistically run as main AoE without red/blue supports. In other words, no melee splash, but xMP is allowed. No comment on how well they run, mind you, only that they can.

Animate weapon? Huh. That's...a thought. (spell)
Burning arrow, explosive arrow, ice shot, lightning arrow, poison arrow, rain of arrows, split arrow
Elemental hit, frenzy
Cyclone
Detonate dead (spell)
Ethereal knives (spell)
Fire trap
Puncture?
Reave, whirling blades
Spectral throw

That's a pretty cool list, actually. Green supports? I mean - relevant green supports?

Added cold (q20 gives chance to freeze, which is pretty cool)
Additional accuracy
Blind
Cast on crit
Chain
Chance to flee
Cold pen
Enhance
Faster attacks
Faster/slower projectiles
xMP
Trap, multitrap
Pierce
Point blank

That means...we're likely looking at a cast on crit build, or maybe an EK-trap build. Pity about not being able to run added fire with the EK. Firetrap's out because it doesn't have enough potential green supports.

Split - fork - CoC - EK - DD - q20 add accuracy for more crit, or anything else. I don't like planning 6Ls. Maybe instead of split, use Reave and a dagger, or WB and the same if you're feeling adventurous. Why not ST, because ST means xMP, and xMP doesn't play too nice with EK. I'd prefer to go with the bow, because fork supports split and does not support reave/WB.

EK - trap - multitrap - standard EK green supports, so faster projectiles and fork...and one more whatever.

Ele hit - xMP - chain - I don't know, does this look like a wand build to you? Maybe use Frenzy with it, because ele hit is just asking for WED and we can't add a WED. (Actually, a lot of things would go well with WED. Too bad.) Accuracy, because @#$% wander accuracy, and as a bonus it gives crit. Oh, I think Enhance would actually go crazy well with Frenzy. Must try that sometime. Should be able to get away with no LL/ML because it's mostly physical, or at least you can flipping well make it mostly physical.

I'll concept these separately. Another day. Yeah, more's the pity I didn't pick up that 6L when I had the chance. If someone knows the unlucky bugger who currently has it, mind pointing them my way?

Monday 9 December 2013

Nailed Fist OP

No, really. I'm posting this directly after a BLAMT race. It became easymode once I got the combination of ST and two Nailed Fists. As easymode as BLAMT ever gets, anyway.

Getting that first ST out really, really hurts though. The idea is to take it slow and pick up a ranged weapon ASAP. Zombies can be melee kited for 0 damage taken, just run away before its animation finishes playing. Sand spitters can't, more's the pity. Once you get a wand/bow, it's free xp and items. Otherwise, BM race though it is, ST is alright to use even on the beach, just don't go overboard.

It's pretty well impossible to beat Hillock as melee, I tried like 3 times and another 2 using ST. Wand and bow kiting both work - rangers and witches have a natural advantage in that they start that way, if you want to just rush Hillock. Hillock involves a bit more running in circles than I like, but it's better than dying.

After the race, I pulled the CSV and played with it a little, so we have stats!

#1 in class: scion/duelist/ranger level 12, witch 11, shadow/marauder/templar 9
#20: ranger 9, scion 7, witch 4 (this one was ranked #585 overall including dead characters)
9 duelists, 10 shadows, 16 marauders and 9 templars made it to level 3 and above (clearly that #1 duelist was an anomaly)
1132 characters made it to level 3 and above before dying; some of those didn't die.

I was going to post a witty comment here, but I think the numbers speak for themselves.

What's the point of Essentia Sanguis?

It's a 130-170 pdps claw (total ~200dps after lightning and quality) which gives Ghost Reaver. But wait, GR is right over near CI and the claw leech nodes. Makes this claw seem almost redundant, yeah?

Well, what if that zone wasn't strictly necessary? Templar's got good ES, and there's a circle above scion, and sure that's pretty close to CI (if 6 or 7 points minimum can be "close") but GR's another 4+ points over. Meh.

So...low life or hybrid build? PA's all the way over there, but unless using crown of eyes that hardly matters. A case could be made for either. 10% additional block chance while dual wielding claws, though the other claw would have to be downright spectacular and you'd be using doublestrike. And half the dw block is over by shadow/ranger. Hybrid incinerate with Cybil's Paw, maybe. Or something. Not sure how well that'd work.

For low life, it'd be all the templar ES nodes, possibly the scion ES wheel too, VP and US on the bottom - we're not going to the GR side, anyway. For hybrid, we can simply ignore that section, and...ohsnap, what if we did neither? Essentia's got a pretty respectable 6.68% crit, for a claw, and...aw man, half the crit is over in shadow, including all the claw crit. Oh well. Next time. Cast on crit's probably out then. Cast on damage taken, however, is not particularly dependent on the weapon. That'd require an Aegis. (I've got one, luckily.)

Tricky tricky. Claw, why?

This makes me want to make a hybrid build. Not necessarily with Essentia, but using a combination of ES and HP. I'll list that in another post another day, I think. 

Friday 6 December 2013

When NOT to play HC

When your mental function is impaired. Now this is a tricky one, it doesn't correlate well with how you feel at the moment. You may feel perfectly fine, but that doesn't mean your decision making is on the ball.

Yeah, I ripped another character doing something stupid again.

Thursday 5 December 2013

Not enough people on the internet appreciate math

Even on Path, and this is one of the most math-heavy games in the mainstream style.

Concepts in this post should all be covered in a basic topology class. Some you might even find in set theory, wouldn't know, never took that.

The question was raised, what is the definition of a "build"? The relevant part of my answer is below.

--
Observations:
- Path build n-space is discrete. It's also closed and contains a (very large, but) finite number of elements.
- Passive tree major or minor changes have some effect sometimes. Big nodes are game-changing, small nodes have cumulative effects, but the only logical way to resolve this is for no nodes to be valueless.
- Skills are different. Support combinations are different, to some extent. As above, there can be only one logical resolution to this. Both these spaces don't necessarily behave all that well if you pick the wrong definitions, so don't pick the wrong definitions.
- For the most part, rare item stats have less impact, and similarly with statball uniques like Kaom's Primacy. On the other hand, there are build definers like Facebreaker. Such items, then, each correspond to a single dimension in Path build n-space, or we can say that Path build n-space is the product space of several spaces, some of which are Boolean in nature and correspond to items. But even statballs have their impacts: physical, spell, both? How much total MF? And so observe: it makes sense to have each individual stat mapping to some interval in R, and the sum total of rares/statballs occupying some point in R^m.

That's...clearly not all different builds the way the term "build" is generally used, though a strict definition should define them that way, so let me introduce the term "setup". Each element of Path build n-space is a different setup.

Luckily for us, setups tend to cluster in different regions of each of the parent spaces that contribute to the product space. Just as an example, on the skills space, taking a subset of all setups which contain the active skill element LA, a lot of xMP, chain, WED, etc are found, while setups containing MPD and splash are conspicuously absent. (If you're wondering, I defined a compact R^6 where "support gem does not work with skill" means that element just maps back to 0 on that dimension. There's a lot of (LA,0,0,0,0,0) in it. Not necessarily the most efficient method, but I'll take that trade-off for the well-behaved-ness of R^6 any day!)

>> I believe it is possible to define a metric on Path build n-space. Just saying. This is important.

But going back to the idea of clusters, that's basically a finer - or coarser, sorry I'm on a timer here not going to figure that one out - version of each of the mentioned spaces. Each element in Path build cluster n-space is a different build. Alternatively, for each cluster choose a point near its centre* to be the element in the corresponding space, and thus define builds.

*this is why the metric was important

Problem solved, right? The only fuzzy bit is the clusters one, and I'm not happy with that either, but how else are you going to draw an arbitrary line? The second method might be better, as in "I'm playing a setup that's close to theWombo's build on Path build n-space".

Monday 2 December 2013

It really does have crap damage.

I was slightly preoccupied yesterday getting an endgame gear set on Nemesis, among other things, but found some time to squeeze in a little bit of testing with my pretty Darkscorn. I chose to use Eye of Chayula for anti-stun instead of taking US, and used Alpha's Howl for anti-chill along with anti-freeze (credit to Perz for pointing that one out to me!) Testing was done in docks, with the rationale that if I can't neatly clear it with a gear set thrown together in 5 minutes, then I can't clear higher maps with gear I spent 6 hours optimising.

Started by respeccing a 70 witch into the dex (evasion) variant of the tree. LA GMP chain WED LL - couldn't sustain mana. Well okay, only 25% of the damage gets 2% leech from Mind Drinker (I later specced out of that, so it's not in the tree linked above, but did take it at some point in time), and I'm not trying for crazy high damage so it's not enough, I get it. Substituted a mana leech for chain - turns out chain's important in getting LA to hit the max number of mobs. Huh. Regardless, even with evasion, I'm taking too much damage and not dealing enough. This sucks.

Well. The easiest option was to go via the str half of the tree and take IG. So that's what I did with my 80 witch. (Recoloured Darkscorn to 5G1R somewhere along the line so I could run both chain and mana leech, couldn't be bothered to recolour further, didn't have a WED gem slotted in but 1.4*terrible is still terrible. I figure as long as I'm doing the testing with the same gem setups on both, I can get a decent comparison in.)

First thing I noticed. Knockback is rolled per attack, like crit, and not per hit. That makes the little 15% knockback wheel next to useless. Next thing: went IR, and it boosted my survivability a ton. I kinda sorta knew that, but still. I'm still not doing enough damage to return as much HP as I'd like, though it's definitely better with IG.

To do:
- try out Blind
- try IR with the dex tree, and ditto the above.
- try socketing Knockback gem along with the skilltree wheel
- try a physical skill (split arrow + chain/fork?) which should allow for better mana return from leech; not sure how I'd pull off static-blows except by gear lightning damage though, and that's always iffy to depend on. Since I don't actually appear to have a leveled split-arrow right now, though, that might be a while in coming.

Where do I go from here, if it doesn't work? Well, I figure a Chin Sol will be a pretty decent damage boost, maybe enough to kick the 80 into viability for maps on this build. Probably better tanking too, with a guaranteed knockback meaning less damage taken. Would like to make it work. On the 70, I still have enough of my free respecs that I can rework the tree into Perz's crit claws tree, or I can do pretty well anything else that involves CI and the dex side of the tree. I don't know yet.

Or I can leave the characters hanging around until I get another idea, while I play Nemesis. Yeah, that sounds better to me. As a bonus, here's what one of my friends asked when I explained my build to him, and my response to it.
Q: This looks like it would be better done from Scion, why didn't you start there?
A: I didn't have a sufficiently high leveled scion with a full respec to play with.
My priorities are most definitely slightly skewed.