Hey Torchlight fans!
Echtra_Jordan here, to share something I dug up from the design team’s document stash. This is a massive write-up done in late January after the crew performed a comprehensive review of class skill tuning. These notes have been implemented and the changes went live with the
Mainline Patch on Feb 4, 2020. This is a very “inside baseball” kind of document, a bit of a Designer Diary - not like a normal blog post or announcement we would prepare. I thought it’d be fun to share for nerds like me who love to look deep into how skills are tuned to create balanced & powerful class combat.
In related news - this work was just one step in a bigger effort. The designers are working through additional class skill improvements, skill synergies, and usability reviews as we speak. I’m poking them to get more of the process revealed to share with the Alpha players.
[h3]Notes on the notes, from game designer Michael May(hem): [/h3]
This document was written internally as a result of a series of number crunching efforts, comparing skills against each other, and adding up all the iterations and changes that have affected each and every skill but may not have affected tuning when those changes were made. This was pre-Update 10, and the numbers specified in the sheet may not reflect the final changes that were made or the most recent build, but still gives an idea of the internal mechanisms of the team when the process was ongoing.
In the future, as the Synergy Milestone bonuses are implemented, further changes to the skills will be made. This includes usability, animation changes, mechanical changes - any skill that isn’t up to par and fun to use is subject to be changed more.
The primary tool that was used for this tuning pass, which some comments in the document mention, is a new global database of each skill. The database breaks down each skill’s mechanics, attack speed, benefits, restrictions, area size, costs - everything about the skill is broken down into categories. Each category is given a damage factor weight, so all skills that are similar in nature are similarly tuned as a baseline. From there, individual tweaks are further made, but it is important to start from an even base to know how much better or worse any given skill is compared to other similar skills.
Designer’s Notebook
General: All base rank 1 (not scaling per point) damage values rounded to nearest 0.1 (10%), to make comparative analysis easier on players at first glance.
Forged
[h2]Overall Notes[/h2]
Mostly minor adjustments for Forged, with the only real big changes being to Rapid Strike and Rapid Fire. According to Discord reports, Forged was considered a little tough to level, but about middle of the pack when the full build settled in. These changes will help to bring up Forged’s early game power, make its core skills more comparable, and improve his non-Vent AOE clearing a bit.
[h2]Moderate/Major Adjustments[/h2]
[h3]Rapid Strike: 90% (was 27%) [/h3]
Large buff here - the skill takes over 1 second to complete and hits 4 times - compare this even to Forged’s own Rapid Fire, which hit over 6 times per second (50% more hits). Rapid Fire was almost double Rapid Strike’s damage per hit despite being ranged, having a 100% damage bonus built in, and hitting 50% more times per second. Though Rapid Strike was more Heat efficient, we’ve declared the Heat costs to be less potent for Forged due to the prevalence of Vent skills (Forged can bypass his resource restrictions like no other class can).
[h3]Shotgonne Blast - 100% (was 54%) [/h3]
Comparable in heat cost and attack speed to Uppercut, but dealt almost 1/5th the damage. AoE’s should be pretty weak by comparison to single target skills, but I feel like this is too much of a penalty, and the formulas back it up. If Shotgunne Blast is still determined to be too much damage at 90%, it likely indicates our overall multipliers for AoE skills need to go down across the board.
[h3]Cyclone Mode - 30% (was 5%) [/h3]
Mechanic Change: Increase heat cost to 10, scale it down with points. The tuning change for this is closely related to the heat cost adjustment - the tick rate for the heat cost is nearly 10 times per second, so this makes Cyclone much more potent at the cost of quickly raising your Heat. It has a custom tuning adjustment specifically for the advantages of moving while attacking, and this is accounted for in the tuning. This will make Cyclone an effective skill in certain builds, but the detrimental massive heat per second will not make any sort of ‘whirlwind-forever robot’ ever possible as a core build early on in the progression.
[h3]Rapid Fire - 40% (was 42%)[/h3]
Mechanic change - Change buff to cause Damage Multiplier instead of Damage Pct - this is an accurate case where we are trying to reward the player with double damage in exchange for moving as little as possible. If this is in the same layer as the additive Items’ Damage Pct, then as item power increases, the reward the player gets for moving very little is reduced, which can create adverse effects in endgame content (reward no longer being worth standing still as incoming damage increases in more difficult content). Additionally, reduce the duration of the buff effect to 3 seconds as it’s far too easy to keep it at 10 stacks currently. Using Holy Bolt as comparative analysis - Rapid Fire will begin outclassing Holy Bolt in terms of DPS at about 5/10 stacks and will remain outclassing it until the stacks are lost. The two have relatively similar effective use times and can be directly compared, factoring in Light Buff against Venting.
[h2]Minor Adjustments[/h2]
[h3]Ramming Robot - 200% (was 108%)[/h3]
Calling this Minor because it’s still a very irrelevant amount of DPS gain given the cooldown, even if the WD total is nearly doubling. Skills whose primary purpose is Movement (such as this one) get their damage set to a fraction of normal, a massive reduction multiplier. But, even still, with the other factors of cooldown and cost applied, Ramming Robot should get a small bump in DPS.
[h3]Servo-Driven Uppercut - 240% (was 229%) [/h3]
Mechanic change: Heat cost increased to 14 (make it a different cost to differentiate it more from Rapid Strike), and raise the base chance to stun to 20%.
[h3]Slug Shot - 200% (was 153%)[/h3]
Comparing it to the new adjustments for Rapid Fire when fully charged, that skill still deals over 50% more DPS compared to Slug Shot. Slug Shot pierces, and has a lower heat per second ratio, but remains the better choice for a hybrid of single-target and area-clearing in a single skill.
[h3]Toxic Dart - Base 90% (was 102%), DoT 90% (was 102%) [/h3]
[h3]Fracking Strike - 90% (was 66%) [/h3]
Not a terribly large increment, but a fair one compared to other ranged skills. The DPS of Fracking Strike is still not close to comparable to fully charged Rapid Fire, but it is more viable for area clear by comparison.
[h3]Vortex Bomb - 200% (was 212%)[/h3]
[h3]Coal Launch - 70% (was 60%)[/h3]
[h3]Furnace Blast - 40% (was 25%)[/h3]
Dusk Mage
[h2]Overall Notes [/h2]
The two main common outliers that resulted in tuning changes were:
- Relation of a given skill’s mana cost to its damage. Specifically, skills of similar mechanical nature that cost more mana did not seem to have the appropriate increase in damage to compensate. If a 2-mana Holy Bolt and a 4-mana Holy Bolt existed in the same tree, the 4-mana one should deal roughly double damage. It would offer stronger burst potential at the cost of being more penalizing if you miss.
- Shotgunning - Skills that are abusable at melee range due to shotgunning have historically been weakened to accommodate. Melee is not the core identity of the Duskmage, and even if it were, it doesn’t feel good to have a skill that appears to be ranged in nature be tuned to only ever be effective at point blank range. If melee is to be an identity of the character, these skills should have a stronger visible identity as melee, such as very brief shotgun-like range projectiles, or short-ranged conical blasts in place of projectiles. Conclusion: Shotgunning will be removed on skills who’s clear primary purpose is to hit a large number of targets.
[h2]Moderate/Major Adjustments[/h2]
[h3]Unholy Bolt - 130% (was 54%) [/h3]
Tuned accurately at 130% WD accounting for built-in shotgun prevention (Now can only hit the same target one time). Unreliable at a distance, but could hit multiple targets. Ultimately, Holy Bolt is still by far the better choice for single targets at range due to its better damage and reliability, but Unholy Bolt should be a fine alternative that sacrifices some single target output for a little better area clear. The previous tuning set made Holy Bolt essentially be all 3 Unholy Bolts hitting the same target (165% vs 54% x3), for roughly equal mana cost and attack speed, making Unholy outclassed completely.
[h3]Absolver - 220% (was 33%) [/h3]
Mechanic change - Reduce mana cost to 3. Make not shotgunnable. Even if the former version was shotgunned (all 3 bolts hitting one target), it was 2/3rds the strength of one Holy Bolt (165% vs 33% x3), and costs twice as much mana. It needed a significant bump to get in line with its cost. It now still costs 2 Holy Bolts, deals a little over 1 Holy Bolt in damage, but still is a better choice for AoE clearing. Holy Bolt remains superior for single target because of its far better efficiency in mana vs damage.
[h3]Energy Spike - 60% (was 10%) [/h3]
Mechanic change - Only create one missile. As it stands, it hits an enemy directly in front of you roughly 6 times, thus the damage has been tuned to assume 6 hits (players abused this when it was tuned assuming fewer hits, and now it is not used at all). We are modifying the power to be less focused on immediate point blank damage, so it can once again be a more reliable ranged attack. At 60% WD, it is expected to hit an enemy in reasonably close distance 3 times, or a further enemy 2 times, putting it a moderate distance behind Holy Bolt in terms of raw damage per hit, but at a cost of lower attack speed, more mana cost, and the ability to pierce.
[h3]Entropy - 90% (was 52%) [/h3]
The tuning for this is assuming 5 hits on a given mob, plus the fact that it is AoE, plus the main missile that also damages and pierces. The reason this still has a damage gain besides that is mostly the mana cost - it costs more mana than almost any other offensive skill in the Dusk Mage’s kit.
[h3]Radiant Blast - 210% (was 100%) [/h3]
I suspect this was tuned low as a result of a broken endgame build for area clearing from the Discord community, which was related to the number of projectiles + the shock affix causing a massive cluster of damage packets. The base damage should be comparable to other skills of the same class, including its moderately high mana cost and low attacks per second. Conclusion: Instead of a cluster of missiles, cause the mob to ‘explode on death’ with a single damage packet and avoid the shotgun chance altogether.
[h3]Light Spear - 280% (was 382%) [/h3]
This was likely tuned too high as a result of its coming late in the tree and being compared to skills with points in them already. At 380%, Light Spear cost 3 Holy Bolts of mana, but also dealt close to 3 Holy Bolts of damage, plus piercing, plus healing. It was a pretty obvious outclassing of Bolt if you made it your main damage skill. Worth noting here that the base damage will go up again once the ‘high tier unlocks get higher base power’ system is in place, which is decidedly separate from this pass. This tuning puts it at costing 3 Holy Bolts, deals a little less than 2 Holy Bolts of damage, but also has utility and AoE mechanics to compensate.
[h3]Spirit Well - 20% (was 40%) [/h3]
In analyzing the total damage provided by summons, factoring in things such as total uptime & expected actual uptime (their health vs incoming damage), it was apparent that even at 40% WD, these summons provide a very large amount of damage, more than any other layer skill. They attack twice per second and there are 6 of them - in the case of them fighting a single target, that was 400% WD per second, or just over 4 Dark Spears layered on top of each other. They should still outclass most layered damage sources, but even at 20%, this skill does a lot of damage over its duration and simultaneously is a large boon to survivability.
[h2]Minor Adjustments[/h2]
[h3]Holy Bolt - 180% (was 165%)[/h3]
[h3]Dark Spears - 140% (was 104%) [/h3]
Mechanic change: Reduce cost to 3. This is a hard one to tune accurately when it costs so much of the Dusk Mage’s total mana (half the bar). The way we tune Damage over Time skills is to assume the skill is not DoT but up-front, then divide the damage out over the entire duration to get the damage per second, then double that result to make DoTs start “paying out” at 50% through the duration. We also factor in the expected uptime, which is low for Dark Spears since it has both a cooldown and monsters can walk out of the area. At 5 mana, Dark Spears would deal even more damage. This may be a case for Dusk Mage’s mana-to-damage ratio needing to be lower, but we’ll see how it plays out.
[h3]Holy Fury - 290% (was 280%)[/h3]
Railmaster
[h2]Overall Notes[/h2]
Railmaster uses a unique formula adjustment that is only placed on the Railmaster’s Endurance-spending skills. This formula adjustment treats these skills more like cooldowns than the equivalent non-cooldown spammable skills from other classes. Spammable skills use their attack speed as a predominant factor in DPS calculations, which Railmaster should not. For example, Ghost Train has no cooldown, but its attack speed has relatively little impact to how much damage the skill should deal, as the Endurance gate is far more impactful than the attack speed gate. As a result, most core Endurance spending skills are recommended to go up in damage. The train cars generally stay the same or are reduced in power, but this is an easy variable to change in the tuning formula if the train wants to be highlighted more.
[h2]Moderate/Major Adjustments[/h2]
[h3]Pound - 300% (was 156%) [/h3]
Consuming 40% of the Endurance bar justifies a very large increase to this skill’s damage. It can be used only twice in a row consecutively without then needing to use only Basic Attacks or Cooldowns, unlike most core skills from other classes. If this was reduced to 1 Endurance, it would deal slightly less than the current damage. Chance to Stun and the small area is accounted for already, but Railmaster’s innate incredibly low Endurance regeneration compared to other classes warrants his spenders being harder-hitting when compared to other classes.
[h3]Blasting Charge - 110% (was 62%) [/h3]
Costing the same Endurance as Pound, Blasting Charge would now deal approximately half of Pound’s damage, accounting for the fact that it is Ranged, has a larger area of effect, and applies a Vulnerability effect. Vulnerability is a rare debuff not found on many skills, and is quite potent to overall DPS on longer engagements, but offers little advantage for general area clearing. If Blasting Charge is considered to be too strong, bringing up the value of Vulnerability in the formula would be advised so it affects other skills similarly.
[h3]Lantern Flash - 40% (was 125%) [/h3]
As with most DoT effects, this receives a sizeable nerf, but in Lantern Flash’s case specifically, it was significantly better than comparables like Dark Spears. Unlike others, Lantern Flash always delivers its entire effect (is not a ground Hazard). Combined with the large area and Stun, this is a justifiable damage reduction by the numbers. It was an incredibly high damage source before, for little opportunity cost, with built-in high utility.
[h2]Minor Adjustments[/h2]
[h3]Ghost Train - 290% (was 222%)[/h3]
This tuning puts it at almost exactly 1 Pound, but costs 2 Pounds worth of endurance, and hits a much wider area. Spending your 4 Endurance on Pound is more than doubling your DPS if you don’t need to hit that many targets, which seems like a fair balance given the Ghost Train’s size - each has its own merits.
[h3]Spike Drive - 150% (was 125%)[/h3]
[h3]Spike Throw - Base 40% (was 33%) Bleed 60% (was 80%)[/h3]
[h3]Hammer Spin - 140% (was 104%)[/h3]
[h2]Train Adjustments (Both Minor and Major)[/h2]
[h3]Cannon Cart - 20% (was 27%) [/h3]
If all Train Car passives need increasing, there is a custom tag to represent the unique mechanics that is the train. It is currently tuned conservatively but can be changed as desired, affecting each of the cars the correct amount to keep them all consistent with their individual mechanics vs damage effectiveness compared to each other.
[h3]Flamethrower Car Passive Base 0% (was 10%), DoT 10% (was 10%)[/h3]
We should consider changing this to just apply a non-stacking DoT effect during its passive. The flamethrower tosses out a ton of missiles, and the damage can vary wildly based on how many of those missiles actually caught enemies. If it just applies the DoT effect, then it doesn’t matter how many missiles connected, just that one did. It will make this car much more consistent in the damage it deals.
[h3]Mortar Cart Passive - 60% (was 101%) [/h3]
Comparable here is that the Mortar Car passively fires half as frequently as the Cannon Cart, and the Cannon Cart fires 3 times per activation, or 6 shots in the same time Mortar fires one shot. Mortar deals half as much damage, but hits more reliably and in a medium-sized area rather than a single target.
[h3]Shotgonne Cart Passive - 70% (was 50%) [/h3]
Comparable here is Mortar Cart. The two attack at about the same speed, with Shotgonne being slightly faster, hitting about the same number of targets per attack, but Shotgonne requiring the train to be a little closer. Similar DPS, with Shotgonne getting a slight edge.
Disclaimer for Train Car Active SkillsThere may be an argument for the Train Car actives to have a unique tag on them just because we want the Train to be better because it’s a highlight of the class. This does come with knock-on effects - if the Train is too good, it can start to become the dominant build, making Hammer users fall behind. But, there is definitely some wiggle room here and this a welcome place for a unique modifier if these nerfs are too problematic, which could be the case.
[h3]Flamethrower Cart Active - 30% (was 50%) [/h3]
This thing fires twenty two missiles, and that’s not counting an extra 2 second DoT it applies. It’s really inconsistent in the amount of damage it actually deals in practice, and is very difficult to tune as a result. Conclusion: We should rework the Flamethrower Cart’s active to be a single burst cone in a direction, instantly applying its effects, and tune it again based on the final version of the new skill.
[h3]Mortar Cart Active - 50% (was 101%)[/h3]
101% per second for 6 seconds is a very high DoT effect for having little to no opportunity cost. As with other DoTs, we calculate out the entire damage of the skill given its opportunity costs and other factors, divide out by the duration, then double it at the end. If DoTs in general are decidedly too weak and need to hit their payout mark before 50% of the duration, we can tune that easily on a global scale.
[h3]Shotgonne Cart Active - 310% (was 300%) [/h3]
[h3]Shocking Rounds[/h3]
This is a tricky one, to the point it’s not worth throwing a number out. The Cannon Cart deals significantly less damage than a given Railmaster cooldown, and with the design as it stands, those two numbers are conflated together. The passive damage being given to the Cannon Cart is +100% WD layered on top, meaning 1 point in this skill by itself increases the Cannon Cart’s damage by a factor of 4x even with the Cannon’s current tuning (27% with another 100 added on top). Conclusion: Rework this skill to split the passive, making it deal ‘Cannon Cart’s damage to another nearby enemy’, and when you activate it, then you get a different buff effect that can deal more equivalent Cooldown-based damage, like an extra +100% per hit. There are other variants available here but ultimately I think this one needs to go back to the drawing board in order for it to be properly tuned and balanced as both a passive and an active.
Sharpshooter
[h2]Overall Notes [/h2]
The main tuning result for Sharpshooter seemed to be Ammo cost in relation to damage.
Sharpshooter’s cost varieties from skill type to skill type have changed in iteration many times to get the fun factor into the right place, but I suspect these changes did not have associated tuning changes to accompany them. Though we factor cooldown into suggested damage formula, mana cost is given more weight, so the general pattern of Sharpshooter’s changes is that ‘big mana spenders and core mana spenders get buffed, non-mana cooldowns remain the same’. The philosophy behind this is that while mana cost and cooldown both speak to a given skill’s usability, skills with mana cost can be thought of as sharing one giant cooldown, where cooldown skills do not.
[h2]Moderate/Major Adjustments[/h2]
[h3]Tight Grouping - 240% (was 107%) [/h3]
Best guess here is that this was tuned based on the Interrupt window, not noticing the SkillCooldown notify that massively lowers the attack speed - an uncommon setup, only seen on a couple of other skills. The base damage of this suggestion seems high when comparing it damage-for-damage to other Ranged starter attacks like Holy/Unholy bolt, but the attack speed of this skill is close to half of those two. It’s a very slow attack speed, thus needs more base damage. This does factor in the AoE nature - if it didn’t, this would be closer to 300% damage.
[h3]Targeted Strikes - 140% (was 32%)[/h3]
When factoring in Attack Speed and Resource cost, Targeted Strikes is about comparable to 2.5 Holy Bolts fired in succession. The new damage for Targeted Strikes puts it into a comparable place, factoring in the extended contact time, occasional chance to miss by nature of the design of the skill, and the damage bonus on subsequent shot mechanics.
[h3]Heartseeker - Base 300% (was 256%), DoT 50% (was 52%)[/h3]
This is a massive buff proposal to the base, but Heartseeker has a very long cooldown and very high mana cost, which makes it want to proportionally be an extremely heavy-hitting skill. The tuning for this in the data takes the total amount of damage for the skill and splits it 50/50 between the base and the DoT portions, and that ratio can be altered as desired. This may warrant a special case tuning adjustment for Heartseeker calling out “a cooldown and ammo cost should not be gaining both damage modifiers to the same degree”, but as it stands, this would be the correct tuning-to-opportunity-cost for this skill.
[h3]Goblin Legion - 30% (was 50%)[/h3]
See Spirit Well notes. Summons are just really efficient and effective and when their damage is calculated over the duration/uptime of the skill, counting incoming damage to them, a reduction is warranted.
[h3]Rizzi's Fate - 70% (was 100%) [/h3]
While this at its current tuning is directly comparable to Dark Spears, Dark Spears costs nearly half of the Dusk Mage’s mana bar. Resource costs are (and should be) worth more base damage than a Cooldown. Rizzi’s Fate is also moderately larger in radius than Dark Spears, which affects its damage slightly.
[h3]Sacrifice to Goose - 40% (was 75%)[/h3]
Three-Fourths of this skill’s damage is located in the additional procs, and like Rizzi’s fate, has no mana cost. Comparing it in total to other similar skills, 300% WD plus the very high vulnerability amount was too high, even factoring in the cooldown. If we wanted to make this more focused on damage, splitting the damage values between the base and the procs would be wise, not having them be the same damage.
[h2]Minor Adjustments[/h2]
[h3]Scattershot - 180% (was 192%)[/h3]
The scaling damage bonus of this needs to go up to +200% to account for the charge time to get to max being equivalent to 3 click-and-release Scattershots. Currently, holding this skill is significantly less DPS, though at better mana efficiency. I would argue a skill that takes this long to charge, dealing exactly 3 spammable shots but at the cost of 1 is a fine starting point (equal DPS but more ammo efficient).
[h3]Onslaught - 120% at 5 Ammo (was 125%), scaling to 420% (was 250%) at 20 Ammo[/h3]
The ammo-to-damage ratio here was not reflective of eating the entire bar, just from a simple math perspective. Eating 5 ammo should be significantly less effective than eating 20 ammo. If we factor this skill as a sister to Reload and always assume that Reload is being used in tandem, we can warrant a numbers drop, but these tuning values will make the skill viable for non-Reload users.
[h3]Scout's Bones - 120% (was 102%), DoT 30% (was 30%)[/h3]
