Skill specialisation: lock two slots, 19 classes in a table
✍️ Alice・愛麗沙
Remember one rule: do not reroll for a full set of three slots - reroll for your two fixed slots. Skill specialisation (the blacksmith’s gear modification) draws 3 of your 6 skill slots without repetition, which gives 20 possible combinations. Hitting one exact triple takes 20 rerolls on average; hitting any triple that contains your two fixed slots takes 5 - the same gold goes four times as far. Korea’s experience in S0, the era before its Season 1 and where Taiwan is now, is that almost all 19 classes follow the same shape: two slots locked, the third decided by your accessory runes. Frost Mage 1 and 5 fixed (third slot 6 or 2), Fire Mage 135, Dancer 1 and 5 fixed (Approach rune takes 3, Freshness takes 4), Priest 3 and 5 fixed (healing takes 2, damage takes 1), Lightning Mage 1 and 5 fixed (6 or 2), Swordmaster 4 and 5 fixed, Greatsword Warrior 145. The third slot moves whenever you swap accessory runes, so it is the last one to spend on. The three splits Taiwan’s forum argued about most all have different answers in Korea: Fire Mage slot 2 is in no Korean recommendation, nobody takes Dancer slot 6, and the Lightning Mage debate is 156 versus 125, not 135.
What specialisation actually does: six slots, three picks, +2, cap 30
First, the terms. What Taiwan’s players call “specialisation” is skill enhancement on engraved gear: every piece engraved with a 4-star or better rune gives +2 levels to 3 of your 6 active skills (elite, epic and legendary alike). The pieces add up, and each slot caps at level 30. Which three slots you get is random; if you dislike the roll, you reroll it at the blacksmith. Korea renamed this function “skill specialisation” in June 2026, so the Taiwanese players’ nickname got there before their interface did. Per the Korean community’s transcription, one reroll costs 2,500 gold on rare gear, 5,000 on elite and 10,000 on epic.
Here is how one slot’s power grows from level 0 to 30, next to the disclosed draw odds:
| Rune rarity | Skill slots enhanced | Levels per slot | Applies where |
|---|---|---|---|
| Advanced | 2 | +1 | KR only |
| Rare | 3 | +1 | KR only |
| 菁英 | 3 | +2 | TW live (official odds) |
| 史詩 | 3 | +2 | TW live (official odds) |
| 傳說 | 3 | +2 | TW live (official odds) |
| Mythic | 3 | +2 | KR only |
Slot count and magnitude depend on rune RARITY, not stars; Taiwan's odds page publishes only the elite/epic/legendary rows. Which slots get hit is uniform (16.67% for the first pick).Odds DB
| Skill level | Gain at level | Cumulative skill power |
|---|---|---|
| Lv.1 | +3% | +3% |
| Lv.5 | +3% | +17% |
| Lv.10 | +5% | +34% |
| Lv.15 | +5% | +51% |
| Lv.20 | +5% | +68% |
| Lv.25 | +3% | +83% |
| Lv.30 | +5% | +100% |
The full 30-level ladder is a KR community damage-model transcription, not an official table; Lv.30 is the cap and lands on exactly +100%.Upgrades
Slot 6 is always the ultimate; slots 1-5 follow the skill bar order. Skills with no damage (heals, buffs) turn enhancement into shorter cooldowns or longer durations, so support classes get real value from specialisation too - it simply pays out in a different currency.
The arithmetic: how many rerolls, and what ten levels are worth
Reroll combinations. Three slots out of six, drawn equally and without repetition, gives C(6,3) = 20 combinations:
| Your target | Matching combinations | Chance per reroll | Average rerolls | Average cost on epic gear (10,000 gold each) |
|---|---|---|---|---|
| One exact triple (e.g. 1-5-6) | 1 | 5% | 20 | about 200,000 |
| Any triple containing your two fixed slots (e.g. 1 and 5) | 4 | 20% | 5 | about 50,000 |
| Any triple containing one main slot (e.g. 5) | 10 | 50% | 2 | about 20,000 |
Chasing a complete triple on one epic piece burns 200,000 gold on average - 1.2 million across six pieces. Locking only the two fixed slots costs 300,000 for all six. The 900,000 you save is enough to push rune engraving up a stage. Fill the third slot once your accessory runes are settled.
What ten levels are worth depends on that skill’s share of your damage. Going from level 20 to 30 moves the skill’s power from x1.68 to x2.00, which is +19%. To get the effect on your total damage, multiply by the skill’s share:
| Skill’s share of total damage | Total damage gained from 20 to 30 | Examples recorded by the Korean community |
|---|---|---|
| 50% or more | about +9.5% or more | Fire Mage slot 5 (more than all other skills combined), Greatsword Warrior slot 5, Longbowman ultimate at about half |
| 40% | about +7.6% | Swordmaster slot 4 at about 40% |
| 10% | about +1.9% | a filler skill in the rotation |
The same ten levels are worth five times as much on a main skill as on a filler. That is why “two fixed slots” holds for nearly every class: the damage is usually carried by just two skills.
All 19 classes: Korea’s S0 consensus, fixed slots, and the third-slot rule
Taiwan’s content sits at roughly Korea’s late S0, just before Korea’s Season 1, so this table only uses Korea’s S0 community consensus. What changes later is covered after the table. Slots are numbered in skill-bar order; slot 6 is the ultimate. Where a skill is named, the name is an English gloss of the Korean community’s term, not an official English name - there is no English service.
| Class | Korea S0 consensus | Fixed slots | How the third slot is chosen | Confidence |
|---|---|---|---|---|
| Frost Mage | 156; 125 if you skip the ultimate | 1, 5 | manual ultimate user takes 6; no-ultimate North Wind build takes 2 | High |
| Fire Mage | 135 | 3, 5 | Korea: “135 whatever runes you use” | High |
| Lightning Mage | 156 or 125 | 1, 5 | ultimate burst takes 6; everyday rotation takes 2 (with Force Field) | High |
| Mage | 124; 126 with meteor on the ultimate | 1, 2 | depends where your meteor sits | High |
| Dancer | 135 / 145 | 1, 5 | Approach takes 3; Freshness takes 4 | High |
| Priest | 235 / 135 | 3, 5 | healing (Wings) takes 2; damage (Bond) takes 1 | High |
| Healer | 135 / 234 | 3 | Vow gives 135; Suppression gives 234 | Medium |
| Monk | 124 / 125 | 1, 2 | whether you wear the Wonder or Wisdom emblem | Medium |
| Bard | 234 | 3, 4 | slot 2 is the second damage source; switch to 6 only for party burst | High |
| Musician | 235; 135 with an accessory that takes slot 3 out of the rotation | 3, 5 | follows your accessory runes | High |
| Swordmaster | 345 / 456 | 4 (Flying Sword), 5 (Insight) | Dancer / Falling Blossoms build takes 3; Flash Strike build takes 6 | High |
| Greatsword Warrior | 145; 245 with the Wisdom emblem | 4, 5 | 1 or 2 by accessories and emblem; 6 only for manual burst; nobody takes 3 | Medium-high |
| Warrior | 124; alternative 134 | 1, 4 | 2 or 3 by accessories (no source backs 234) | Medium |
| Archer | single-target 134 / mob-clearing 123 / raid 136 | 1, 3 | single target → 4; clearing → 2; raid ceiling → 6 | Medium |
| Crossbowman | 234 | 4 and 5 share a level | only the higher of 4 and 5 counts, so one of them is enough | High |
| Longbowman | 136; alternative 356 | 3, 6 | 1 or 5 by accessories | High |
| Rogue | 134 (Toxicity build) / 234 | 3 (ambush) | whether you run the Toxicity accessory | High |
| Fighter | 145 (most players) / 245 / 456 | 4, 5 | slot 5 carries 70-80% of damage | Medium |
| Dual Blade | 156 / 136 / 356 / 456 | 5 | the build names are the accessory rune sets | Medium |
Taiwan’s three splits, and what Korea said
Fire Mage “5 > 2 > 1”. Putting slot 5 first is exactly right - the Korean community describes its damage share as higher than every other skill combined. But slot 2 appears in no Korean recommendation: once you equip the Fire Pillar rune, slot 2 itself is cut in half, and it sits in the rotation to build heat, not to deal damage. Korea’s S0 answer is 135, because slot 3 out-damages slot 1 over a fight and slot 1 gets reset by explosions. That also settles the “Fire Pillar or Heat Wave” question: if you pick Fire Pillar, slot 2 deserves specialisation even less.
Dancer 156. No Korean source recommends specialising the ultimate - the finale “does not need it”. The only split is the third slot: an Approach accessory means slot 3, a Freshness accessory means slot 4. The Taiwan post saying “I have 156 because I could not afford to reroll past 1 and 5” is describing a stopgap, not an answer.
Lightning Mage 156 versus 135. The real Korean debate is 156 versus 125: 156 gives the bigger ultimate burst, 125 gives the higher everyday rotation damage (slot 2’s automatic damage with the Force Field rune). Slot 1, the lightning strike, must be at 30; slot 5 at 30 extends overcharge so you fit more slot-1 casts. Slot 3 is not in any recommendation.
One more common mix-up: Frost Mage 145. Some Korean players did take 145 in S0, but that was a minority route built on the elite accessory rune Winter, which the community sheet ranks below North Wind; the 145/146 that took over in Season 1 needs the Season 1 accessory rune Glacier, and Taiwan does not have it yet. Taking 145 in Taiwan now means putting specialisation on a slot with no damage rune behind it.
Do now, skip for now, and what would overturn this
Do now:
- Find your class’s two fixed slots in the table above; reroll each piece at the blacksmith until the result contains both, then stop.
- Only once both fixed slots reach 30, start spending on the third; the third slot follows whatever accessory runes you are wearing now.
- Do not ignore support classes because they “have no damage” - enhancement turns into cooldown or duration and is just as valuable.
Skip for now:
- Rolling a complete triple in one go. It averages 20 rerolls, four times the price of locking two, and the third slot has to be redone whenever your accessories change.
- Auto versus manual. No Korean source splits its recommendations by auto or manual play; the one consistent rule is that “slot 6 is only worth it if you fire the ultimate by hand”.
What would overturn this:
- When Taiwan gets Korea’s Season 1 accessory runes (Glacier for Frost Mage, Blue Flame for Fire Mage, the halo rune for Healer, and others), several answers flip: in Korea’s Season 1, Frost Mage moved to 146, Fire Mage to 156 or 136, Healer to 345. On that day this whole table gets rebuilt.
- If Taiwan’s reroll cost or its 3-slots-+2 output turns out to differ from the Korean transcription. Trust the in-game figures for the cost column - but “locking two is four times cheaper than locking three” is a count of combinations and does not change.