Tree of Savior Forum

Design Breakdown - Skill Points

Last week we’ve talked about the class system on the way the progression is done within it and combining it with the fundamentals of Identity, Theme and Fantasy it should be enough to make classes interesting and viable even for few aspects of the game yet there’s one thing that can break a class, that is the skill management and how points are evaluated. For this discussion we’ll talk about the power of skill points and how they are distributed, what a leveling component tells about a skill design, how attributes solve an agency issue and provide more features and other possibilities to expand the system.

Skill points have value relative to the skill yet some may not notice how that can be designed to get players to always pick certain skill and even maximize them, that isn’t a shady practice nor anything new in game design as the goal is to establish a foundation for players to work with but it may also lead to death of choice. To understand the values of skill points we must first know the nature of scaling and have some way to measure their relative power, it can’t be done with absolute precision as many factors rely on the context the skill is being used ranging from null in the worst conditions to extreme in the most favorable scenario. Don’t forget that the development team is subject to limitations and some ideas may not work due engine, practicality and balance issues. Sometimes it isn’t IMC’s fault.

Scaling and point allocation

Even classes being the first choice in game skill points are the first decision that comes with an instantaneous payoff, since class doesn’t matter without them, as the other ‘soon to be removed’ is stat points don’t have value on their own and need a large quantity to be noticeable, that said skill points will be the defining factor to distinguish two identical class builds and make them unique. In theory that would happen as a natural response of skills having different high and lows yet for that to be true they must sell their unique effects as part of the skill value so players can pick based on that, the elements responsible for choice are components, all the individual effects that said skill has, and scaling.

To discuss scaling it’s interesting to understand the one point phenomena first which isn’t exclusive to Tree of Savior, one point wonders can be found in any game that care lacks attention to some skills design or done strategically as point dumpsters which can be seen when only one of the skills within the class has such trait. Any skill that is leveled from 0 to 1 will have a leap on efficiency as it goes from nothing to something instantaneously and after that the increments will be added in a linear progression increasing in +X per invested point, it doesn’t mean all skills progress linearly as some scale with stats, others have multiple components affected by level and some even have their effect to be stacked. One pointers are born in this gap that some skills has as they provide all components straight away and usually focus on damage as the sole leveling factor, thus the only element to compete with other skills it damage and it can’t be paired to pure offensive skills.

By tracking down a wide range of skills we can find some golden ratios regarding skill value based on the starting level so is the case that you can pick skills yourself and the values on the chart below to do the test, from a conceptual side it is obvious that such behavior will be more efficient towards skills with higher base output, attack multiplied by overheat, numbers as they always increase in percentage.

Type 1 per lvl 1 total 2 per lvl 2 total 3 per lvl 3 total
Circle 5.5% 77% 7% 63% 12% 48%
Tier (ToS 2.0) 3.33% 46.2% 5% 45% 10% 40%

A point spent into a tier 3 skill adds the same effectiveness than 2 points in a tier 2 and 3 points in a tier 1, that addressing the raw value alone, which makes it easy to notice a huge cost benefit advantage towards higher skills without even taking in consideration their better components, range, ratios and execution. By default almost every offensive skill has the growth peak at the very start, circles almost double the value while tiers doesn’t even get to half, and this makes them more useful to be unlocked only as the investment is too low for the output and never invest on them as any further significant improvement. This tiny element, known as base value or floor, is responsible to allow players to leave their skills in any level they want, it makes skill unlocking greatly effective cost benefit wise and doesn’t punish those that don’t plan to maximize certain skills.

Unfortunately that can’t be said about buffs as they either fall into the full effect with increasing duration, which has variable value until refresh time point as any excess has zero value, and scaling effects, which often lacks floor making each point invested have 100% efficiency compared to the starting level and that can provide variance between 400%~1400% based on circle/tier. For the Rank-Circle system this weighted a lot as players had lower caps to work with and each extra circle adds 100% of the introduction circle limit, in mere skill points value a Thaumaturge 3 maxed Swell Left Arm has 300% of the attack power of a Thaumaturge 1 and since their buffs don’t stack it has 0 value in this scenario. That’s also the case with selfish buffs such as Gung Ho and Concentrate as they don’t offer enough at the start, especially as they scale on top of basic stats, and don’t feel significant at all unless taken way mid to high levels.

As a matter of fact classes with great buff skills tend to be more popular and their value is multiplied to the party, those that couldn’t go deeper into a class to have those points as null unless said class is absent yet the option to not maximize isn’t available as the skill growth is so wild. If a buff has competitive nature and no floor it tends to end up with virtual value of zero unless maxed, this is the case with all Pardoner sellable buffs as they do make all these have zero value on skill points but that’s more of a mechanical issue than scaling itself even tho it would have less of an impact if they had floor. By adding a floor and decrease the leveling power, maintaining the same output when maxed, skills can have interest ranges instead binary distributions and and only lose the exceeding cap values yet it won’t diminish these skills since buffs have persistent effects that empower the character as a whole.

Another element to be considered is the overscale as that changes the growth from linear to exponential and that happens to all skill that have convergent components increased with skill level, each element enhances the next and the point value is greatly amplified. Cryomancer’s Subzero Shield is a great example as it scales on duration, freeze chance and hit count when leveled, at level it can only freeze up to 6 enemies at 15% chance under 18s but when maxed it goes up to 60% chance on up to 15 enemies over 45s - such is the case that the the transition from 9 to 10 is 3x the value of 1 to 2.

In this case in particular is interesting to point out that the skill has an outside scaling as well since it requires Block to be activated which means the skill also scales with the character stats and level in bigger snowball increasing or decreasing the point value beyond the skill itself. Overscaling in itself isn’t a bad feature but it has to be used carefully to prevent the must max scenario, that may not be the case with minor elements such as target count/AAR which can be applied to lower circle/tiers and increase their value by the time advanced skills are introduced.

When the base value, circle/tier growth and absence of floor are taken in consideration it seems like the choice in skill distribution may not be that free, later skills have better growth and are a best investment, skills that provide special effects seldom follow up on it and those that do are the ones more likely to be seen in more distribution ranges instead of the binary state. The natural outcome makes those with better availability, execution and biggest numbers get on top with no hassle unless confronted with skills that have special mechanics or have overscale. It’s important to question if players would ever sacrifice points on higher skills for basic and mid skills, real choice is provided when the is exchange is interesting or beneficial in a context.

At last one key change Re:Build introduced to the game was the character level based SP cost, in the old system leveling a skill would cause the SP cost to rise along the skill benefits but this doesn’t happen anymore. By removing having this element to be static towards skill level it makes skill distribution more hostile to point scattering, after all it is more convenient to have that heavy SP consumption to generate highest output as skill with few points can be as costly.

Skill’s heart

Identity can also be applied to skills and as such each one has a different thing to contribute to the game even if they have the same outcome, if that wasn’t the case it wouldn’t be possible to have that many offensive skills that feel different. The easiest way to spot is to track all components of a skill and check which of those scale with level, aside damage, as that is the most important element along the skill mechanics.

As an example we can use Quarrel Shooter’s Scatter Caltrops since it has damage, object count, duration and slow as base components, it is by all means a trap skill, as does nothing until an enemy interacts with it, that also knocks down the enemies hit making the only way to use aggressively is by luring, which is quite ineffective. The mechanics point out that the skill was never intended to be used that way which is quite ironic as they’re both damage and object count, not only the skill overscales but also has no floor for caltrop.

Caltrops are meant to control the terrain and prevent enemy movement be that by targeting the enemy directly and lock it in the position or block a path completely, the skill doesn’t prevent ally movement as walls do and enemies have the choice to break through if they want to. With that in mind the object count does the job as the defensive component and the extra damage punishes those that dare walk over it, if the real purpose was to serve as an offensive skill the damage could be redistributed to be higher on early levels and have a fixed count for caltrops but that doesn’t appeal as much to those that want to keep distance from their enemies.

As a thought experiment we can take Peltasta’s Guardian launch version as it increased defense at the cost of lowering the damage output, for a defensive oriented use the skill can have the PAtk penalty static and get more PDef when leveled which makes it interesting to be leveled for those that want to be bulkier. However the skill nature changes completely if the PAtk penalty is reduced by skill level, completely mitigated when maxed, and have PDef bonus as static, in this case those that want to be tanky can take it at low levels while damage dealers will prioritize to get it at the highest values or simply ignore it if the point cost is too high. The current version however has no defined audience as the penalty was removed, instead duration does the job as the leveling component and all player that took the class aim for the refresh level as anything past that is pointless.

That said not all skills have that flexibility to promote their unique components which is the case with most of debuff and control skills, those can trivialize enemy treat and allow players to slay enemies without drop a sweat. Effect efficiency isn’t always an option as it can easily overscale making investment mandatory, it isn’t impossible to balance out in a way it can be dropped at any level but if their value is multiplicative it usually backfires. Duration however is less impactful but depending on the effect power but still can provide unwanted results, not to say it may have to be subject under 1s per level and/or abide the floor and neither feel that satisfying to level. An alternative to that, specially to control skills, is to expand the magnitude of the skill by applying to more enemies as this feature that is great in crowded places yet useless against few enemies, that’s further supported by the default AAR mechanic as each 3 points can vary from 1 to 3 additional enemies and can be implemented as +1 AAR per invested point. It can be a small trait to be ignored by some portion of players yet it can enhance the utility nature of those skills significantly, after all it’s more interesting for a stun skill to get one or two enemies more than competing for damage.

Death to passive skills

Games that have passive skills often deal with an issue regard point distribution, it’s a basic matter of the nature of passive effects as they’re applied all time with no cost. This means a point spent on a passive has a permanent effect at the same cost of an active skill, depending on the execution, value and bonus it can outvalue active skills. When passive skills are in the same sphere as active skills we either end up with them being too good that dominates point distribution or lackluster and be completely ignored, both are equally bad.
The other aspect of passive skills that is often seen as bad design is how they lack engagement to players, as much as having +X Crit Rate for the whole game is great it can also be completely forgotten as stats become invisible the more efficient they are.

Buffs may be close to passives but they often apply to party, have uptime and downtime or explore elements, mechanics, limitations and penalties that wouldn’t work that well as passive effects. Oddly some buffs have simplistic effects that aren’t enough to set them as full skills, this is the case with any simple self buff that is always uptime and has short cooldown. If a buff is a stat boost effect only, is always available and can’t affect a party it doesn’t need to be delivered as a buff, that said it

The way Tree of Savior deals with those passive skills is clever, they take silver instead of skill points in the form of attributes we know today. With a simple context change players can always have the passive effects by spending their money and focus on active skills as their real choices, it takes away the issue of power scaling from these elements as any player that has the class will be subject to it. That said there are some limitations as requiring a certain Circles or base level as in the upcoming update yet that doesn’t diminish the importance of such division, in fact they did a great job by applying this concept into skills to give additional options and features.

Despite the attributes fail to provide tolls when used since most of them only increase the SP consumption by a bit and their effects are great overall, it is a kind of penalty but not in the same sense as having cooldowns increased or one component to be disabled. The best way to do attributes is by giving them functionality changes or the classic enhancement, even it benefitting those that maxed the skill more than anyone, as additional effects, at least for 2.0, with the current setting won’t provide reasons to not be taken.

Bindings and groups

With the Re:Build changes some features can be added in order to spice up the way points are taken but before that we have to take a look at Necromancer’s skeletons. As the highest tier reward the class received Raise Skull Wizard and all the skeleton summoning skills were changed to share cooldown and overheat (may have been changed), this is a brand new addition to the game mechanics as it binds three skills into variances but take them as a single unit. With that in mind it’s possible to expand that concept into the skill distribution and introduce conflict unfitting for Rank-Circle, that is to bind skills into groups and have those capped with a maximum point count.

For that let’s simulate a random class with 3 random skills named A, B and C within bound inside a group, they have their max levels as 15, 10 and 5 respectively and the group can only support 20 points. That said if a player decides to maximize skill A it can only spend 5 more points inside this group and that’s enough to max C or get B to mid levels, for a more distributed arrangement it can get A to 10, B to 5 and C maxed. Even if B and C are maxed players can decide to take not fulfill the group point count and spend them elsewhere, that said it’s also possible to have more than one binding group inside a class be that by taking less skills grouped or the presence of more skills within the class.

It may sound stupid to bind skills but this conflict can be quite important to add variance to skill distribution that won’t happen spontaneously unless the skill balance is done perfectly. Just imagine having the two best skills of a class bound together in a way only one can be maxed, would it be best to master one of them and drop the remaining points on the other one or balance those two in an unusual and more equal distribution to extract more of both. In a way it can also work for other classes as players will need tools outside a class bubble, of course that has to be done carefully to not become a plain frustration.

Free skills

Historically there have been skills that don’t really fit into satisfying balance state that allows it to have multiple levels and have invested points felt like well spent, which often are close to or full essential features to play the class itself. Just to name few that is the case with Warlock’s Evil Sacrifice, Cannoneer’s Bazooka, Falconer’s Roost, Necromancer’s Disinter and Gather Corpse. It isn’t fair to force players into certain skills that can’t be used on their own but are also a vital component of the class itself, Unbind is the best example on that group as it does nothing aside undo other skills of its class and in most cases isn’t that needed yet players have to take it for the 1% times the skill is required.

However classes are the exception as they own free control skills even tho that falls into the same category, in the old system that could be used as a way to promote some Circles depending on their relative power but in the current one there’s no reason to not give them away for free. As a matter of fact all single point skills are seldom ignored due the low investment they have so making them free can be a great way to have classes more satisfying at no real cost, it is a small trait for something players will get them anyway but won’t take require their precious points that could be spent somewhere else.

With the new system it is quite easy to deliver some skills for free and it can be done in different ways depending on how important they are, among the easy options it can be done by using the base level, class tier or even by getting more classes. Let’s take Hand Knife as an example, only usable after Palm Strike, as the new mechanic can allow it to be locked away when Monk is taken as the starting class, then once this character gets the second or third subclass the skill can be handle out as a gift.

Of course some can argue that those traits can be too much of free power provided to certain classes and have the others to stay behind but skills that are part of the core experience can’t be taken in the same level as skills that can scale and have value on their own. It’s important to not forget that these are either follow ups, mandatory setups or punctual gimmicks.

We can’t forget that Ram Muay and Double Gun Stance are both vital requirements for their respective classes to work and, slightly delay it, yet are not that interesting to level, for Bullet Marker some skills are still available to use but Nak Muay is completely in dark without Ram Muay. Helmets that can be delayed as they aren’t as important in the earlier stages but still offer some nice effects to the user.

Helmets

In the Rank-Circle system helmets were valuable buffs that balanced out earlier Circles, for Murmillo it was a way to even out Peltasta’s and Rodelero’s attack to pair late game damage as Dragoon has it own skills enhanced. However neither is required in the in Re:Build as all classes have their power at fullest and are subject to the same power ceiling, Plague Doctor’s Beak Mask is the only skill from this line that still makes sense as it works with a new mechanic instead of damage multipliers.

That said there’s no reason to keep them especially when a considerable part of the players is unpleased with their visuals but are forced to use them for the effects, if those buffs are that important it would be better to change them into attributes, weakened to compensate, and apply those effects whenever a cosmetic helmet is equipped. With that all the skill helmets can be changed into equipment either provided upon class change or through quests, they may even be storable so other characters can use them for the aesthetics. As a major gain players can use other helmets, such as Velcoper, and benefit from those buffs, it won’t overwrite the cosmetic anymore and still be readable in PvP.

Transformations

There’s a group of skills that operates as transformations, those overwrite hotkeys temporarily to use monster skills. As a whole it ain’t a bad mechanic but it isn’t practical either, the key targeted are A to F~G and for the whole skill duration players can’t use the real assigned skills or items. The only reason this is being brought up is because they can fit way better on ZXCV (and the transformation key to undo) as the attack commands don’t work properly and end up clunky when they do, in addition it would be interesting to have the new skills displayed on top of the hotkeys with name, damage ratio, attack property and cooldown so players can be properly informed in game.


With that all the guidelines, and some pre-set comments, we should be done for the setup of following classes Design Breakdowns, as a perhaps lighter discussion next week we’ll talk about weapons and their identity in game, or lack of. There might have some things that may not be as clear and for that i ask you guys to point out so it can be improved, other comments regarding the discussion and even some ideas on how to improve it are always welcome. If want to show your support for this kind of thread but don’t know how to engage in the discussion a simple like can do the job.

3 Likes

What about Crafting skills and shop skills? Shouldn’t they be part of your skill point analysis as well?

1 Like

It’s better to take them individually as they often come with different mechanics (like Magnum Opus and Enchant Armor) that doesn’t allow them to be taken as a group in the same way as offensive skills, buffs and stances, meanwhile Transformations and Helmets are fewer in the game but still share common the same core traits that can be addressed with a single solution (just a quality of life change).

For Repair, Tincturing and Spell Shop the same concept of floor can be applied since will push players to maximize skill due effect scaling when the skill itself could be enough (only preventing early levels to be useful). I’d add them into the thread but i couldn’t find a fitting approach that won’t have them to be discussed in details.

This topic was automatically closed after 60 days. New replies are no longer allowed.