Tree of Savior Forum

Design Breakdown - Class Systems and Progression


With the previous Design Breakdown covering Theme, Identity and Fantasy we can get into a more visible and practical discussion, the real plan was to have a lightning round to talk about all classes and issues before the system change but almost at its completion i realized a common element among them so why not having it unified in a more interesting discussion.

Tree of Savior had for quite some time one of the most interesting, if not the most, class systems ever implemented, it expanded the basic concept of linear advancement of classes that is seen in many MMORPG’s and pushed it the extreme by having multiple advancements that weren’t related to the previous choice. A player could start as a Cleric, progress into a Priest to protect its allies, release itself from the physical plane with Sadhu knowledge and even become a werewolf by learning the way of Druid, all in the same character. Yet no system is truly perfect and for that we will discuss how each one of the systems, rank-circle and full-dive, and here we’ll discuss it along how progression is handle within classes. 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.

Rank-Circle, commitment and abandon

The Rank-Circle system worked with two core elements, those were progression and specialization. Any game ever made has progression in it and those with class systems do set out some checkpoints so players can feel motivated and be rewarded by playing but here it is a little different, Ranks worked as a ladder that provide at least two new classes once completed. It could be compared to be more in line with traditional systems with the difference being the possibility to pick previous unlocked classes as new advancements or extend the power of picked class through Circles. Unlike the previous element Circles worked as checkpoints in a class itself and is a direct advancement into it and can be taken twice, advancing into circles was equivalent changing into a new class in the ladder structure but instead of a full new skill set it will add a couple more skills and more levels to the earlier roster.

Those checkpoints served as a scale inside classes and allow players to take which degree of each single class they want, instead of going all the way in as most games do. As a major advantage it provided full control on the Identity amount wanted, needed and/or desired for a lower cost, it’s was feature that had building itself as a balancing factor to be considered when aiming for a goal. For those that enjoyed diving into Theme and Fantasy it was the best way to build it by taking only the desired bits, it was all about how much you will commit to a class and when will you move on from it.

With a proper class distribution among ranks it could be possible not only to provide the tools required to cover the basic needs for the whole game but also implement an interactive hidden tutorial introducing skills and mechanics that will be expanded later on, if done right it could be so natural that players wouldn’t even notice. It seems like this aspect was taken in consideration and for that let’s take a quick look at archer’s first 2 ranks and what new elements they provided, there were indeed other mechanics that are explored later on but the basics of targeting are all covered at this starting ranks.

Archer - Single targeting in basic attacks, Oblique Shot (with bouncing as well) and Heavy Shot, ground targeting in Multi Shot, direct party buffing in Swift Step and stances in Kneeling Shot.
Quarrel Shooter - Unit deployment in Deploy Pavise and Scatter Caltrops and directional targeting in Rapid Fire and Teardown.
Ranger - Directional targeting and hitboxes in Barrage and High Anchoring.

As another advantage the system slow pacing helps new players to not be overwhelmed by the class roster as the first split had only 3 options from which two have the element of discovery, it was also implicit that new classes would be unlocked every following rank and those already taken could be expanded. Each advancement brought excitement as new options were presented and that was made fully effective with the low level cap for classes in a fast pacing until mid ranks, of course those arguments applied mostly to those that started playing blindly and other interactions outside game environment could have the opposite effect.

One of the reasons the system didn’t sustained itself is due a blend of misconception and expectations towards the nature of its ladder-like structure - a need for power scaling of higher rank classes. As classes were introduced in specific level ranges their power was paired to the content available in this same range, it’s a natural conclusion given the fact that any advancement is expected to be better and stronger than the predecessor but that is a fallacy regarding ranking, that wasn’t the case with circles as they are higher peaks of power. If we compare a scenario where have build A2 vs AB, the first one can have its skills in higher levels and a couple more skills than AB which can be a considerable advantage yet AB has a wider roster of skills to pick from since it has two different classes, even if A2 is overall 50% stronger it will have less tools to use and have to deal with fewer rotations with more cooldowns opposed to a less specialized variation. If we take most conservative build A3 will have to get a new class at rank 4 while AB may have transited into ABC, A2B or AB2 and can reach a similar power range at this same rank.

It would be possible to balance out the system with a power ceiling and that’s what they plan to do, yet for Rank-Circle it can be awkward when questioning why classes certain classes are unlocked by ranking up. That could be the case if the introduction of those classes is based on harder playstyles, solve issues presented in late game content, prevent crafting classes to be taken in stages they won’t contribute and have behemoths, high damage and higher cooldown with few overheats, locked till they are useful.

Now if we assume that the power scaling was indeed inevitable the other answer is to change the guideline and curve of escalation, taking the base class as 100% each subsequent rank can add 10% more on top of it being capped at 190% as a C3 rank 8 class (subject to target count, damage and other factors). In the latest rank-circle standards, comparing lvl 15 Multi Shot with lvl 5 Zenith (rank 3 skill) in raw numbers the first one, if all 10 hits lands on a single enemy, is equal to 95% of the damage of a single arrow, however if Zenith land hit with all 10 arrows it deals 9.6x times the damage of a full Multi Shot in a large radius and hits multiple enemies - that was the escalation ratio that made the system inviable.

After all the praises it’s time to take the negatives to the spotlight and we have to start with one that is quite visible and bothersome as it was a natural byproduct of such system, the presence of class fillers. As classes were limited by rank they couldn’t be picked before their appearance and it lead to scenarios where no class proved itself satisfying as an advancement yet one had to be taken in order to reach the next rank. One of the reasons this happened was due the class and skill distribution not supporting certain higher rank classes be that by low or none synergy or in some cases weapon incompatibility, however that was mostly an issue with the particular high rank classes were too restricted to work out with the earlier options. Even if such issue were solved on the mechanical side, by shifting class ranks and skills, that wouldn’t do it for the Fantasy as some players builded characters based on it and the system couldn’t integrated them without bending itself.

It’s also important to keep in mind that a system like this need to provide not only an Identity route but also some Relief options in the first two Circles, for many players a class could be picked for what it contributes to the game instead of completely over its Identity which was the case of said fillers and/or strategic choices planned to stop midway. On that matter Relief worked as a safety device within the class to prevent a specific skill distribution to be static in all scenarios and have transitions to be smoother, this was something a significant amount of classes did and some skills such as Blessing could be taken as the primary focus regardless of the build. That said Relief could also be taken as an alternative branch of the class playstyle to be explored since other classes can explore those skills or even synergise with them, it was a good element to be present even being a minor in value, quantity and impact.

For the other instance where the system couldn’t be balanced, and this time with no workaround, we had PvP as a whole. A fundamental characteristic to make a PvP game is to have consistency, once you get in a combat you’ll be put to face real thinking humans and react to their actions but that requires information as fighting without a plan is often not the best idea. Any PvP dedicated game has as one of its core design elements the availability of information, the elements with high variance are often visible so opponents can check them anytime and feel they had a fair defeat when taken out by something completely unexpected such as an enemy smart play. Even then the major elements are often static and know by all players in the match, that means the unpredictability is always created by a player action and not the system itself which is essential to make the rules consistent and fair.

To have it in raw numbers any PvP match in Tree of Savior has the minimum of 3 players and with the most conservative builds each one would had at least 12 skills in a total of 36 skills that had to be recognizable and kept track of, if this same scenario is taken into the wackiest arrangement this number goes up to 30 per player in a total of 90. Since any player could build anyway they want the PvP experience can be quite inconsistent, even after the match is over the next batch will have new players with new builds that you have to learn about on the fly and by the time you can catch up it will be already too late.

That all said it wouldn’t be impossible to keep the system by doing changes on the same scale as flipping it out completely, yet no adjustment would make the PvP factor to be less of a hassle. Also it would be better to cover the synergetic potential of this system but that isn’t really easy to discuss when talking about the class system itself and, despite being ignored on basically every single Design Breakdown, isn’t as appropriated then in class discussion itself.

Full-dive, flattening complexity

Re:Build came to streamline building using classes at full power from start, it is an effective way to simplify the advancement as a whole and makes the systems be more similar to traditional ones. That said it still keeps the same unbound advancement that is the heart of the game. For that to work the restrictions on where to pick each class have to be lifted which can cause new players to be overwhelmed by options on the first advancement, going from three choices to a minimum of ten without the safeguard provided by the second Circle of the starting class.

There are no doubts the system is easier to understand as a whole and may not be as scary for new players once they research about the game online, that said it may not be the case for a blind play as the first advancement is not only quite early in game but also huge load of options to check and analyse. Some restriction on the starting pool may be welcome on the first advancement as classes as Sorcerer, Oracle, Templar and any crafting class aren’t suitable for the first promotion, in fact comparing to the old system all the essential skills have to be crammed in the base class and the system can’t develop the same learning curve present in Rank-Circle which means players that chose their class order poorly can get stuck in a limited class till their next advancement unless they gather enough points to reset.

As a counterpoint the system also breaks out the unlock limitations allowing impossible combinations to the Rank-Circle structure, that said it doesn’t mean the full-dive offers more variety but rather different options. If said change was made keeping the Circles as part of it then it would be truly a more diverse system where players can pick everything on the desired amount, however that’s likely to be more frightening for new players and as complicated, if not more, to balance.

By removing the commitment and abandon features it is easier to handle synergy as a whole and have party and PvP interactions more readable, this implies in builds having some skills granted whenever picked which can also harm skill distribution as certain class combinations can end up always with the same skills in the same levels. Without Circles Relief elements will lose their value as the major need of them is to be an option for those that won’t stay the for the complete class, in fact it may be better for classes to be more focused on their Identities as those can’t be repeated in a system that has everything available from start.

To adapt Circles into full-dive skills are unlocked on the same level range as the old system, it keeps excitement of progressing to a new tier and prevents the best skills to be taken from scratch even tho it’s likely to keep the same point saving feature that allows players to max those skills once the requirement is met and have the following levels to be the same as before. At last the fixed stat distribution that does solves the issue of stat dump the game had for long, as even if the substat effects were greatly improved players would still focus on one or two stats in particular points always provide +1 in the invested stat. However by being fixed it takes an useful expected ratio set that may not fit the character when trying to do something unusual, this is rather true for cleric classes as they have the most diverse stat chart in the game and prevents player to be successful in hybrid builds which was one of the intended directions for the tree. A better way to do it is by having at least two stat distributions to pick from each class or allow players to disable certain classes to be factored on the ratio distribution (even with some penalty), anything else can either turn back to the same issue or be too hard to deal with when class reset is factored in.

The infamous twist

Regardless of the system used there is one element present in the way classes are designed that is the way skills are arranged considering their identities. For a brief moment let’s imagine a bookstore is making a sales deal for recipe books with the following guideline.

20$ - 2 items of your choice, 3 main dish books
50$ - 5 items of your choice, 6 main dish books and 2 dessert books
100$ - 10 items of your choice, 8 main dish books, 4 dessert books, 2 sci-fi comics and a toaster.

It can sound absurd but this kind of thing happens in the game, classes start with one Identity direction and transition into something completely different that is only related to the Theme. That said some classes have been reworked and don’t present the same condition as their Rank-Circle counterparts, yet this can still be seen in some classes such as Squire and Fencer. Some could even argue that the reason the same system failed in the first place was this questionable execution of the Circle progression, however this is a may be a conscious design decision as at least half of the classes that have more of a defined Identity suffer from it. We can prove it by using Inquisitor as an example since it won’t be changed on that matter, it starts as a physical anti magic class that shifts into utility and PvP in the second Circle/Tier and shifts again into crowd killing in its conclusion.

On a way the class does progress by getting stronger and fancier skills but when those aren’t connected to the main reason a player picked the class or the way the class is played, it’s a power fantasy at cost of cohesion and Identity. Dievdirbys is also a good example as the last skill grants mobile invincibility yet the class is all about gaining control of the terrain, it is undoubtedly better to be able to move freely while invincible and a major improvement on the class power yet that ignores the inherent weakness it has of being forced to stay within a zone which is a core element of its Identity.

Another less common scenario, likely to be solved, was classes with two directions transitioning from A to B between Circles and have a counterpart opposite arrangement of in B to A requiring players to pick both for a full set. That wouldn’t be much of an issue if it weren’t for the fact that by not going full class the base skills can’t be leveled further, even if you match A to A or B to B it won’t reach the full potential of an united Identity class as it requires both classes to be maxed. There’s a difference between progression spikes and direction turns, the latter punishes both abandon and commitment, which are two essential elements players have to use, when building their characters.

Since the only relevant element to be discussed is skill scaling, with some math, and the impact of skill points distribution will be the last fundamental part required for future analysis (even tho this one wasn’t really needed for that). I may have missed something so if that’s the case please point out in the comments along other arguments about the qualities and flaws of both systems, if you don’t have anything to add up to the discussion but still want support this kind of thread leave a like here and on other comments that are interesting.

Design Breakdown - Skill Points
What Makes a class design good/bad?
closed #2

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