Creating Cool Boss Fights

Banzai Games game designer Dmitry Aborilov talks about inspiration, cooperation between the teams, and game mechanics used to achieve the WOW effect in boss fights.

A Shadow Fight 3 Developer Shares His Experience


Banzai Games game designer Dmitry Aborilov talks about inspiration, cooperation between the teams, and game mechanics used to achieve the WOW effect in boss fights.


Are you a game developer who has faced the challenge of making the final boss of a blockbuster game? One that will be long remembered by a multimillion audience of fans? I was surely not.


But when I first started my job at Banzai Games, it quickly turned out that this would be my new task. And, by listening to my inner voice, I realized that I madly wanted to do it. And I wanted to do it well. So the sixth chapter of Shadow Fight 3 and its three bosses stood in front of me. This unique challenge will be the topic of my essay.



About the game

Shadow Fight 3 is a Fighting/RPG game that was released a year ago and has over 50 million installations at the moment. In this game you can create your own character and choose a weapon, armor, and fighting style from a selection of a few dozen. The game has a strong emphasis on the plot, where the player meets many NPC characters and fights against various dangerous opponents. After the battle, random rewards are given: new armor, weapons, and/or additional abilities.


How to start working on the bosses


At least for me as a game designer, that task starts with the plot. For the plot, our writers invent different enemies, define their role in the world, and describe their appearance. Only then a game designer appears. In fact, the game designer’s task is to take this foundation (appearance, personality, character motivations) and wrap everything using the right fighting mechanics. There is a close interaction with all other departments on all stages of the task - animators, artists, FX’ers, builders, and programmers.


It is important to unite many ideas and demands from all parties into one big picture. Of course, all of this needs a lot of meetings and brainstorms. But the result takes into account different points of view, and that’s important.


In order for the discussions to be effective, we need to think in advance. Therefore, I began with the analysis of bosses from other games and constantly discussed plot drafts with the writers. This gave me extra time to think. It also made it possible to better understand what kind of idea and emotion the enemy would translate to the player.


What are the bosses in fighting games?


The bosses are usually opponents with increased health, high damage, and unique abilities. Often they have an interesting appearance and a good plot motivation. What makes a cool boss really cool? Why will the player remember that fight?


I asked my colleagues to list the best bosses in Shadow Fight 3, plus I looked for references on YouTube. As a result, there were three bosses who were discussed the most: Gizmo, Sergeant, and Itu.



Why do players remember them? Of course, they appear very often in the plot and the game had already imprinted their images into the player's mind. But I'm interested in something else: what are their features in the gameplay?


Let’s take a look at the mechanics:


Gizmo. He fights with a two-handed sword, which he throws away during combat and continues the fight unarmed. The player must learn to fight against the sword and against the fists as well.


Sergeant. From time to time this boss becomes fully invulnerable and starts moving forward. The player has to avoid collisions, and then the boss performs a dangerous superattack.


Itu. When low on health, this boss triggers a regeneration stance. In order to interrupt that, the player needs to get behind his back and attack him, otherwise, Itu will fully recover.



All of these bosses are designed with a same idea in mind. They require the player to change his gameplay during the fight. They change the state of the battle in the middle of it.


Also, the Sergeant and Itu have a kind of a built-in puzzle. If you do not solve it, then the boss can’t be defeated. This condition ensures a memorable gaming experience because it is different than any other fights in the game.


It is extremely important to make a puzzle that is understandable to the target audience and can be solved without guides. Otherwise there will be dissonance. The player wants to beat some bad guys, but they are forced to play chess instead.


When it comes to understanding what the players like during boss encounters, the requirements for me are:

  • a puzzle

  • an increasing difficulty between rounds

  • integrity between boss appearance and gameplay mechanics


Start of a discussion


I do not want to go deep into the details of our endless conversations, brainstorms, controversies, and alterations of the concept. This is quite a difficult period, as in most cases each specialist thinks that his own part is more important than the whole boss image. Therefore, it is necessary to discuss a lot, but what is more important is to agree on a common vision so that everyone has the same idea about the boss encounter.


It is important that before starting the chapter development we had to design three bosses at once: Xander, the Queen, and the Shadow Mind. This is because content teams (animators, artists, FX specialist, and many others) work effectively only when everything is planned out in advance. Therefore, it was impossible to experiment on one boss and take into account that experience while designing the second one. It was necessary to agree on three concepts at once.





The first boss, Xander, was already well-described. The player has met him before during the plot, when he was a “loser-fighter”. After his defeat, he spent a lot of time studying various martial arts and now is ready to come back to the fight.


The best way to implement such an idea is to give the boss several weapons at once. So we gave him a staff, a sword, and a sharp blade called “sai”. We only had to select the best attacks from each of those styles and combine everything. A bit of framedata balance magic was needed there. To make the boss more interesting to play, blows with different attack distances were selected. The staff covered long distance attacks, sais for close ones, and the sword was used for the middle distance.


Is this enough for a good boss? Partly yes. That way Xander would definitely be remembered for its uniqueness, he even may be difficult ... but there won’t be any puzzle, it will not be something really new. The player has fought against these styles before, and only the difficulty has changed. Therefore, it was necessary to think further.


Searching for an inspiration


To come up with a mechanic that fits the boss well and creates a puzzle, you have to dig into the plot, the lore, and everything that has already been done. We were lucky. The writers suggested an interesting clue: it turns out Xander refused the use of shadow energy!


Shadow energy is the magic element in the world of Shadow Fight and at the same time is used as a comeback mechanic, which allows you to turn around a losing situation. Something like X-Ray from Mortal Kombat. When you get beaten, you get some energy. With full energy, you can use extremely effective special attacks. But since Xander refused to use it, it means that we have an unnecessary interface element (a shadow energy bar) that can be used for new purposes.



This character is hot-tempered and hates the player. And the fight takes place in prison. Therefore, I wanted to create a mechanic associated with aggression, strength, and danger. This brought me to the thought: let's make a two-state boss — ordinary and insane when he falls into a fighting frenzy.


Two states


My thought was to combine this idea with the following rage bar mechanics: Every time you hit a boss, he gets a bit of additional rage that decreases with time. If you make a combination of blows, then the rage bar will soon become full, and the boss will enter his “enraged” form. This means an increase of his damage and fortitude (the blows received won’t interrupt the boss's attacks), and the ability to attack the player boldly.


This mechanic is great for Xander from a plot perspective, and it is also great in terms of gameplay. If the player decides to attack the boss with a series of punches, then Xander will go insane and tear the player apart. This means, you have to beat him carefully, only one or two hits at a time.


The player’s emotion


The rage really forced the player to think and allowed him to play around with the boss during our playtests. But the playtest also showed an interesting phenomenon: when the boss caught on fire, the players were enraged too and wanted to beat him even harder!



But this was an extremely useless approach. Enraged Xander had good protection from damage. It was added to emphasize the fallacy of the player’s attacks during rage mode. But after the tests, it became clear that the protection should be removed, as it spoils the overall impression.


But how to make the rage state dangerous if there is no more defense buffs? Increasing the damage did not help much; the boss didn’t become more challenging. I had to implement something different than that.


The decision came by itself, when the animators completed a unique superattack during which Xander beats with all kinds of weapons in one turn. Since there are a lot of strikes, there is a lot of damage. This attack dealt more than 60% of the player’s health bar as damage while Xander was infuriated. And this is often fatal to the player.



We added a mechanic that the boss is more likely to use this technique in his frenzied state. And everything immediately fell into place. Now the player should avoid making Xander angry. Otherwise, he will not only flare up and get all sorts of bonuses, but he can also kill him at once.


Players now had a choice. It turned out to give the desired emotion.


Finish line


According to the original checklist, we hit two points out of three. There is a puzzle and there is the integrity of gameplay and the plot. But to pass the boss you have to beat him in three rounds. And playing the same thing three times is pretty boring. So it was necessary to add a difficulty progression between the rounds.


We decided to make it simple - with each round the player needs fewer hits in order to infuriate the boss. Technically, we needed the boss to go berserk in the second round with 70% rage, and in the third round with 50%. But just changing the numbers is not interesting and not transparent to the player.


So we added two animations where Xander beats himself and gets extra fury. In the second round, when he had 70% rage, he began to beat his staff on his own forehead, so that he would get angry and gain a full rage bar. And in the third round, reaching 50% rage, Xander would pierce his own leg, enraging himself immediately.



It turned out that we actually shortened the progress bar, but made it memorable using the personality of the character.




I was happy with the result. We managed to implement all the elements of our checklist at once:

  • Combine story, visual, and gameplay components of the character;

  • Implement the mechanics of a rage mode, which adds a puzzle element to the fight;

  • Achieve a difficulty progression between rounds, changing not only the numbers, but also adding new memorable animations.


The original statement of the requirements and focus on the plot made it possible not to lose the image in search of ideas and not to quarrel when discussing all sorts of details.


I would also like to note that the rage mechanics are in control of the player, not the game. We could use a rule like "the boss goes into a rage when he has 30% of his health left". This would have worked, too, but the player couldn’t play around with the mechanics then. There would be no learning curve and so there won’t be any sense in experimenting.


Queen Iolanda


With the second boss things got more complicated. At the beginning of the boss development, we didn’t have a vision of her. The writers described two other bosses in detail, but they had no ideas about Queen Iolanda. Therefore, I had to develop it according to a very short description: “a robot girl”.


Was this enough to come up with some good mechanics? Yes and no. Your brain immediately begins to draw images from “Ghost in the Shell”, all sorts of augmentations, but because of this it is easy to drift too much into sci-fi, which has nothing to do with that story chapter.


It was necessary to take a break, think about which mechanics could be developed based on the given description, and try to find something most suitable for the boss implementation.


Searching for a mechanic


How can you show that a character in the game is not a person, but a robot that pretends to be human? I think the easiest way to do this is to remove some of her body parts. This is a classic - as a robot you can take your hand off and install something else instead.


Immediately plenty of ideas came to mind:

  • Special moves where hands or legs are thrown (or shot)

  • The possibility of the boss to fight without important body parts

  • Replaceable body parts

  • Self-repairs

  • Wires, bombs, etc.

That’s a bunch of interesting ideas to work on. But you can only do a small number of such mechanics (our team resources are limited), and the ideas should also integrate well into the rest of the game.


Therefore, only two of these mechanics made it into the game - superattacks with shot arms / legs and self-repairs. The latter has undergone so many changes that I don’t even remember how it was originally described. In any case, it had to change the player's fighting style.



Repair form


In contrast to Xander, who should not to be made too angry, I wanted to make a character who needs to be barraged with various blows. So the idea of the ​​self-repair came to be.


The boss is given a self-repair ability, which turns on for a couple of seconds after the last dodged attack. So, she can’t be beaten if the player stops his attacks. This mechanic was easy to implement, and it immediately began to demand certain behavior from the player.

But in this form, Iolanda had one big problem. It was difficult to balance her difficulty. If she was too weak, the mechanics would not work, and if she was too strong, the player would have to kill the boss in one single combination to have a chance to win the fight. This is virtually impossible, and is an example of a bad design, too. The boss fight should be long enough so that there is a feeling of facing a greater challenge than usual.


It was necessary to somehow break her repairs into stages, saving the player progress in between. So repair became the three “batteries”, which were placed on the back of the boss.


Each battery has its own health segment: the first from 25% to 50%, the second from 50% to 75%, the third from 75% to 100%. If the health of the boss falls below the operating range of the battery, then it falls off. For example, if a player has dealt only 15% damage, then the battery will recover up to 100% again. But if a player could inflict as much as 25%, the battery will break, and the boss will never regain health above 75% anymore.


The mechanic itself is not really remarkable, but it copes well with its main tasks:

  • Conveys the spirit of robotics. And the fact that there are three batteries at once

makes the mechanic even more suitable;

  • Generates a puzzle. The player must find out the strongest attack combinations or hope for an occasional critical hit, which will deal a lot of damage, since the standard blows are no longer effective;

  • It is easy to balance, since the mechanic was divided into separate pieces. In fact, there is a slight difference in the healing rate of the batteries: it is more difficult to destroy the last battery. But the difference is hidden from the player and is only about 10%.


Special Moves


We have a very cool team of animators, which can quickly make a few new moves and attacks. And they usually make them really awesome.


This happened here, too. The only thing that was in the initial requirements - the strikes must be very effective and work at different distances - medium and long. In the end, after a couple of weeks, I was called to review the results. The robot was already starting to show her arsenal of tricks.


So what about the requirements?

  • Reducing the effectiveness of strikes is easier than increasing. Therefore, if you first make an imbalance in favor of the boss, it would be corrected easily. It is impossible to make perfect balance from the start, so it’s not even worth trying;

  • The difference in effective attack distances is needed so that the blows would not interfere with each other. And the player would be confused if he realizes that the boss uses the same attacks with the same range but visually they are different.


Progress in the rounds

The mechanics began to work, and I had only to unite everything into a coherent picture and add more difficulty. From round to round, it was planned to include additional special moves. So the following scheme turned out:

  1. Batteries for a self-repair

  2. Batteries + Medium range attack

  3. Batteries + Medium range attack + Long range attack


Such a banal method of increasing the difficulty gave a very interesting result. Players could easily understand the basic mechanics in the first round, and then solve a new puzzle and look for other ways to win in each of the next rounds. The boss simply did not give them the opportunity to continue to use the tactics of the previous round.


That especially affected the players with long-range weapons. They could play relatively relaxed for two rounds, but then on the third round Queen Iolanda became an extremely dangerous opponent for them. And to pass the boss, you need to win in three rounds.







Here, once again, it turned out to meet the requirements of the entire checklist. But the most important thing that I understood when working on this boss is that we always need a clear concept. Otherwise, everyone will generate new fancy ideas, which are too difficult to use.


As designers, we must start from the restrictions. If there is only a short description - work with it. If it’s possible, invent your own terms and pass them to the whole team.


Batteries seem to me as the weakest part of this boss, as they do not change the gameplay very much. It could be worth experimenting with their different capacities a little longer. Plus they do not actually create unique battle situations. They force the player to “behave this way — the rest is wrong” and that's it. I think it is much better when mechanics allow you to approach it in different ways and get different results.


Shadow Mind


The last boss of the chapter should be the coolest one. This is his objective. And given that this is the chapter before the release of the big final, which will take quite a while, you need to do something super awesome. So that every player wants to go through it, see it with his own eyes, and tell his friends about it. And continue playing, of course!


On the other hand, Shadow Mind is our last chance to practice before the big showdown in Chapter 7. Of course, there will be more bosses there, but it will be necessary to invent and develop all final ones at the same time. Therefore, we needed to experiment right here and now.


So that the boss is guaranteed to be remembered and stand out against the previous two, you need some kind of stunning visuals. Therefore, the original checklist contained a very simple point: we need to have a wow-effect. How to achieve this was not so important; the main thing was to really surprise the players.




The search for a mechanic, as always, began with a plot and a lore study. Shadow Mind has long been in the game. He can slip into the mind of many characters and exist there. A fight with him takes place in something like the player’s mind, where, of course, the boss has full power - this is his habitat.


The first thing that comes to mind when you talk about a deity is all sorts of natural disasters. After all, this is a pure demonstration of power unknown to man. So there was an idea. Let the boss throw any kinds of shadow whirlwinds, typhoons, and all such stuff against the player. This is definitely WOW, since everything will be huge, dangerous, and unusual.


The idea was obvious for everyone, but immediately raised the complexity of implementation. For example, in the existing locations it was impossible to show something that huge. The camera is close to the characters, so a typhoon just does not fit, and you can’t move the camera away. The boundaries of the setting and various auxiliary elements that normally were hidden from the player would immediately come out.


Therefore, we decided to make something like an “infinite location” - a bridge that does not begin anywhere and never ends. Players could always go in any direction, well, or fly endlessly high. This made it possible to overcome all the potential problems with the camera in the future and provided truly unlimited possibilities for HUGE EFFECTS.



The second problem was founded in the balance of such typhoons. How much damage will they do? How can the player escape them? This wasn’t easy to answer. But there was a good alternative - we just replaced typhoons with a “bullet hell” explosion with thousands of projectiles flying around - and the player had to dodge them.



So it was possible to show something huge, but not in the form of one big element, but many small ones. It turned out that the damage will be small and can be avoided by hiding in the right place.


In order to finally create the WOW effect, we decided to create some kind of a superattack, during which the boss throws the player over a huge distance and directly goes after him. It would have turned out to be something like an epic superhero move, and at the same time an impressive demonstration of the infinity of the map. But when the animators took that task, it looked like a huge cartoon movie with a series of punches, take-offs, and other things. They even had to lower the damage received from each hit, since it would instantly kill a player with full health.





The visual expressiveness of a bullet hell strike is very cool. But it does not add a puzzle to the boss fight. Of course, the player will have to remember exactly where to stand and how to move so that he doesn’t get hit. This is something like a mini-game, but not a full-fledged mechanic. I wanted to add a more complex element.


Shadow Mind is constantly in shadow form and inhabits the player’s mind. We can assume that the laws of physics in this setting will differ from the real world ones. You don’t fight in your own head very often.


A simple mechanic for this was obvious: what if a player can only deal damage to the boss when he is in shadow form? It turns out that in order to interact with the “shadow world”, it is necessary to be a shadow yourself. In my opinion, this sounds logical.


We needed to elaborate this idea a bit more. I wanted the battle to consist of two phases which are constantly changing: the player in normal form versus the player in shadow form. But for this, we would need to give the player a lot of shadow energy, since under normal conditions the shadow bar will barely fill up more than once or twice per round.


This modification was easy to do: every time a player hits the boss, he gets a lot of shadow energy. So it makes sense to fight against the boss in the normal form, but instantly switch to shadow mode when the bar is full. The fight was effectively divided into two states that constantly replaced each other.


By the way, an unexpected additional result came with the new mechanics: once the player was in shadow form, he could prolong it by hitting the boss. So in certain cases it was possible to fight almost twice as long than in ordinary battles. This kind of “reward” for doing the right actions looked great with the overall concept.


The final touch for the mechanics was to reduce damage from the blows of the boss and the shadow moves of the player. This was a necessary rebalance so that the battle did not last for 1-2 shadow form phases, but for 3-5. In the end, after a few hours of fiddling with numbers, the fight became a real “shadow war” and became cool, even without bullet hell and other WOW-effects.



Progress in the rounds


At this stage we had to add some increase in complexity to finish the boss. For Shadow Mind this task was the easiest of all. First, you can simply increase the difficulty for the bullet hell from round to round, and second, you can reduce the cooldown on the use of his shadow abilities.


The point is that shadow abilities are difficult to counter. Therefore, it could be estimated that that the boss will kill the player with approximately 8 attacks. Consequently, if the cooldown for each ability is 6 seconds, the player will lose, if he won’t defeat the boss in around 1 minute (taking into account the time of the strike animations). Therefore, we simply reduced this cooldown from round to round and obtained an increasing difficulty as a result.




Most importantly, the WOW effect has impressed the players. On almost all YouTube videos with the walkthroughs of the sixth chapter of Shadow Fight 3 there is a bullet hell screenshot as a preview. The players were thrilled! Once again, everyone in the studio felt the immense power of visuals in games.


Although, of course, the effort to implement this battle was high. It’s hard to do something in a project that hasn’t been done by a team yet. And it’s even harder to make it playable. Our Shadow Fight 3 fighting mechanics did not provide character skills or moves for escaping a heap of bullets falling from the sky. There were a lot of problems that arose during the development.


But the puzzle aspect worked. Players began to share the most effective gear and weapons setups that gave bonuses for entering the shadow form or increased the damage in it. This is cool - it means this boss has created a different approach for the game.


Unfortunately, some players thought that the boss simply had much higher level than the players, and thought that was the reason why they caused so little damage that it won’t appear in the interface. This situation often occurs in normal battles in Shadow Fight 3, but at the design stage we missed this point.


I had to make a special fix - to lock the boss's health bar with a special visual effect when the player is out of the shadow form. So the players had a better chance to understand the mechanics.



A few months have already passed since we made this chapter. At the time of writing this article, the final boss of Shadow Fight 3 was already in production. But there are a number of lessons that I have learned for myself and I still use them:


  • To come up with a good result quickly and effectively, you need a concept. If it does not exist, we need any information possible to anchor your imagination. Three goals are achieved this way - it is easier to work, a better concept is obtained, and it is much easier to accept or reject new ideas.

  • A WOW-effect is extremely important. There is no sense in creating a boss that is not surprising and not memorable. It is necessary to have both an awesome visual and interesting mechanics.

  • Integration of the puzzle element and round progression should be thought out right from the start. If there is a suspicion that the mechanics are not well thought out, then, most likely, this is true.

  • While there is not even a rough understanding of how to change the numbers to make a mechanic work - don’t believe in it. It may be worth thinking twice or change the concept altogether. With Queen Iolanda, we managed to make a good boss, but it took some luck and a few sleepless nights.

  • Communicate with the team. A team performs much better when everybody understands each other correctly. Often, the time to explain the vision and ideas takes most of the concept development time. But it is still extremely important - because ideas do not make sense if they are not understood by a team. And usually in game development there is no time for playing around with many ideas. The deadlines don’t wait.

Latest Jobs


Playa Vista, Los Angeles, CA, USA
Senior Level Designer (Zombies)

PlayStation Studios Creative Arts

Petaling Jaya, Selangor, Malaysia
Lead Concept Artist

Digital Extremes

Lead AI Programmer
More Jobs   


Explore the
Advertise with
Follow us

Game Developer Job Board

Game Developer


Explore the

Game Developer Job Board

Browse open positions across the game industry or recruit new talent for your studio

Advertise with

Game Developer

Engage game professionals and drive sales using an array of Game Developer media solutions to meet your objectives.

Learn More
Follow us


Follow us @gamedevdotcom to stay up-to-date with the latest news & insider information about events & more