EPT![[image loading]](http://i.imgur.com/HEPsL.jpg)
Ivoldi knows the importance of the unit death system, do you?
If you read my last post you will recall that I suggested the following as some very promising features for a competitive RTS title:
+ Show Spoiler +
Utilizing a combat system wherein:
(1) every ranged unit fires an actual projectile that behaves according to a physics engine and has the potential to damage both enemy and allied units regardless of targeting
(2) every ranged unit has a minimum range on their projectile attacks
(3) most ranged units are slow or medium movement speed
(4) most ranged units are relatively fragile and easily killed if caught up to
(5) every ranged unit can be commanded to “attack ground” and fire once at a specific point on the map
(6) the aforementioned changes are balanced out by appropriately strong buffs to range, damage, special abilities, and AOE beyond what is the norm in a game like Starcraft 2
(1) every ranged unit fires an actual projectile that behaves according to a physics engine and has the potential to damage both enemy and allied units regardless of targeting
(2) every ranged unit has a minimum range on their projectile attacks
(3) most ranged units are slow or medium movement speed
(4) most ranged units are relatively fragile and easily killed if caught up to
(5) every ranged unit can be commanded to “attack ground” and fire once at a specific point on the map
(6) the aforementioned changes are balanced out by appropriately strong buffs to range, damage, special abilities, and AOE beyond what is the norm in a game like Starcraft 2
Now that we are also interested in how such a system would affect a spectator’s experience of the game, I propose some additional features, the role of which I will elucidate in this post:
(7) every unit death in the game produces some number of discrete objects which behave according to a physics engine in a context-sensitive fashion
(8) distinct causes of death can produce different numbers and/or types of discrete objects upon unit death
(9) the objects produced by unit death stay in the game as discrete objects for some appreciable amount of time (and possibly indefinitely)
(10) artillery fire and powerful abilities alter the appearance of the terrain if they hit the ground, leaving behind an obvious trace of some sort
But before we delve into the implications of these extra features, we need to appreciate the way in which unit death generally works in Starcraft 2, that we might tease out the relevant differences and how they affect the spectator experience.
Unit Death in Starcraft 2
Unit death in Starcraft 2 occurs in the following way:
- A unit is killed
- Depending on what killed that unit one of a variety of set death animations will ensue
- The remains of the unit quickly disappear
This is the case whether a marine is being blown to smithereens by a siege tank or a zergling is killing a probe. A given animation unfolds as a result of the cause of death, and what objects are produced do not interact at all with other “forces” in the game (such as explosive blasts). They do, however, sometimes interact with terrain by rolling down hills and so forth. This is actually the only time that anything semi-unique occurs during unit deaths; by and large they are the exact same animations recycled over and over in response to cause of death.
The Trouble With Monotony
These death animations are often quite well done in the sense that they are artfully animated and dazzling to behold. The problem, of course, is that they are always the same. Once you’ve seen a stalker get shot down by a marauder (or a roach, marine, immortal, etc.) a few times, you’ve seen all you are ever going to see. That this is troublesome becomes increasingly obvious as the number of units of the same type dying of the same cause close in time increases. It’s one thing to see marines dying to banelings in various small pockets and skirmishes throughout a game, and quite another to see twenty of them die to a baneling bomb and perform what I like to call “synchronized acid death,” wherein every single unit undergoes the same death animation at the exact same time and put on what looks suspiciously like a well-coordinated dance routine.
![[image loading]](http://i.imgur.com/iQku2.jpg)
Michael can’t compete with this kind of timing
That some of the most amazing and game-changing moments in Starcraft 2 are made to look so silly by the way unit death operates in this engine strikes me as something of a tragedy. What ought to be incredibly exciting and uniquely interesting events are dumbed down and made to look odd and repetitive, the result being that the sheer importance of the event to the state of the game (in a more abstract, competitive sense) must pick up the slack in order to make it truly exhilarating. This will make it harder for less hardcore fans or the uninitiated to jump in and really enjoy the spectacle that is a well-played match. The death animations in a competitive RTS should certainly make these sorts of moments as thrilling and gripping on a primal, visceral level as is feasible.
![[image loading]](http://i.imgur.com/wJlLj.jpg)
![[image loading]](http://i.imgur.com/7nDiA.jpg)
This sort of thing seems… underwhelming. Numerous identical piles of rubble, not even a hint of the debris being affected by the concussive force of the blast; a blast which happens to be the most devastating weapon in this game.
Learning from Myth, Part II
Fortunately, this sort of monotony does not have to be the case. I’m going to argue that Myth II, an RTS released in 1998, had a much better unit death system in place, one that any game which seeks to attract and entertain spectators (and boost replay value generally) ought to emulate. Let us consider the major differences:
#1 Upon death units decay into objects that are subject to manipulation by in-game effects
When a unit dies in Myth it produces objects which can be subsequently affected by events in the game. Were a dwarf to toss a bomb and kill a thrall, the thrall as a unit would disappear from the game and it would be replaced by several “chunks” of thrall that exist in the context of a physics engine. These objects are then given an initial velocity (which includes speed and direction) as a result of the angle from which the bomb hit the thrall. So the unit disappears while an arm, a shoulder pad, and a head take its place and are imparted an initial velocity that is sensitive to the directionality and power of the attack which killed the thrall. These objects will then begin to move as the physics engine dictates. If unaffected by other events in the game their momentum will play out roughly as you would expect in reality; they will glide so far before hitting the ground, bouncing, rolling, and finally coming to a stop. If, however, a secondary blast were to go off in the path of one or more of these objects, the objects would be affected, again per the specifications of the physics engine, and thereby receive a new trajectory and velocity. The core idea is that units decay into discrete objects which can be manipulated by subsequent events which occur after the death of the unit in question despite them not being the cause of death.
#2 Cause of death determines the number and/or type of objects produced on death
Whereas in Starcraft 2 the cause of death determines which animation will ensue, in Myth the cause of death determines what number and type of objects will be created. When a baneling kills a marine, the “marine acid death” animation unfolds. When a wight explodes on a dwarf, a given number of dwarf chunks are created and these objects are given a certain character based on the cause of death; they may turn a charred black, or they may trail smoke, blood, or puss as they fly through the air. (There are several combinations of these depending on the unit killed and how it was killed.) So the cause of death in Myth determines the number of objects produced when a given unit dies and what the character/appearance of those objects will be.
#3 Cause of death determines the initial behavior of the objects produced
That which kills a unit in Myth will also determine how the objects produced on death behave in the context of the physics engine. That is, the cause of death will determine the initial velocities of the objects produced. If a thrall is shot down by archers, it will simply collapse into a heap of thrall chunks. One or two items might roll a tiny ways, but they are essentially given no velocity and just drop from a given height. Were that same thrall to be killed by a fireball that exploded several feet in front of it, the objects produced would violently shoot off at high speeds. Further, the direction in which the objects shot off would be relative to the angle of incidence of the fireball. If the explosion came from behind the thrall the bits would fly forward; if from the left of the thrall the bits would careen off to the right, and so on. So that which kills a unit will also determine the initial velocities of the objects produced on death in a context-sensitive fashion.
#4 Objects produced by unit death remain in the game until destroyed by in-game effects
The objects created by unit death do not quickly disappear like the remains in Starcraft. They stay in the game and continue to respond to what is occurring on the battlefield. The head of a dwarf blown off five minutes prior will still hop off the ground and start rolling again if an explosion goes off close to its resting place. The blood that was smeared all over the hillside stays there, and the earth remains scorched where a lightning bolt struck several minutes before.
In fact, in Myth it is even possible for the refuse of the battle to take part in the battle. Discarded swords, helmets, and even limbs can be launched into the fray and cause small amounts of damage if they hit a unit. Objects are only removed from the game entirely if something extreme happens, such as a lightning bolt landing on the object. This will typically remove it from the game instead of launching it. The overall effect of this is an extremely gritty, realistic, and dynamic battlefield atmosphere. One can often get a sense of how the carnage unfolded in some detail just by looking at the distribution of gore on the terrain.
Comparing the Systems – Which is More Spectator Friendly?
I have included this video demoing some of the different scenarios in both games while noting any relevant differences and what effects those might have on a spectator’s (or player’s) experience of the game. I have provided a summary of each general unit type in the two systems if you don’t want to watch the whole thing, if you hate my voice, or if you just want to be more certain of what I’m claiming and how I justify it.
Explanations in Text + Show Spoiler +
Low Tier Melee Units
Low tier melee units killing one another in Starcraft 2 leads to one of maybe 2-3 death animations occurring, the remains of which quickly dissipate. In Myth 2, when a low tier melee unit kills another it creates some number of persistent objects which then drop onto the ground, having been given no initial velocity upon creation. This leads to a pile of objects in the spot that the unit was killed. The primary difference is that these objects can be affected by future events in the game, such as an explosion. While they do not do anything dramatic upon creation they have the potential to be flung across the battlefield later.
Suicide Units
The baneling in Starcraft 2 does damage to everything in a given radius when it explodes, and if it is enough damage to kill the units then those units will play out their specific acid death animations and quickly disappear. The wight in Myth 2 similarly does damage in a given radius, though it also shoots out smaller chunks of itself that can create miniature detonations beyond the initial blast. It will also blacken the ground and temporarily paralyze anything that does not die from the blast. However, when the wight kills a unit those units degrade into a number of objects and are imparted an initial velocity relative to the center of the wight explosion. So a warrior on the left of the wight will be shot off to the left, while the archer on the right of the wight will be scattered off to the right, and so forth. These body parts will trail blood and interact with the terrain according to the physics engine after being shot off from the point of origin, the net effect of which is a unique and context-sensitive patch of persistent gore. Which is just to say, the spectacle produced by a wight exploding is vastly more entertaining to witness than that produced by a baneling.
Artillery Units
Artillery units in Starcraft 2, such as the siege tank, have appreciable AOE damage and often lead to the spectacle of limbs flying. However, it achieves this spectacle in a very narrow fashion which keeps it as repetitive and dull as every other death in the game. While a siege tank shooting a marine can and often does lead to an arm flying up into the air and splattering onto the ground, the scene quickly loses its charm when you realize that the arm in question is the only limb that ever goes flying, and, moreover, that this arm always follows the same trajectory. It picks a direction to fling to, follows a fixed trajectory every time, and ignores every other in-game effect that occurs as it flies to its destination and after it has landed. Which is just to say it is but another recycled animation that is in no way context-sensitive or interesting to behold beyond the first few times. In Myth 2, artillery units such as the dwarf are constantly painting complex and context-sensitive works of strategic art with body parts, blood trails, charred ground, and bits of armor. The behavior of the gore is intimately related to the positioning of the shots loosed by the dwarves, and the objects produced by one death always have the potential to be disturbed and tossed about by further events in the game. This results in, once more, a far more complex, fresh, and entertaining spectacle.
Giant Units
The ultralisk, the quintessential giant unit of the Starcraft universe, actually behaves precisely the same (insofar as unit death is concerned) as the low tier melee units. The ultralisk kills a target which then goes through a death animation in the precise spot it was killed and quickly decays thereafter. An ultralisk killing a marine looks suspiciously like a zealot killing a marine (and that’s because there isn’t actually a difference in what can occur). In Myth, however, giant units such as trow or forest giants bring a unique and terrifying flavor to the destruction of lesser units. Rather than the unit simply plopping into a pile of objects as would happen in low tier versus low tier battles, the trow’s kick or the forest giant’s swat send body parts flinging away from the point of impact. The sheer force of their blows leads to a unique sort of gore where some body parts will stay close by while others will be launched away, trailing blood. One can tell simply by seeing the aftermath of the battle that something far more formidable than a low tier melee unit killed the units in question, as it leaves a relatively distinctive pattern of body parts and blood (in the notable absence of scorched earth). (As an aside, the trow has a very ominous and powerful movement sound, as any properly-made giant badass should have. The ultralisk in Starcraft 2 does not, and I fear I will never understand why.)
Spellcasting Units
Spellcasters in Starcraft 2 like the high templar and the infestor can cause devastating, game-changing damage in a matter of seconds. While this is undoubtedly important for the competitive aspect of the game, the way in which it has been executed as a spectacle to enjoy is woefully dull. Fungal growth will lock down units and they will play their generic death animations when enough damage has been done. Psionic storm operates exactly the same way, minus the locking down. Both produce scenes similar to what the baneling brings; namely, synchronized generic death rather than synchronized acid death. This is boring to see more than a few times, and the sheer importance of the spell’s effects for the state of the match have to pick up a lot of slack in order to make things exciting. Yet in Myth 2 the spellcasters produce uniquely interesting carnage when they kill units. The fetch, for example, fires an extremely loud, fast lightning bolt that sends smoldering, burnt husks of their targets flipping through the air and bouncing across the map. These chunks will, of course, continue to interact with subsequent lightning blasts or other disturbances. The spot where the lightning hit is clearly marked by charred earth and the whole scene is, once more, sensitive to the angle that the bolt came from. Here again we have complexity and dynamism instead of repetition.
Low tier melee units killing one another in Starcraft 2 leads to one of maybe 2-3 death animations occurring, the remains of which quickly dissipate. In Myth 2, when a low tier melee unit kills another it creates some number of persistent objects which then drop onto the ground, having been given no initial velocity upon creation. This leads to a pile of objects in the spot that the unit was killed. The primary difference is that these objects can be affected by future events in the game, such as an explosion. While they do not do anything dramatic upon creation they have the potential to be flung across the battlefield later.
Suicide Units
The baneling in Starcraft 2 does damage to everything in a given radius when it explodes, and if it is enough damage to kill the units then those units will play out their specific acid death animations and quickly disappear. The wight in Myth 2 similarly does damage in a given radius, though it also shoots out smaller chunks of itself that can create miniature detonations beyond the initial blast. It will also blacken the ground and temporarily paralyze anything that does not die from the blast. However, when the wight kills a unit those units degrade into a number of objects and are imparted an initial velocity relative to the center of the wight explosion. So a warrior on the left of the wight will be shot off to the left, while the archer on the right of the wight will be scattered off to the right, and so forth. These body parts will trail blood and interact with the terrain according to the physics engine after being shot off from the point of origin, the net effect of which is a unique and context-sensitive patch of persistent gore. Which is just to say, the spectacle produced by a wight exploding is vastly more entertaining to witness than that produced by a baneling.
Artillery Units
Artillery units in Starcraft 2, such as the siege tank, have appreciable AOE damage and often lead to the spectacle of limbs flying. However, it achieves this spectacle in a very narrow fashion which keeps it as repetitive and dull as every other death in the game. While a siege tank shooting a marine can and often does lead to an arm flying up into the air and splattering onto the ground, the scene quickly loses its charm when you realize that the arm in question is the only limb that ever goes flying, and, moreover, that this arm always follows the same trajectory. It picks a direction to fling to, follows a fixed trajectory every time, and ignores every other in-game effect that occurs as it flies to its destination and after it has landed. Which is just to say it is but another recycled animation that is in no way context-sensitive or interesting to behold beyond the first few times. In Myth 2, artillery units such as the dwarf are constantly painting complex and context-sensitive works of strategic art with body parts, blood trails, charred ground, and bits of armor. The behavior of the gore is intimately related to the positioning of the shots loosed by the dwarves, and the objects produced by one death always have the potential to be disturbed and tossed about by further events in the game. This results in, once more, a far more complex, fresh, and entertaining spectacle.
Giant Units
The ultralisk, the quintessential giant unit of the Starcraft universe, actually behaves precisely the same (insofar as unit death is concerned) as the low tier melee units. The ultralisk kills a target which then goes through a death animation in the precise spot it was killed and quickly decays thereafter. An ultralisk killing a marine looks suspiciously like a zealot killing a marine (and that’s because there isn’t actually a difference in what can occur). In Myth, however, giant units such as trow or forest giants bring a unique and terrifying flavor to the destruction of lesser units. Rather than the unit simply plopping into a pile of objects as would happen in low tier versus low tier battles, the trow’s kick or the forest giant’s swat send body parts flinging away from the point of impact. The sheer force of their blows leads to a unique sort of gore where some body parts will stay close by while others will be launched away, trailing blood. One can tell simply by seeing the aftermath of the battle that something far more formidable than a low tier melee unit killed the units in question, as it leaves a relatively distinctive pattern of body parts and blood (in the notable absence of scorched earth). (As an aside, the trow has a very ominous and powerful movement sound, as any properly-made giant badass should have. The ultralisk in Starcraft 2 does not, and I fear I will never understand why.)
Spellcasting Units
Spellcasters in Starcraft 2 like the high templar and the infestor can cause devastating, game-changing damage in a matter of seconds. While this is undoubtedly important for the competitive aspect of the game, the way in which it has been executed as a spectacle to enjoy is woefully dull. Fungal growth will lock down units and they will play their generic death animations when enough damage has been done. Psionic storm operates exactly the same way, minus the locking down. Both produce scenes similar to what the baneling brings; namely, synchronized generic death rather than synchronized acid death. This is boring to see more than a few times, and the sheer importance of the spell’s effects for the state of the match have to pick up a lot of slack in order to make things exciting. Yet in Myth 2 the spellcasters produce uniquely interesting carnage when they kill units. The fetch, for example, fires an extremely loud, fast lightning bolt that sends smoldering, burnt husks of their targets flipping through the air and bouncing across the map. These chunks will, of course, continue to interact with subsequent lightning blasts or other disturbances. The spot where the lightning hit is clearly marked by charred earth and the whole scene is, once more, sensitive to the angle that the bolt came from. Here again we have complexity and dynamism instead of repetition.
The Battlefield Atmosphere
![[image loading]](http://i.imgur.com/KJBv4.jpg)
One minute after slaying a horde of Myrkridia
![[image loading]](http://i.imgur.com/635dM.jpg)
One minute after a maxed PvZ battle on Cloud Kingdom
In Starcraft there is actually zero evidence of a battle having occurred anywhere on the map if more than a minute has elapsed since the fighting stopped. This is because nothing mars the ground and what gore there is disappears very quickly into nothingness. This has its advantages: the scene is less cluttered and the corpses of giant units and the like are not clogging up the screen nor creating additional hardware strain. But it has some very significant drawbacks as well, especially when the game we are playing is designed to be pleasing to a spectator.
In Myth the battlefield gore is remarkably persistent. Between the burn marks, the chunks of flesh, the pieces of armor and weapons, and the pools and trails of blood, you can tell at a glance that some serious shit has occurred. The designers of Myth actually had a rather elegant solution to keeping the battlefield gore persistent enough that it lends a very gritty, realistic, and even historic sense to the game without making it so persistent that it clutters up the field of play and begins to look silly. The solution was that explosives usually terminate any organic chunks they hit, while leaving intact pieces of armor, weaponry, and the blood stains. This works out so that if a battle is being waged with artillery or spellcaster units they are continually “mopping up” the many hunks of organic gore that they are creating. The battlefield still looks very grisly, but the amount of leftover corpse bits is almost always kept at a manageable level anywhere a battle is occurring.
Imagine a professional player holding off a very nasty timing attack as he takes his natural expansion in a game with this sort of unit death system. Imagine that there are bits and pieces of enemy units strewn across the ground outside of his building wall, blood trickling down the ramp that leads up to his natural, maybe even some smoldering limbs and spatters of blood finding their way onto his buildings. The terrain in front of his base is scorched and smeared with blood, the patterns in the gore making it obvious that a tough fight was had in this very spot, and artillery played no small part in the defense. Or it could be, as it is in Starcraft 2, that you notice nothing but a damaged nexus and some hastily placed gateways and pylons. Of the two, I think we have a clear winner in terms of the spectator’s experience.
Blizzard Already Taking Some Steps in This Direction
And you thought this was all just eloquently presented nostalgia
Observe the burrow charge on the marines. This is the direction Blizzard (and any other developer) needs to keep pushing in order to make their game as spectator-friendly and all-around-kickass as possible.
Great Potential for Future Titles
I don’t really expect this will be implemented to the extent that I would like to see it implemented in Starcraft 2. (One can always hope, though.) However, the potential for using this basic kind of spectator-friendly (and player-friendly, for that matter!) unit death system in future RTS games is tremendous. There are so many variables that can be tweaked, so many graphical subtleties that can be added, and so many different ways to make the chaos wrought by any unit in the game a lot more exciting to behold than what we are accustomed to.
Some ideas off the top of my head:
+ Show Spoiler +
- Different corpse breakdowns based on the velocity of the killing blow. If it was an explosion of high velocity, break down the unit into a greater number of smaller pieces which are imparted a large initial velocity. If the cause of death is lower velocity, such as a weaker explosion or a smack from a giant unit, keep the unit largely intact, save perhaps a piece or two, impart them a moderate or small initial velocity, and let rag doll physics determine what the larger portion of the body will do thereafter. If a low tier melee unit kills another low tier melee unit, perhaps leave the corpse completely intact (some of the time, not always) and have it crumple to the ground as a large, persistent object. I can envision a system of 3-4 different tiers of “violence of killing blow” for each unit determining what objects they decay into upon death.
- For the laudably ambitious, one could even have corpse breakdown depend on the particular attack animation used to deliver the killing blow. If a giant were to chop downward with an axe and kill a warrior, have that unit split down the middle into two discrete pieces that then enter the game and behave according to the physics engine. Were the giant to fell another warrior with a horizontal swipe, have the warrior cut in half at chest level instead. One could also implement this with exclusively normal-sized units, as one warrior might take off another warrior’s head as the killing blow, or take off an arm, rather than having them just collapse into a heap every time.
- Unique object characteristics if certain kinds of spells dealt the killing blow. If there were a spell similar to fungal growth it might leave behind chunks of half-digested limbs or torsos while trailing an eerie greenish slime in place of blood. It would depend on the kind of spell in question, and the limit would be the imagination in this case.
- Have certain types of units react more strongly to certain causes of death. If there were some sort of mummy unit (or other dry and flammable unit), have it so that it explodes into a blinding rain of flaming cloth and bones. That the same fireball might just char or singe the chunks of less flammable units makes this event especially exciting when it occurs. (I HIT FIVE MUMMIES WITH A FIREBALL ZOMG FIRE!!1! LOOLL – a crude approximation of a casual gamer’s reaction.)
- Different colors of blood for different types of units, to further diversify the scene of the battlefield and tell an even more detailed story at a glance.
- For the laudably ambitious, one could even have corpse breakdown depend on the particular attack animation used to deliver the killing blow. If a giant were to chop downward with an axe and kill a warrior, have that unit split down the middle into two discrete pieces that then enter the game and behave according to the physics engine. Were the giant to fell another warrior with a horizontal swipe, have the warrior cut in half at chest level instead. One could also implement this with exclusively normal-sized units, as one warrior might take off another warrior’s head as the killing blow, or take off an arm, rather than having them just collapse into a heap every time.
- Unique object characteristics if certain kinds of spells dealt the killing blow. If there were a spell similar to fungal growth it might leave behind chunks of half-digested limbs or torsos while trailing an eerie greenish slime in place of blood. It would depend on the kind of spell in question, and the limit would be the imagination in this case.
- Have certain types of units react more strongly to certain causes of death. If there were some sort of mummy unit (or other dry and flammable unit), have it so that it explodes into a blinding rain of flaming cloth and bones. That the same fireball might just char or singe the chunks of less flammable units makes this event especially exciting when it occurs. (I HIT FIVE MUMMIES WITH A FIREBALL ZOMG FIRE!!1! LOOLL – a crude approximation of a casual gamer’s reaction.)
- Different colors of blood for different types of units, to further diversify the scene of the battlefield and tell an even more detailed story at a glance.
Concluding Remarks
The features of the Myth franchise I have discussed in these two posts seem to me to have been pointlessly lost to posterity, rather than carefully explored and discarded on account of demonstrable shortcomings. I do not think every single aspect of the way unit death was handled in Myth is ideal or even proper for a game like Starcraft 2. After all, keeping persistent gore on the field with 10 ultralisks and 8 thors doing battle could become extremely crowded extremely fast. There would have to be compromise. But the general features I listed at the beginning of this post are, in my opinion, the direction competitive RTS games should be headed. Blizzard seems to have an inkling of this already, as you can see from the HotS demo video. I’d like to know what you all think about the relative merits of the two unit death systems, so please post below!
Thanks for reading.
TL;DR
![[image loading]](http://i.imgur.com/rp8CW.jpg)
Perhaps you can enlighten me on this score.
