|
On January 27 2011 06:58 TimeSpiral wrote:Show nested quote +On January 27 2011 06:46 VelRa_G wrote:
Great stuff. It's good to see a rigorous analysis of these fundamental, core aspects of gameplay. If you want to be extra baller, immortalize this thread on Liquipedia. I'm unfamiliar with this process. Care to elaborate? Is this something you guys think the community would consider useful enough to have in that type of directory?
Of course. There is already a Unit Positioning article there, but its current situation is barebones.
 Unit Positioning
Head here and login with your TL account. If you are unsure about how to contribute to a Wiki, you can navigate the numerous help pages available there.
|
You mention Fig. G and Fig. H, but there is only two E's and two F's.
|
On January 27 2011 08:03 dbosworld wrote:
You mention Fig. G and Fig. H, but there is only two E's and two F's.
Wow. I know exactly why that happened too. Shoot.
I've put in a temporary edit. I will fix it tomorrow :/
Thanks, Dbosworld.
|
Don't forget the Protoss forcefields which can be used to split up a bioball into 3 or 4 parts if used creatively in conjunction with terrain. This will majorly screw the positioning of terran. Psi storms will wreck the trapped marines, and the part nearest to the P army will get surrounded by zlots.
What I don't get is why pro players (Sjow, demuslim, even the ogs terrans) don't load their trapped guys up into medivacs and escape, instead choosing to stim, run and let them die. Surely the micro can't be THAT tough? If you can drop, surely you can load trapped guys into medivacs.
Btw, jinro is the only terran player I have seen escaping FF trap by evacuating marauders with medivacs, but he does it rarely because he relies more heavily on tanks, and Toss are forced to move back due to the long tank range.
|
Spread your arc for my ball so I can focus fire.
Stupid things aside, this is a good post for newbies. You should also integrate some analysis of Terrain because formation and Terran go hand and hand. For example, there's a reason why you set up your ranged units behind a choke because they end up in the 'ball' formation.
|
On January 27 2011 08:44 Antisocialmunky wrote:
Spread your arc for my ball so I can focus fire.
Stupid things aside, this is a good post for newbies. You should also integrate some analysis of Terrain because formation and Terran go hand and hand. For example, there's a reason why you set up your ranged units behind a choke because they end up in the 'ball' formation.
Don't assume someone is a noob just because you already know something  But I get your point.
That is why I labeled the knowledge classification as "Intermediate." I could definitely see myself expanding this type of article if I see there is a calling for it.
|
Oh, before I forget, I want to provide a quick answer (in my perspective, so don't take it too seriously) to the zergling baneling combo against marine ball/spread formation.
As the T, I will ball up my bioball against the zerg for maximum damage output, but will assign 1 hotkey for marauders and 1 hotkey for marines. When the banelings come rolling, I will select the marine hotkey and pull back, while the marauders will scoot and shoot the banelings (1 marauder 2 shot 1 baneling w/o upgrades). All this while maintaining ball formation (2 balls). Hopefully this will be enough to take down the banelings. If not, then I will lose all marauders to zerglings but still have full marine ball to take down the weakened zerglings.
If there was nowhere left to run, then I will spread the marines using mouse dragging, and hope to escape with at least one portion of the marines.
I rarely face such massive head-on attacks by the zerg on naked bioballs, as I use drops and airforce to hurt zerg, while using bunkers as defense (not many banelings left if they rushed my bunkers and depots). Hence, do take what I wrote with a grain of salt...
|
An important thing to add is that the disadvantage of Ball vs Arc is greatly reduced if the distance between both armies is smaller then the range of the units. For example, if you have two pure stalker forces, and your enemy has an arc on you, it can be beneficial to blink right on top of him, of course only if you are sure that you have at least similar amounts of stuff, since this will make a retreat of any side very hard and costly.
The main goal is always to maximize the amount of your stuff hitting his stuff, and ideally at the same time to reduce the amount of his stuff hitting yours. While this is quite obvious, it can´t hurt to state it, because i think while it is intuitively clear, since i have the feeling that a lot of people do know which positions they prefer their army to be in from experience, but do not exactly know why these specific positions are better than others. And this always comes down to maximizing the amount of time your units hit the enemies, and minimize the amount of time the enemy units hit yours. Ranged want to kite melees, because this way they get more hits for each hit the enemy gets. Melees want to surround, because this way they can prevent the enemy from kiting. Siegetanks want to be spread out, because they can still fire all the time because of their large range, but the enemy needs to run from tank to tank to kill them. No unit should be wandering around behind the lines looking for a spot to get in, because then they are not firing, which is bad. All comes down to the same basic principle.
So even if you don`t have an arc, an engagement where your ball is near to his arc is far less bad for you than an engagement where your ball is at maximum range of his arc, with all of your enemies stuff firing at you, while half of your guys run around aimlessly, and generally are not firing.
|
On January 27 2011 04:43 Skyro wrote:Show nested quote +On January 27 2011 03:39 .kv wrote:
Nice post but you should also inform techniques on how to get this arc instead of creating a ball when you move out and attack In that coaching video that was posted with InControl he pointed out a nice way to achieve a nice arc/spread while moving your army by first spreading out your army and then clicking attack move to a point very far in front of them, causing them to move in a more uniform line, rather than converging to a point close in front of them and forming a ball.
remember to make sure there arent any ramps or any other obstacles between you and you attack move location, your units will converge on a turn or a ramp regardless. This tactic works best in open ground but can be tailored to fit many scenarios... Its crucial in PvP collossus wars to use this imo.
|
On January 27 2011 08:53 TimeSpiral wrote:Show nested quote +On January 27 2011 08:44 Antisocialmunky wrote:
Spread your arc for my ball so I can focus fire.
Stupid things aside, this is a good post for newbies. You should also integrate some analysis of Terrain because formation and Terran go hand and hand. For example, there's a reason why you set up your ranged units behind a choke because they end up in the 'ball' formation.
Don't assume someone is a noob just because you already know something But I get your point. That is why I labeled the knowledge classification as "Intermediate." I could definitely see myself expanding this type of article if I see there is a calling for it.
I'd say unit positioning is a pretty basic skill because good positioning will beat mindless good macro if you have decent macro.
|
Speedlings are almost 5x better against the spread than they are against the ball It would be better if you could avoid things like that. 5 is a number you pulled out of... nowhere. It depends on the number of lings, marines, marauders, quality of spread, positioning of marauders, zergling micro, terrain, usage of stim, creep presence...
|
On January 27 2011 11:09 Cano wrote:Show nested quote +Speedlings are almost 5x better against the spread than they are against the ball It would be better if you could avoid things like that. 5 is a number you pulled out of... nowhere. It depends on the number of lings, marines, marauders, quality of spread, positioning of marauders, zergling micro, terrain, usage of stim, creep presence...
It is definitely a generalization, but in the example I created the formula yields a result of 4.59. Yes, there are many factors, but I'd be willing to be that if I ran this permutation through statistical rigor we'd see a nice standard distribution with 4-6x being the majority.
I think it is unfair to say I based it on "nowhere" when I show you the exact formula used to generate the number.
|
Yes, but you need to realize that your formula does not directly apply to a real world situation. Your formula describes the surface area of the terran bio army. While this is certainly relevant, it only directly translates into the effectivity of the zerglings if they magically appear with instant 100% surface area, and only applies for as long as there are enough zerglings to cover the whole of the surface area, and the area does not change, for example because of stuff dieing.
In a real world situation, the zerglings need to flood in, probably from one specific direction, wrap around the bio, and kill it. All this time you are losing lings, and you will especially lose the foremost lings first, thus making the time to get the surround even longer, so you will need more zerglings to ever achieve that perfect surround, decreasing the efficiency of them even more. Encounters between bio and lings do not tend to take a long time, since both sides, but especially the bioball, do insane amounts of damage to each other, so the time to get the surround is not insignificant. This time also is different when surrounding a ball when compared to surrounding small blobs, and especially the zerglings do not gain any efficiency from the surface area they have while moving in to surround.
So if you lose half of your lings before you even get a surround, you might not even have enough lings left to even use all of that surface area, thus gaining nothing at all. You only gain the full potential increase if the time running to the enemy is insignificant when compared to the time spend fighting, and also you have enough lings to cover the whole surface for the whole time of the engagement, which basically is not a real-world situation.
Edit: This is not meant to say that this is not important. Just the plain "5 times more efficient" feels strange. I just did some testing, and, for example, you can fit around 23 lings around a ball of 20 marines, but if you spread those rines out into groups of three, each group can be surrounded by about 9 lings, making a total of about 63 lings able to hit them. Note how this, for example, is not an increase of *5.
For all of the following tests i used stim + combat shields vs lings with speed, no upgrades on each side.
The influence of the formation is still very important, for example, those same 20 marines spread out like that die to 40 lings, with about 11 lings at half health still alive. But clustered up, more than half of the marines survive, and all the lings are dead. Your lings are far more effecive against spread out marines. But do not make up random numbers.
For example, those same 20 marines spread very barely defeated 36 lings. If those 36 lings were 5x as effective as they would be against a clustered force, this would mean that the 20 marines clustered would be able to barely defeat, or maybe barely lose against, 180 lings. This obviously does not happen. About 150 lings survive. The reason is obviously that marines start dieing, and there are still enough lings to cover the whole surface area for the whole fight.
|
On January 27 2011 12:11 Simberto wrote:
Yes, but you need to realize that your formula does not directly apply to a real world situation. Your formula describes the surface area of the terran bio army. While this is certainly relevant, it only directly translates into the effectivity of the zerglings if they magically appear with instant 100% surface area, and only applies for as long as there are enough zerglings to cover the whole of the surface area, and the area does not change, for example because of stuff dieing.
In a real world situation, the zerglings need to flood in, probably from one specific direction, wrap around the bio, and kill it. All this time you are losing lings, and you will especially lose the foremost lings first, thus making the time to get the surround even longer, so you will need more zerglings to ever achieve that perfect surround, decreasing the efficiency of them even more. Encounters between bio and lings do not tend to take a long time, since both sides, but especially the bioball, do insane amounts of damage to each other, so the time to get the surround is not insignificant. This time also is different when surrounding a ball when compared to surrounding small blobs, and especially the zerglings do not gain any efficiency from the surface area they have while moving in to surround.
So if you lose half of your lings before you even get a surround, you might not even have enough lings left to even use all of that surface area, thus gaining nothing at all. You only gain the full potential increase if the time running to the enemy is insignificant when compared to the time spend fighting, and also you have enough lings to cover the whole surface for the whole time of the engagement, which basically is not a real-world situation.
Edit: This is not meant to say that this is not important. Just the plain "5 times more efficient" feels strange. I just did some testing, and, for example, you can fit around 23 lings around a ball of 20 marines, but if you spread those rines out into groups of three, each group can be surrounded by about 9 lings, making a total of about 63 lings able to hit them. Note how this, for example, is not an increase of *5.
For all of the following tests i used stim + combat shields vs lings with speed, no upgrades on each side.
The influence of the formation is still very important, for example, those same 20 marines spread out like that die to 40 lings, with about 11 lings at half health still alive. But clustered up, more than half of the marines survive, and all the lings are dead. Your lings are far more effecive against spread out marines. But do not make up random numbers.
For example, those same 20 marines spread very barely defeated 36 lings. If those 36 lings were 5x as effective as they would be against a clustered force, this would mean that the 20 marines clustered would be able to barely defeat, or maybe barely lose against, 180 lings. This obviously does not happen. About 150 lings survive. The reason is obviously that marines start dieing, and there are still enough lings to cover the whole surface area for the whole fight.
Nice tests!
I like to see that. In one of my responses I did indicate that in order for the numbers to be precise you would need to use in-game unit measurements in relation to specific unit collision masks, etc, etc. So I get that.
Why do you think the numbers are random? The formula states that units are 4.59 times more exposed to melee attack in the spread formation than a ball of similar forces. I really don't think this needs to turn into an argument. I've edited the original post to alleviate some of the confusing wording. But the fact that there is a potential to be nearly five times more effective is true. I'm not sure why you think this means the number of units killed or defended is multiplied by 5 (as alluded to in your last post). In the nicest way possible, I think you are just slightly confused about that one aspect.
Now, granted, when comparing equal food counts of lings and marines, and setting it up in the lab, you get results like the ones you described. However, that is not what I was saying in the OP at all. I'm talking about the importance of unit positioning, and in this particular scenario, the vulnerability units have against fast moving melee units when in the spread formation, most notably, speedlings. Nowhere in the OP was I comparing "equal army marines to equal army speedlings." If I implied that, I will revise the wording.
|
Well, yes, the last part of was merely something fun i tried out, and is obviously not very useful, or intelligent.
The point i was making is that it your formula states that units have 4.59 times more surface area in that specific case, but not always. For example, if i oversimplify it to the marines taking up a specific area which could be distributed anyway you like, instead of them being specific units each, and assuming circular spaces covered, the following formulas for a circle are well known:
A=Pi*r²
C=2*Pi*r
With A being the area of the circle, and C the surface (don´t know the english word for that on a 2-dimensional figure)
So, if you split the area of one circle out into x different equal circles, you gain
A'=x*Pi*r'²=A=Pi*r²
since r=C/(2*Pi), and C'=x*2*Pi*r' ---> r'=C'/(2*Pi*x)
==> x*C'²/(4*Pi*x²)=C²/(4Pi)
C'²/x=C²
C'=C*sqrt(x)
C'/C=sqrt(x)
This means that the change in surface area is proportional to the squareroot of the amount of pockets you split the circle to. Obviously, this gets less precise the smaller the numbers of units involved are.
But generally, for example, if you split your stuff into 4 groups, this increases their surface area by 2, if you split it in 9 groups, by 3, and so on.
I, too, do not want to overcomplicate this, however, i believe that the formula you used to derive the surface area was based mostly on those circles which you draw by eye, if i am right. But maybe this formula i derived will help you out a bit.
Also, the problem that other guy and me were complaining about, is not the exact number of 5, but the general fact that you used a number there at all. As you can see from my results, the surface area increases with the amount of pockets created, and your 4.5* increase means that the enemy created about 20 pockets, while different results are gained at different amounts of unit groups. Also, the surface area, while being important, can not be directly translated into speedling efficiency.
Sorry for intruding and overmathing on a very specific and not necessarily very important part of your OP.
Edit: Stupidity
|
great post man. Think little tips should be outlined for noobies, like having an "arc" of marines at the top of your ramp on hold position to maximize dmg to a stalker that wants to peek up to harass. Also makes it super easy to pull a weakened marine away.
|
On January 28 2011 03:15 Ponyo wrote:
great post man. Think little tips should be outlined for noobies, like having an "arc" of marines at the top of your ramp on hold position to maximize dmg to a stalker that wants to peek up to harass. Also makes it super easy to pull a weakened marine away.
I plan on expanding this into more of a comprehensive Guide. I think the community will enjoy it.
|
Really good job man. It's always cool to see intuitive concepts analyzed like this.
|
UPDATE
========
Hey everyone, I just updated this guide with new graphics and whole new sections!
Let me know what you think.
|
Love it! Gonna study this loads, gonna help a ton in ZvZ roach battles.
|
|
|
|
|
|
EPT
