EPTedit: I stand corrected by brickrd, thanks for the tip because I haven't watched that many games lately.
LotV Beta Balance Update — September 3 2015 - Page 17
| Forum Index > Legacy of the Void |
|
Salteador Neo
Andorra5591 Posts
edit: I stand corrected by brickrd, thanks for the tip because I haven't watched that many games lately. | ||
|
brickrd
United States4894 Posts
On September 09 2015 19:30 Salteador Neo wrote: So why don't just give Overseers drop capability? Makes sense to me, they are upgraded overlords. Could also remove the dumb contaminate spell, if that thing is still in the game (haven't seen it in ages). If it's too good in early game then make them cost more or Lair tech? contaminate is used regularly at pro level especially to delay upgrades in zvz | ||
|
Tenks
United States3104 Posts
| ||
|
Big J
Austria16289 Posts
On September 09 2015 23:32 Tenks wrote: Has anyone figured out if there is a rhyme or reason to where the MULE is called down to from an OC? Yesterday I was playing on Dash & Terminal and a Zerg went Ravager and it seemed without fail my MULE would spawn on the mineral patch most exposed to the Ravager skill shot. I wanted it to spawn on the other side of the mineral field but I don't think I have any control over that. The usual SC2-criterium is closest target from the caster. So in this case it should simply be the closest mineral patch. If there are two mineral patches equally close, I believe the tiebreaker is some rule like "right before left". It probably roots in the way the search algorithm for targets is implemented. For example a simple search algorithm could look like this: ![[image loading]](http://i.imgur.com/gNyw1lP.jpg) So in this example the CC would search for a target in concentric circles and end its search when doing the third round and running into a mineral (which must be closest, otherwise there would have been a mineral within the first 2circles). That mineral would now be selected for muling. Note that the mineral is not the only closest mineral in this example (bottom blue is equally close). But everytime this algorithm is performed from this CC (unless the CC could turn, which it can't in it's static form) it would find the same mineral patch first. | ||
|
sharkie
Austria18522 Posts
| ||
|
Roblin
Sweden948 Posts
On September 10 2015 00:22 Big J wrote: The usual SC2-criterium is closest target from the caster. So in this case it should simply be the closest mineral patch. If there are two mineral patches equally close, I believe the tiebreaker is some rule like "right before left". It probably roots in the way the search algorithm for targets is implemented. For example a simple search algorithm could look like this: So in this example the CC would search for a target in concentric circles and end its search when doing the third round and running into a mineral (which must be closest, otherwise there would have been a mineral within the first 2circles). That mineral would now be selected for muling. Note that the mineral is not the only closest mineral in this example (bottom blue is equally close). But everytime this algorithm is performed from this CC (unless the CC could turn, which it can't in it's static form) it would find the same mineral patch first. this is very close to what I think it does but more specifically I believe it uses Dijkstra's Algorithm in a nutshell, its a pathing algorithm that searches a nearby space, then if it didn't find anything it searches a bigger space, repeat until you find what you are looking for or until all space has been searched. the main difference between your explanation and dijkstras algorithm is you mentioned a circular area, dijsktra uses "adjacent nodes" instead. | ||
|
Big J
Austria16289 Posts
On September 10 2015 00:40 Roblin wrote: this is very close to what I think it does but more specifically I believe it uses Dijkstra's Algorithm in a nutshell, its a pathing algorithm that searches a nearby space, then if it didn't find anything it searches a bigger space, repeat until you find what you are looking for or until all space has been searched. the main difference between your explanation and dijkstras algorithm is you mentioned a circular area, dijsktra uses "adjacent nodes" instead. Interesting one, do you happen to know whether this is the current state of the art? Also, does Dijkstra's algorithm find the closest mineral patch amonst multiple ones though? Because the way I understand it what it does is that it finds a way to one specific target and once it has found that target, it also knows the shortest way to the target. But I wonder if it always finds the closest target first if I gave it multiple possibilities.  But yeah, obviously they are using some more advanced algorithm than what I painted. One that is using the underlying board-structure of maps and maybe other SC2-specific stuff like Dijkstra's. | ||
|
Tenks
United States3104 Posts
| ||
|
Cascade
Australia5405 Posts
On September 10 2015 01:10 Big J wrote: Interesting one, do you happen to know whether this is the current state of the art? Also, does Dijkstra's algorithm find the closest mineral patch amonst multiple ones though? Because the way I understand it what it does is that it finds a way to one specific target and once it has found that target, it also knows the shortest way to the target. But I wonder if it always finds the closest target first if I gave it multiple possibilities. But yeah, obviously they are using some more advanced algorithm than what I painted. One that is using the underlying board-structure of maps and maybe other SC2-specific stuff like Dijkstra's. I don't think we can assume they use advanced algorithms. They probably just take the minimum distance nodes checked within a region and in case of ties they may very just pouch the one that happens to be first in their vector. I guess we should try to place the minerals in different order in the map editor and see if the mile always good to the first place one.  not sure that the map making order will be retained to the mile search of course, they probably resort them by x-y coordinates or something well before that. | ||
|
TheWinks
United States572 Posts
On September 10 2015 01:27 Tenks wrote: I wish the MULE would get called down on the SCV rally point instead of using some way to scan for which patch to drop the MULE upon If their goal is to eliminate the screen movement and number of clicks, then they shouldn't do that. It should just call down on the patch with the highest mineral amount. If they want to let you pick the mule calldown spot, it should fall on the highest patch in the currently rallied mineral line. | ||
|
Charoisaur
Germany16009 Posts
| ||
|
Big J
Austria16289 Posts
On September 10 2015 06:51 Charoisaur wrote: having to relocate orbitals at mined out bases is a nightmare. terran macro surely hasn't become easier with this patch. The idea is that you simply don't. If you start relocating to get extra value you are either doing it wrong, or the patch didn't nerf it hard enough. | ||
|
cheekymonkey
France1387 Posts
On September 10 2015 01:27 Tenks wrote: I wish the MULE would get called down on the SCV rally point instead of using some way to scan for which patch to drop the MULE upon There are several problems with this, if you misrally you will call down your mule somewhere random. And usually you will rally to nearly finished bases, and sometimes away from patches in case of harassment or attacks. But you don't want to mule there. There are just so many problems with auto-muling. As much as I hate the radius restriction, I'd rather keep it and have the mule going back to energy. | ||
|
Endymion
United States3701 Posts
| ||
|
Athenau
570 Posts
As far as muling goes, it almost certainly just choose the closest node that it can reach (in other words, the closest node for which the pathfinder can find a path). Unless you're in the habit of building sim-cities around your mineral patches, that's just the closest node. | ||
|
KT_Elwood
Germany1088 Posts
Zerg macro is all about not forgetting injects, and creep. Even under stress, even if not all injects are lined up. and can be be done 1,2,3. Doing multiple things under stress perfectly is the definition of the mechanic in SC2. It's hard when you are getting attacked some secondes before you had to inject. Yeah. So is forgetting 1 Round of marines from 12 Rax, because you just had to defend some mutas. But with Auto inject its SUPER FORGIVING. It's like Archon mode for 1v1. Protoss and Terran are given a faster rate at the income, at best. Zerg gets a fully automated Production, if you add a hatch and a queen, once in a while. That's plain BS. Its boring for Good zerg even in 1v1 it's making zerg archon obsolete. Yes its a mindnumbing taskt by it self, but it added balance to the game, in a sense that Macro was hard for every race. But since the Races play out so uniquely, you can not throw Mule,chrono and inject in the same category. Mule made up for slowest worker production, and slowest remax, and repair-Ability for burning Buildings. Chrono can be used to boost eco. Or sharpen timings, or getting even on upgrades...but it requiered decisions. Inject is stupid (almost) no decsion, but its unforgiving to those zergs who forget about it. And that was okay. Starcraft is an serial game. You can not give 2 orders at once, you have to do them one after another. Critical micro in two places + Macro is almost impossible. Inject was one of the critical tasks, you could always inject over micro, or you can go for the micro, and inject later. And once on the task of micro, bad players almost always delay their inject. TL;DR Unless you give terran the possibility to spent 4000 mins from automule instantly into 80 Marines, remove the auto-inject from the Game. | ||
|
Vanadiel
France961 Posts
On September 10 2015 19:57 Endymion wrote: blizz honestly needs to wipe all LOTV changes, go to dh9/10, and balance from there.. otherwise i'm convinced LOTV will be dead on arrival in korea That would be so good. DH 9/10 model (with a spice of Blizzard own change on the economical model, after all I do not think that they are incompatible), removed auto-cast stuff but maybe improve macro mechanics so these are less punishing if they really want it. I feel sad because I actually like all the new units except the cyclone and I feel they improved the game quit a lot, but as as KT_Elwood says, this autocast macro mechanics are a deal breaker for me. | ||
|
DooMDash
United States1015 Posts
I switched out liberators for medivacs to heal hellbats and pick up tanks, as well as mixed in some cyclones, but nothing can seem to compete. Any advice to mechers or Zergs? | ||
|
FabledIntegral
United States9232 Posts
On September 11 2015 15:55 DooMDash wrote: My TvZ mech went from a 75% win rate to about 10% since the Ravager change went live. Anyone else feel this was a completely ridiculous change? Feel like it hard counters everything mech has. I switched out liberators for medivacs to heal hellbats and pick up tanks, as well as mixed in some cyclones, but nothing can seem to compete. Any advice to mechers or Zergs? I've been playing Zerg recently with limited success with the ravager upgrade specifically due to medivac tanks. They simply move it over where I mortar and then drop the tanks and fire on my units themselves. Note: my opponents are using some, but notably less liberators as well as a result. | ||
|
DooMDash
United States1015 Posts
On September 12 2015 10:19 FabledIntegral wrote: I've been playing Zerg recently with limited success with the ravager upgrade specifically due to medivac tanks. They simply move it over where I mortar and then drop the tanks and fire on my units themselves. Note: my opponents are using some, but notably less liberators as well as a result. Yeah I've been trying this, and so far I haven't had much luck. I will keep trying though. | ||
| ||
