Send maps that break the analyzer to firstname.lastname@example.org, I will fix the bugs.
Special thanks to Louder, Wihl, PrinceXizor, Logo, monkh, chuky500, kraemahz, CharlieMurphy, konicki, Barrin, Orion-TF, funcmode and everyone else who is helping improve the map analyzer.
SC2 Map Analyzer 1.4.3 Released!
1.4.3 identifies the main chokes and calculates how much space is in each main base, counted in Command Centers (CCs, 25 cells each).
+ Show Spoiler [South on Steppes has more space to build] +
1.4.2 highlights destructible rocks in more output images.
1.4.1 fixes a bug when rendering the main2main shortest paths image.
To get started with the new release, download and unzip it. In the release directory, edit to-analyze.txt and output.txt with paths for your system. The comments within the files explain the options and provide examples.
1.4.x is a major update. The analyzer has been violently eviscerated and patched up with cybernetic implants:
+ Show Spoiler [Before] +
+ Show Spoiler [After] +
- The whole usage model has changed---the analyzer is no longer intended to ever be run as a command line tool. Instead config files control the analyzer and you double-click the executable (or a shortcut to it) to start it up.
- Map formats from beta phase 1, beta phase 2 and post-release are supported now.
- The analyzer's model for pathing is much more precise; just compare the ramps in the before and after above. The analyzer also recognizes pathing fills and the painted pathing layer.
- The footprints.txt config file allows pathing footprints to be defined for any unit or doodad. With some help from the community we can fill this file out until the analyzer is aware of every bitty doodad that can affect pathing. For now it has the resources, watchtowers and destructibles.
+ Show Spoiler [Analyzer 1.3.x Ladder Map Info] +
Want to know what the watchtower coverages are for the ladder maps? I knew you did! As always, I put the latest tables for the ladder maps in this Google doc. And what I mean by watchtower coverage is the percentage of pathable cells the watchtowers cover, not the percentage of the total map area. So yeah, watchtowers cover air, too, but I think the ground-cover measurement is harder for a human to judge by eyeball as easily. If you really want to know the other type of coverage I can add that to the tool, too, just let me know.
It looks like Blizzard's sweet spot for tower coverage is 20-25% which sounds reasonable to me. Now you can put a tower over 25% prime center space or 25% of no-resource corners or whatever, so don't use these numbers to cry about anything until you think about it. Just be careful when you make a map and your tower coverage comes out to 62.2% like Kulas Ravine! At least most of the towers are behind D-rocks to start, otherwise my first batch of Zerglings would just be chilling on those hogs.
Who comes in last? You might guess Incineration Zone, which is right (14.5%) but get this: Metalopolis is right behind at 15.6%! I was surprised at how tightly placed the towers are (either one scouts both golds, remember?). And here is a case where the coverage percentage is surprising, but with a worker at either tower you do have an eyeball on the only paths from one half of the map to the other.
In any case, if you're not sure about adding or subtracting a tower from your map, erring on the side of ~22% coverage like the rest of the ladder maps will at least be consistent with players' experience.
+ Show Spoiler [Original Map Analyzer Post] +
SC2 Map Analyzer was created to aid competitive players and map designers by computing all kinds of fun data about melee maps, WICKED! Map nerds gather round!
What is this crazy thing? You may have read the thread about my experiment comparing map rush distances which was apparently of some interest to the community because several people asked for more measurements. OK! SC2 Map Analyzer is the SkyNet-like next step!
Download it here. It's command-line (or a bit o' drag-n-drop) so you'll need a manual. We're having fun already!
If you want the long and juicy write-up of what internal map files the analyzer reads, and the gory details of how it computes all of its info, go over to SC2Mapster where I'm hosting the SC2 Map Analyzer project.
Okay, great, but what are we going to do with this thing? I'll tell you--ANALYZE THE LIVING SNOT OUT OF THE LADDER MAPS! ugh, ugh, ugh--HA, sicker than your average...
A ZIP of the analysis outputs (CSVs and PNGs) for all the ladder maps is here, the spreadsheets are all posted as a Google doc but you can just read on for now and I'll cover interesting bits I teased out of it already.
So let's get started!
+ Show Spoiler [Bases] +
SC2 Map Analyzer, I command you to find and count all the bases on a map and tally the resources and compute interesting information from it for me. NOW!
If you didn't already know (but we know you did) Blizzard's convention for gold expansions is to place only six mineral patches, so even though they mine at a similar rate but for less workers, overall you get less total minerals.
And did you realize the semi-islands on Scrap Station only have six regular mineral patches? Or the semi-islands on Kulas Ravine only have seven regular patches? Islands on Lost Temple are 7 regular, too. And every other base on the ladder maps is the standard 8 patches. Nice to know, like realizing that the island expos on Desert Oasis are as good as ladder island expos get.
Or check this: Kulas Ravine has the most total minerals and gas (156K/70K) but is tied for the lowest average minerals per base (16.1K, tied with Scrap Station).
That's a crazy stat, what's that for? I think what it tells you is that there is a significant disparity among the value of bases on Kulas, so if you are able to pull it off you should take only the richest bases (naturals and mains) as expos to get an edge in the long term. Or use the knowledge when you scout your opponent taking his semi-island as his first expansion--contain him on his low-rate/low total expo and get an economic lead. It's not like the disparity is so large you MUST play this way, but I think data like this can help you get the type of edge your opponent isn't even aware of.
Another interesting observation is that Blistering Sands has the lowest total minerals at 90K but the highest percentage of which is high-yield, 20%. What can you do with this info? Could you work out a strat to take the gold bases as your first and second expos while keeping your opponent at his natural? The data implies that a form of this strategy could work the best on Blistering Sands because the effort of controlling both golds is simultaneously giving you a more efficient economy and controlling a significant percentage of the total bases. Terran could PF both golds, Zerg could nydus entrace one gold, nydus worm the other. Of course, those golds run out quick--so you gotta strike while the iron is HOT! Will it work? Who knows! But look for possibilities in the data!
And now, an earlier experiment, revisited.
+ Show Spoiler [Shortest Paths, AKA Rush Distances] +
As I mentioned, I have already conducted an in-game experiment comparing rush distances of the ladder maps. SC2 Map Analyzer can do this job no problem, as well as compute the cliff-walk and air shortest paths, too!
SC2 Map Analyzer can also automatically find another important measurement: the distance from natural to natural, which some posts in the experiment's thread asked for in comparison to the main-to-main measurements. Let's revisit the conclusions from the other thread while consulting these SUPER FUN and EXCITING charts (last tab) that were generated by SC2 Map Analyzer.
Previous Conclusion: Incineration Zone is as short a rush as LT or Metalopolis, but you don't have to scout to find that out. Not a big surprise, but its nice to know for sure just how short it is.
Considering Nat2Nat: Incineration Zone nat2nat is just about the same as the shortest nat2nat on Lost Temple, the shortest nat2nat on Metalopolis, which are all roughly the nat2nat on Steppes of War. So from nat2nat perspective, Incineration Zone isn't so short after all. But here's the kicker: every nat2nat on Kulas Ravine is shorter than Incineration Zone! Do you really think about that when playing Kulas?
Previous Conclusion: Desert Oasis is a whopping 36% further rush distance than the 2nd longest, Blistering Sands.
Considering Nat2Nat: Desert Oasis is still the furthest, but not by much because Scrap Station nat2nat is a hair longer than Scrap Station main2main, both of which are close to the D.O. nat2nat. Combine this with...
Previous Conclusion: ...it feels to me as though Scrap Station has a significantly longer rush distance than Blistering Sands, when in fact Sands is a little longer.
Considering Nat2Nat: As Bigpon86 and PrinceXizor suspected, Blistering Sands nat2nat is much closer to the longer Lost Temple and Metalopolis nat2nat distances.
Other conclusions: Kulas is the 4-player map with the least variation in main2main and Metalopolis is the 4-player map with most variation in main2main. Considering Nat2Nat, all of this is still true.
Big Takeaway: Main2Main is probably meaningful for the earliest rushes, where Nat2Nat is more useful for planning something like a mid-game timing attack. And of course, having the hard data is good for answering any future questions you have about whether executing or defending a given rush might be affected by the maps.
Ground/Cliff-walk & Ground/Air Distance Disparity
In addition to the shortest distances by ground, SC2 Map Analyzer computes the cliff-walk shortest distances and air distances. Your intuition tells you when there is a big disparity between the air and ground distances (Desert Oasis!) but it's interesting to see that on most maps and most spawn configurations the cliff-walk and air rush distances are usually in the ballpark of only 25% shorter than the ground distances. This suggests the ladder maps aren't too complex in terms of winding ground paths and that's probably a good thing in general for game length and spectating both.
Which maps are the outliers when it comes to disparity between ground rush and cliff-walk/air rush distances?
Desert Oasis, not surprisingly, is 50.3% shorter by cliff-walk and 67.8% shorter by air.
Scrap Station is 60.6% shorter by air and only 2.1% shorter by cliff-walk until those rocks get knocked down, because there is no short cut for cliff-walkers to take. I think you can take an unusual observation like this and use it to reason about what an opponent is likely to do--say a Protoss going Colossus. "Hey, I'll be sneaky and use the Colossus to break the rocks!" Silly, rabbit, we knew you were thinking that all along.
Steppes of War is only about 5% shorter by cliff-walk or air, and it's the only spawn configuration on any ladder map that has this property. What can you do with that? I don't know--think about it!
Now let's break some new ground with SC2 Map Analyzer!
+ Show Spoiler [Openness] +
Openness - An attribute of a game cell stating the distance (in cells) to the nearest unpathable cell.
Openness, chokiness, whatever you want to call it is an important aspect of a melee map, but I'm fairly certain most players and map designers judge openness purely by feel (myself formerly included). What are we doing? Openness is something we can clearly measure! SC2 Map Analyzer, stop watching Nal_ra Old Boy reruns and get to work!
I never thought about how much asymmetry there really is in Metalopolis until I looked at this image of its openness. The mains are quite differently shaped and I'm going to have to extend Mr. Map Analyzer to specifically measure "space in main" to see if they balance out that way. Check out the lanes going past the watchtowers--the bottom lane has hints of blue and is wider than the top lane. It also looks like the open area near 8's natural is the most open, slightly. And even though we don't need openness to tell us, the gap between mains 8 and 6 is definitely wider than the gap between 1 and 2.
We can use openness like I just did to more easily judge a map just by looking at it, or we can crunch some numbers, too. Let's make it a quiz!
Which map has highest average openness?
+ Show Spoiler [answer] +
Which map has lowest average openness?
+ Show Spoiler [answer] +
You said Kulas Ravine or Incineration Zone, didn't you?!
+ Show Spoiler [actual answer] +
+ Show Spoiler [actual answer] +
Which map has the single most open area?
+ Show Spoiler [answer] +
Steppes of War
Don't forget all this raw data is available either by running the tool yourself or checking it out here.
Openness is really useful, I think, for comparing maps and managing asymmetry in new maps. Read on for another awesome use for openness!
An awesome use for openness!
+ Show Spoiler [Positional Balance] +
Positional balance is critical to a fair melee map, but let's be crystal clear--SC2 Map Analyzer doesn't even try to make any claims about racial balance. Period. So on with positional balance.
For the details of how SC2 Map Analyzer calculates positional balance, check out the analysis details page at SC2Mapster.com. The right-side navigation lets you jump right to it. Examples included, yummy!
An almost perfectly symmetrical map like Kulas Ravine has good positional balance:
No big surprise, but even Kulas isn't perfectly symmetric, and from the analysis we can see the little fluctuations have virtually no impact at all.
On the other hand, if a map has a different number of bases reachable by ground paths, you have problems. There were threads about this Novice version of Desert Oasis having destructible rocks in different configurations at the top versus the bottom, and even though we don't care about Novice maps, it's a good example of a map that has positional imbalance:
I haven't played a game on a novice map since I first loaded up the beta, so I may have never downloaded the newest version of this map. Still, what was Blizzard thinking? Did they intentionally inject positional imbalance to see if their stat collecting systems could measure it?
There are two ladder maps with slightly lower scores by SC2 Map Analyzer's reckoning: Incineration Zone and Metalopolis.
Incineration Zone is an odd duck because the map is a little wider than it is tall, so one spawn is on the shorter side of the rectangle and one is on the longer side. Now, in terms of distance from the start locations to bases, wackily this map is dead-center 100% balanced. The positional imbalance comes from the slightly differently shaped areas, by openness, of the bases. So it's not even clear which spawn is better or worse, but in terms of positional balance there is a significant difference between the shapes of bases. So actually, as a community we may decide Incineration Zone is actually perfectly balanced, but we're at least aware of how much variation it has compared to other maps. Keep an eye on this sucka!
Metalopolis has the worst positional balance score at 84.2% in favor of the 8 o'clock spawn versus the 2 o'clock spawn.
Here is the map with axes of symmetry manually added by myself. The gold expansions and watch towers form a perfectly symmetrical center arrangement where the axes cross at 45 degrees from horizontal. Okay, but look at those third expansions at the top and bottom. The top expansion's ramps are nicely centered, but the bottom expansion's whole land mass is shifted towards 2 o'clock. It's not the distance to bases that causes the imbalance, but a difference in the shape of bases on the halves of the map.
In the end you have to consider the data carefully. I'm not saying Metalopolis is garbage and someone should take a wrecking ball to those skyscrapers. This analysis, I think, is best taken as a guide. Think about your strategies and the data can inform you which maps they might work better or worse on, or help you debunk myths you have about which maps are the longest rushes, etc.
That said, you always have to consider the intangibles, too. DJ Wheat and Chill keep discussing with Weapon of Choice callers which spawn positions feel the best, and for whatever reason almost everyone likes the bottom-right spawn on Scrap Station (I think the reigning theory is that the wedge-shaped field-of-view maps nicely over that spawn's main area). If you consult SC2 Map Analyzer, it'll tell you the base shapes are very much the same, but the top spawn has an advantage by being slightly closer to the bases on its half of the map. Maybe that doesn't matter if your main fits in your field-of-view and it gets you in a zone, intangibly improving your play!
I hope that SC2 Map Analyzer is a method for the competitive community to extract some hard data when questions about maps come up, and of course it can be a valuable tool for designers creating new maps.
One last note: this stuff gets my engines firing on all cylinders. If you are a game developer and want me to do stuff like this for you with vigor and a smile, send me a PM because I'll need a job after grad school.
If you find a bug or want to request a new feature we can always discuss it in this thread, but I will consider the master list of issues to be the ticket system at the tool's project site on SC2Mapster.com: