I just want to show you all a quick guide on how to set up mineral boost patches as Zerg because I don't see many non-Koreans doing this trick.
In this video I show a couple of example patches for how mineral boosting works when the minerals are on the left and how to do the return cargo setup for minerals on the right side of the hatchery!
It's a little troublesome to explain so if you have any questions I might be able to answer them or someone else here may know also!
You'll find that typically when the minerals are on the right its always the far top right patch and the far bottom right corner patch that you can use the return cargo trick with, so experiment on different maps until you understand which patches to do this on!
A simple trick to get a small advantage in the early game especially if you split your drones efficiently.
Thanks for posting this - when you first told me about it I was still a bit confused, so I just never ended up using it. I still find it bizarre though.
The return cargo thing seems like an awfully over-specific (and probably suboptimal) solution to the common problem of metastable worker paths (as I have come to call the phenomenon). The crucial point here is that workers mining at a Hatchery (not necessarily Drones! It is the specific collision box difference between the resource depots that is responsible here for racial differences) get a faster turnaround when they approach those specific diagonal Mineral fields from the corner (as opposed to the front edge). Hence this is not possible on mineral lines where those patches lack a free corner, like in FS mains. Probably the fastest way to achieve this is the following: + Show Spoiler [show schematic] +
Step by step: (i)…(iii): Split the closest workers to the closest patches (3,4,6). Don't split to (1) directly, because it is too far away from the worker spawn to enable a fast first return, and the difference between a straight path (patches (3,4,6) and a a fast diagonal return (1, as desired with regards to the OP) should be minimal, with straight paths being shorter and hence tending to be slightly faster in general. [It might be ever so slightly faster to send (iii) to (8) and (iv) to (6) instead, but at least I seem to lack the proper reflexes to pull that off consistently]. (iv) Send the farthest worker to patch (7) (v) When it reaches the yellow circle, send it to (8). It should stop and start mining the patch from the corner immediately, which not only gives us the desired approach angle but also cuts out the deceleration time (This is an interesting article for further reading on this). (vi) After the first return, once the topmost worker [from patch (3)] has dropped its mineral load at the Hatch [at the green circle], order it to go to (vi). (vii) Once the worker reaches the other green circle, order it to mine patch (1). The same principle as for patch (8) applies. (viii) Send your 5th worker to mine from patch (3) again. (ix) next best patches to mine from are (5,7,9). The exact order should not make too much of a difference with these. (2) should be used last as it is far away from the worker spawn and the slowest mining patch in general.
EDIT: After some additional testing on that CB position specifically, I figured out that the desired angle is not quite the corner of the patch but a few pixels below it, with quite a bit of leeway, so it still easy to hit manually via a move order with some practice, which is still faster than having to rely on an intermediate long return...
liquipedia disagrees with the bottom left mineral on the left mains being bad, as does my own data below (number on the mineral patches is what your drone priority should be, green indicates first 4, then yellow is next few, then red)
the trick is useful for patches on CB and the right of FS though i think, i will mess around with it
liquipedia disagrees with the bottom left mineral on the left mains being bad, […]
I think those graphics should just be removed from the article as they are misleading. It is not even documented how those numbers were arrived at and the presentation suggests that the creator of those graphics made several naive assumptions which any one willing to spend a few minutes in-game worker microing with different races can easily disprove:
That specific worker behaviour and mining rate depend only on patch position, not the race (or more precisely type of resource depot present), whereas in reality this can have a huge impact (some specific instances shown in the OP, where single worker mining from certain patches can be permanently sped up through precise positioning, making these patches first priority patches for Zerg, and only Zerg, specifically)
That pathfinding is location-independent and thus top and bottom mains can just be considered identical (it is not, the specific layout of pathfinding regions around the mineral patches, resource depots and worker path can make a huge difference
[…] as does my own data below (number on the mineral patches is what your drone priority should be, green indicates first 4, then yellow is next few, then red)
the trick is useful for patches on CB and the right of FS though i think, i will mess around with it
Same mistakes here: No documentation of how this was generated make it worthless data. How were those Best/Worst values come by? Is it for different worker angles? In that case: For which race? Or is it for different races? In which case: How were worker angle variations accounted for, if at all?
my data is just for zerg. i timed the return time on each trip after sending them from the starting position, i didn't consider different angles because my understanding is that after the first trip/return it will always reset to a standard angle (a false assumption i think given the info in the OP's video). the best worst values are just the highest and the lowest times, to give an idea of the variance that i was looking at.
edit~ also all 1.16.1, i haven't tested any of this on scr
On January 19 2018 20:22 Endymion wrote: my data is just for zerg. i timed the return time on each trip after sending them from the starting position, i didn't consider different angles because my understanding is that after the first trip/return it will always reset to a standard angle (a false assumption i think given the info in the OP's video).
There are many things that can happen, depending on patch position, resource depot type, other patches in the vicinity blocking certain angles and pathing region layout:
Some patches will naturally default to a stable standard path from the second trip onward.
Many patches have two or more meta-stable paths that workers will stay in from their second trip onward, depending on their initial positioning. Different metastable states on the same patch can vary widely (especially around region borders) and worker migration at higher saturations can force workers out of particularly good paths (like the ones shown for CB in the OP) or, contrarily, out of particularly bad (bugged) mining paths (thus creating self-mitigating mining bugs which do not necessarily need to be addressed in map debugging).
Some few patches produce periodic paths where worker mining position cycles through two or more discrete states over multiple mining trips (this has generally to be considered a mining bug because at least one of the paths tends to be a really slow one).
the best worst values are just the highest and the lowest times, to give an idea of the variance that i was looking at.
So you used a stop watch or something like that, not frame-counting or long time mining averages? Workers on a specific stable or metastable path should not have any variance in their mining speed!
edit~ also all 1.16.1, i haven't tested any of this on scr
Makes no difference since pathfinding is exactly the same.
yeah it was just manual timing via stopwatch, i don't know enough about broodwar's api (or whatever the bot writers use) to write something to count the frames. i did it mainly for ZVZ since that's the only time low workers materially have an impact on the outcome of the game, but i'm not really sure how we would measure the individual paths (or how you would even take advantage of that really, you would need to know how to enter into the specific patterns). in my anecdotal experience on live fighting spirit games though, the data works when you issue the order directly following the egg hatching on the left side of the hatch. after drone drilling/transferring drones i haven't seen massive problems either, but at that point i normally have more than 9 drones so it's harder to keep track of mining efficiency.
however, I always prioritize sending drones first for the "green" patches in my data even if i have 12+ drones. so say the #1 patch finishes mining and i have another drone returning to the adjacent patch. i'll send the adjacent patch drone to the #1 patch and send the original #1 patch drone to the adjacent patch, which i'm pretty sure speeds up your mineral income rate. However, i have seen the different paths that you're describing mess up the optimization and probably result in a wash or a decrease in overall efficiency so i'm not 100% on it. it would be nice to have some more hard data for sure
I've never used this specific tactic, is this only for drones or hatcheries, or does it work for terran? Toyed around with it for a while but couldn't see a difference.
Thanks for the video, I didn't know this, it will be useful!
I also don't see many players do what I call "worker cycling", where you manually send workers to the patches that are going to be free soon, so you rotate 1 worker out of like 3-4 on 2-3 minerals (depends on the positions), to decrease their wandering time. It's only feasible to do this while there's nothing else going on though, but it will give you a small mineral boost. It's possible to do multiple of these simultaneously. It's even better if you manually move the drone so it doesn't slow down. I've seen Flash and others doing it.
On January 22 2018 09:49 quirinus wrote: Thanks for the video, I didn't know this, it will be useful!
I also don't see many players do what I call "worker cycling", where you manually send workers to the patches that are going to be free soon, so you rotate 1 worker out of like 3-4 on 2-3 minerals (depends on the positions), to decrease their wandering time. It's only feasible to do this while there's nothing else going on though, but it will give you a small mineral boost. It's possible to do multiple of these simultaneously. It's even better if you manually move the drone so it doesn't slow down. I've seen Flash and others doing it.
Very common on higher levels, especially in games that matter. And only relevant in first minute of a game basically. There is also a downside of it, when you manually change the path for an SCV, it will "freeze" for a quick moment and it acutally takes it longer than if uninterrupted. So if you are 100% sure about where which scv will go to find a vacant mineral, go for it, but if there is little doubt I would just let it go. (what I mean is when you have 10-12 SCVs sometimes you send the SCV to the mineral that is still being mined and the SCV arrives just when other SCV leaves)
On January 22 2018 09:49 quirinus wrote: Thanks for the video, I didn't know this, it will be useful!
I also don't see many players do what I call "worker cycling", where you manually send workers to the patches that are going to be free soon, so you rotate 1 worker out of like 3-4 on 2-3 minerals (depends on the positions), to decrease their wandering time. It's only feasible to do this while there's nothing else going on though, but it will give you a small mineral boost. It's possible to do multiple of these simultaneously. It's even better if you manually move the drone so it doesn't slow down. I've seen Flash and others doing it.
it's material in zvz and low income games when you're talking about an extra drone somewhere. i'm not sure how effective it is, but i find it more fun to do in the early game compared to just spamming my apm
I made an updated video on zerg mining techniques and efficiency, its not super complex but maybe slightly more advanced than before. https://www.youtube.com/watch?v=4Us3GWUa9c0
On January 22 2018 09:49 quirinus wrote: Thanks for the video, I didn't know this, it will be useful!
I also don't see many players do what I call "worker cycling", where you manually send workers to the patches that are going to be free soon, so you rotate 1 worker out of like 3-4 on 2-3 minerals (depends on the positions), to decrease their wandering time. It's only feasible to do this while there's nothing else going on though, but it will give you a small mineral boost. It's possible to do multiple of these simultaneously. It's even better if you manually move the drone so it doesn't slow down. I've seen Flash and others doing it.
Brood war bots (the ones created with the BWAPI) use this a lot. I made a video about this a while ago:
As koget mentioned, as a human player this is only feasible to do in the early stages.
EPT![[image loading]](https://i.imgur.com/qiONuOX.jpg)
File:Fighting Spirit top and bottom left main.jpg![[image loading]](https://i.imgur.com/3dP3CnR.png)
