Page 1 of 1

Muta stacking command processing and other questions.

Posted: Thu Jan 19, 2017 7:06 pm
by kovarex
Muta micro
So I started to do my own bot, and now I'm in the phase of the muta micro implementation.
I expected the muta stacking part to be easy for the bot, but for some reason, it doesn't work that well.

I have a test map with 11 mutas and overlord in the corner, I always order all the mutas together with the overlord in a group and try to stack them by moving around.

The only strategy that works for me now to stack them really tight is to issue commands around the center of the mutas in a circle. The problem is, that it takes quite a long time to stack them, and when I move around after that, they tend to unstack quite fast (one or two attacks and they are too unstacked). At this phase, if I had to start stacking them again, it would be highly ineffective as I would spend most of the time stacking the mutas.

I know that it is possible to do it much better, as shown in this video of killer bot from 2013:

The questions are:
a) Is there source code available of this bot?
b) Are there replays available of the bot doing the muta micro, so they could be analyzed?
c) Might it be required to move individual mutas and take latency and future position into consideration to keep them stacked all the time? This would increase the APM requirements per group of mutas a lot. Also, what is the maximum APM that is considered to be OK?

Command processing
How exactly does it work in starcraft frame by frame? When I tried to issue attack command that was followed by move command in the very next frame, the attack command was always ignored in my tests. Aren't the commands processed in batches (of 2 probably)?

Re: Muta stacking command processing and other questions.

Posted: Mon Mar 27, 2017 7:05 pm
by bftjoe
I thought that the method that bots use is to tell the mutas to move to the same spot, they don't micro like humans.