Skip to Content
 

Using the computer to create units for me

4 replies [Last post]
larienna
larienna's picture
Offline
Joined: 07/28/2008

One problem I always have in game design is designing the data. I can design the rules, like how units battles, but selecting which values those units have is complicated. There are various ways that I know how to deal with those situations, but feel free to give additional suggestions:

A) Improvise, give semi random values and make some tests. The problem is that I might be a bit too much OCD to be able to completely improvise. I am constantly thinking balance in my head.

B) Give all units the same stats, then vary slightly as the design progress. This solution could be more viable, as you only need 1 unit to design, and then you introduce new stats variation one by one as you make more tests. Still those variations would require at some point improvisation.

But I was thinking of another way, make the computer chose for me. The idea is simple. I wanted to create a computer program that simulates battles between designed units to determine the efficiency of each of them in order to balance the game.

Now I could do the other way around, generate units randomly, make them fight each other and keep the most efficient designs. Still, it will select the most efficient units according to the currently implemented rules, change of rules will require re-battling the units.

Maybe I am overthinking too much and should just use method B, unless you have other ideas for people who cannot improvise.

X3M
X3M's picture
Offline
Joined: 10/28/2013
I design with a weight factor

I design with a weight factor in mind for all statistics.
Then I can easily calculate their value.
I have a balance of 99%.
The last 1% is done through testing.

larienna
larienna's picture
Offline
Joined: 07/28/2008
Once I have a set of units

Once I have a set of units with working rules, I could make combat and probability simulation to determine what is the weight of each stats.

But I am not there yet, I am at early unit design. I need units to test if the rules works. Then determine which changes should I make to the rules.

X3M
X3M's picture
Offline
Joined: 10/28/2013
Make a good plan

If you have all statistics set to 1 time a number.
And then double a number.
Do the same test, but have the number be 3 times as high.
Or even 4 times as high. Maybe the effect is exponential.

Another thing to check is, what if a statistic has synergy with another statistic.

I have seen this with fodder against tanks. So, it is something to keep an eye on.

But yeah, having a basis. And then work from there can work.
Sometimes, if the effect is twice as good. It is also possible that the adjustment has to be the square root out of 2.

Just giving some pointers to look at.
So, when you think you got a weight, test it first by actually applying it.

larienna
larienna's picture
Offline
Joined: 07/28/2008
I tried to start with 3

I tried to start with 3 simple stats:

Casualty: Does the weapon KILL or STUN
Target: WHO select the target, the PLAYER or the OPPONENT
STR: a value between 1 and X, which is the TN to roll under to hit.

Now the STR value is the counter weight for the rest of the stats. So I get here 4 possible combination:

KILL + PLAYER
KILL + OPPONENT
STUN + PLAYER
STUN + OPPONENT

STR will be adjusted to how efficient the mechanics above are.

I could simplify even more and remove STUN, because the consequence are more long term than short term. Unless STUN can prevent units to attack in the same battle. I will have to think about it.

Then I could generate battles, using an average TN of 5, see which unit is more efficient via stats, then make adjustments in consequence. I will need to determine what statistics should I be recording.

Maybe I could only compare final battle outcome in matter of casualties and victory. Maybe both side have all the same units. Then add 1 new unit to the pool and see if the combat results change. The problem is because it's group battle, it's hard to know who is responsible for the outcome.

Then I could add 1 stat at a time and see the impact on the casualties and victory as this is in fact what I am interested to know: Who dies (or which side lose more units) and which side wins.

Syndicate content


forum | by Dr. Radut