Mac Gaming Mods

These tags are marked as "experiment with" tags.

I tested these experiments by running the beach storm in the beginning of a b30.map mod called "Beachhead" over and over again and seeing how the AI reacted differently each time.

So far, I've seen that:

-No matter how high you make the number, it caps off at 2147483647. For example, if you type "9999999999999999" and enter it, it will automatically make the number 2147483647. If you type 2147483646, it stays 2147483646. If you type 2147483648 it makes it 2147483647.

-Negative numbers cap at -2147483648. (Really weird, the last integer is only one value higher than the positive number cap.)

Number 1: Has to do with range at which the AI will shoot at an enemy from. The higher the number, the farther the AI will start taking shots at the enemy. The number was originally a very low negative number, but when I made it extremely high, the AI would shoot at an enemy no matter how far away it was. When I made the number 0, the AI seemed to act normally and still shot at the enemy but at closer range. The original number was -1287930578.

Number 2: After LOTS of testing, I have no idea what this does. The AI show no noticeable changes after each test, but it has to do SOMETHING. After maxing out both high and low numbers, no difference in behavior, though the enemies died in different ways each time. Original number was -1287930578.

Number 3: When the number is maxed negative, the AI acts normal. When the number is maxed positive the AI either prefers closer combat, or always rushes any enemy. When number is 0, AI acts normally. Original number -1287930578.

Number 4: The only different original number out of the four. The number is also the only positive: 1056964607. Same results as number 2.

EDIT:

-I maxed out the all numbers (positive), and the AI always rushed and always shot at the enemy no matter how far it was. The AI sometimes started shooting at nothing and at walls at times. I think they were trying to shoot enemies on the other side.

The reason for the caps is because they're treated as signed 32-bit integers (hence int32). Can't speak to the actual utility.

Thanks for the testing, Monk34!

You can check out those offsets in HexEdit.

Can you test the max vision distance, hearing distance and peripheral distance values in the actr tag and see if you get the same results when making those really high?

Thanks!

The AI used in the testing had INF hearing and vision distance. The first unknown tag apparently hightened the actual shooting distance. I'll do more testing with a regular version of b30 though.

EDIT: I made the marines in an unmodded version of b30 deaf and blind (Zero hearing distance and vision distance) and tested with Unknown #1 maxed and at normal. Both tests were identical: The marines ran past enemies as if ignoring them, but sometimes knew where they were and fired at them.

Monk34 wrote: EDIT: I made the marines in an unmodded version of b30 deaf and blind (Zero hearing distance and vision distance) and tested with Unknown #1 maxed and at normal. Both tests were identical: The marines ran past enemies as if ignoring them, but sometimes knew where they were and fired at them.

Significant proof that there is in fact a 6th sense.

Yeah it was weird. Once the deaf and blind marines somehow found an enemy, it shot at it until it died or until the enemy died. It used tracking and everything.

Maybe we should label it "Sixth Sense". What do you think?

The unknown tags didn't really have an effect, the deaf and blind marines did this regardless of what numbers were in the unknown boxes. Ill so some more experimenting though.

Could the "sixth sense" be some kind of motion sensor, or the equivalent of a red reticule?

^That's interesting. The problem is theres no settings in the actr tag that relate to a "sixth sense" unless it's in the actv tag or one of the unknowns. There are lots of those unknowns in both the actr and actv tags, and they're usually unreasonably large numbers.

Those unreasonably large numbers might need to be converted, like the health & shield levels of some old coll plugins, and the AR display.

Ok, so it seems I botched the deaf and blind experiment by only giving the major class the correct actr. I redid the experiment and the AI just ran forward not noticing anything that was happening. I'll fool around with the unknown tags and see what happens though.

EDIT: Blind and deaf marines still dive out of the way of grenades, and if you shoot them they will look at you. That's about as responsive at it gets.

Did the unknown values have any effect, and how did the deaf/blind test affect those values, if at all?

I only had time to test the first and second unknowns and so far there's no effect. I'll keep trying new things though.

I am actually very interested in this work you have been doing with the AI... specifically for this reason:

In my tests, I have found that it is impossible to spawn working AI in any map constructed entirely of scenery. They spawn perfectly fine, actually, but it is their behavior after spawning that makes them incompatible. AI spawned on top of scenery seem unable to run or walk or move in any way aside from jumping. I have tested this with elites and flood, and vaguely recall the possibility that something similar happened with marines when I tested a long, LONG time back in an early beta version of Paranoia, when I was deciding on easter eggs for it.