[SOURCEMOD] Brutal-OXCE 7.12.1

[human being]

I'm playing this with xfiles and half the zombies are just hiding and running away when I come close.

[Vakrug]

Can we just agree that seeing enemies  using medikits to help one another is nice?

No, absolutely not! I saw that in Phoenix Point. In 99% cases enemy using medikit is an (another) act of suicide. And after seeing it more then 10 times it is even no longer amusing. AI in XCOM is not even closely smart to use medikits effectively. Just like "close quarter combat" mechanics is always beneficial to a player, enemy using medikit also will be.

- and more importantly, civilians! -

Only doctor civilians. And those doctors should heal any humanoids, not just player units.

[Xilmi]

I'm playing this with xfiles and half the zombies are just hiding and running away when I come close.

Thanks for the report!

This was not intentional. Since the Zombies use the isLeeroyJenkins-Flag, they should indeed always be moving forward. It's fixed in:

Brutal-OXCE 4.1.2 - Zombie Fix Patch

Updated base to OXCE 7.8.15.

AI now also watches the energy of their units when peaking as to not get stranded out of position due to a lack of it.

Fixed a recently introduced issue leading to the isLeeroyJenking-flag no longer leading to units always acting aggressively.

Now using the armor of a blockage in order to tell whether walking into a proximity-mine should be save or not.

[Xilmi]

AI in XCOM is not even closely smart to use medikits effectively.

Well, one of the reasons why I didn't already do it is because in order to do it effectively, there would have to be made a lot of situational considerations.

It's not just "when someone is wounded, try to get close to it and use the medikit."

If they are still near a fight it might very often be the better idea to participate in it and only do the treatment when the threat is gone.

Also the wounded unit and the healer need to coordinate where they do the treatement.

[haja0011]

Nice mod. I watched some Streams from Marbozir and another guy. The AI is now very powerful.
But in terror missions the civilians have no chance. Is it possible to enhance their behavior as well to hide?
So that it is more likely to rescue some? Maybe as option or so?

Regards
Jan

[Xilmi]

and another guy

Do you know whom and/or what it was? I wanna see the vod.

Also about the civilians: There is an option for letting Brutal-AI control the neutrals. This should make them try to hide but they don't know about running towards X-com. The logic also has to support stuff like when they are a third allied faction, as it happens in WH40k-mod, for example or some other mods, where they start with weapons.

[Vakrug]

This should make them try to hide but they don't know about running towards X-com.

Some of them are smart enough to climb X-COM vehicle, grab the biggest gun they find (including those that took month for brightest minds of the planet to figure out how to fire), then try to shoot an alien just to hit one of X-COM operatives in the back. On the positive side, after the end of the mission that gun don't registered as stolen.

[Xilmi]

Yeah, bringing some extra-weapons to distribute them to the civilians is something that you can do, if Brutal-AI for neutrals is enabled. Deliberately or not.

What I'm working on right now is "carefully progressing". This will mainly be an improvement for the auto-play-AI but the aliens shall also have some benefit from it.

It consists of an enforced investment into peaking while saving 25% of the available movement resources (TUs and Energy) and if nothing is found spend the rest of the TUs for getting in the position with the best cover.

This also required me to do some energy-management. Where they won't do peaking if the energy is too low and they are in a save position. This currently has the issue that they then sometimes go backwards to where they came from last turn as the best cover is now there. So units with low stamina are not really progressing. So if they think their position was "safe enough" to recover energy they also should skip the get to best cover-part.

For the enemies, this new technique shall replace the ON/OFF-flipping of the sweep-mode by having a dynamic ratio of moving forward vs. saving TUs for hiding. The sweep-mode was a bit too extreme in how much they'd expose themselves. So having something that ramps up more smoothly and only acts like the sweepmode, when the enemies morale is completely broken and they are basically guaranteed to panic seems better.

However, during my tests I found one big issue with peaking in general: If it doesn't reveal anything new because your friends already peaked into the same area, it's just a waste of TUs and energy, that could have been used for reaction-fire. For example, when breaching an UFO-entrance right now and hiding from them that they won't see you when they open the door, you can mostly count on everyone in the entry-hall of having no TUs for reaction-fire. Because all of them probably peaked to the door and then went back to a better spot. It's not peaking in general that is useless but peaking something that has been peaked at in the same turn by someone else. So I need to find a way of how a unit can identify whether someone else has already peaked the same stuff they are about to peak and if so stop them in their tracks. This won't be easy but I think it's worth the effort.

[Xilmi]

I got the "will my peaking reveal anything I don't already know?"-mechanism to work yesterday.

However, it's a bit contradicting the goals of the abolishment of sweep-mode. When a higher ratio of peaking vs. hiding is a way to advance but peaking can be discarded alltogether, then units will move forwards more in piecemeal instead of a combined effort.

I think that when the AI is already aware of at least one unit or has the conditions that previously triggered sweep-mode, it's mode of operation shall be one that is more advance-friendly.

Not only should it spend the peaking-TUs for moving forward, it should also not put so much weight into cover, that it'll go back to where it was before. Like it should have a much lower threshold for how good the cover needs to be compared to how it values being closer to the enemy. Maybe I can use the same morale-ratio as a multiplier for the cover-quality-requirements.

[Xilmi]

Brutal-OXCE 4.2.0 - Combat AI Enhancement Patch

Fixed an issue that could make the game lock up during auto-play.

AI is now able to determine whether expending the TUs for peaking is a worthwhile investment depending on whether it would actually reveal something new.

AI now can decide to move forwards even without being able to discover anything new, when some of their units are already engaged in combat or in the situation where it earlier would have gone into sweep-mode.

AI will no longer refrain from peaking based on the maximum range of their weapons based on their current TUs but based on what it would be if it had full TUs.

Instead of sweep-mode being a hard on/off-switch the AI becomes gradually more willing to move forward faster as it gains a morale-advantage. It will usually still try to find some cover.

Non-aggressive-forced auto-play-AI has a fixed value of how much it wants to spend on moving forward vs. saving to hide. It will also always consider being in combat as to not leave units behind.

AI will now take their energy into account when it considers peaking and rather not do it, if the energy is too low.

AI will now move in to attack instead of pre-priming a grenade if it can attack.

Fixed an issue where AI would not always turn towards a door it wants to open.

Melee-units now take potential reaction-fire into account too and might refrain from peaking. It should not prevent them from trying to move in for a kill though.

When trying to hide always prefer locations with cover over those without and where the enemy couldn't reach.

The regular cover-seeking score which is usually used to get into some cover after moving forwards and not finding anything to attack will now prefer to conserve TUs and no longer has redundant score-modifiers.

AI shall more actively try to avoid standing in the blast-radius of its own grenades or in fire.

[panzer]

This is a really cool AI improvement. Now every fight is a real challenge)))
The only bad thing is that AI rarely ambush and
reserve TU for return fire.

[Xilmi]

This is a really cool AI improvement. Now every fight is a real challenge)))
The only bad thing is that AI rarely ambush and
reserve TU for return fire.

Can you elaborate what your understanding of an ambush is? According to my own understanding of the word, I'd say that the AI in 4.2.0 definitely does this.
The basic idea is that if they attack the player too close to the Skyranger or other starting-zone, the player might just flee or in case of the Triton be tempted to use the invulnerable cover of it to their advantage.
So what they do is to use the information of whether a unit was spotted far enough away from the starting area that it shouldn't be able to get back to it within one turn. If not and they also don't see anyone currently, they'll try to make more space between themselves and the starting-zone. Then they wait for contact somewhere and reevaluate. If they find units exposed far away from where they start, they shall try to attack it and will even hunt them down. That's my understanding of an ambush.

When it comes to reserving TUs for reaction fire it is important to realize that there's a game-mechanic called "mutual surprise". It means that when two units see each other at the same time, the one who's turn it is always gets to decide on what to do first regardless of reaction-score. You basically need to see the other unit spending TUs in order to react, not just spot them. When the player knows about that, there's quite a bunch of ways of avoiding reaction-fire. Like avoiding night-missions where the aliens have a sight-range-advantage, always covering your path with smoke-grenades and using the 45° cornering-technique. Therefore having the AI rely on reaction-fire would make it overall weaker. I took some measures to make it preserve TUs if spending them would serve no purpose so there should be cases where reaction-fire happens. But also cases where you'll get the first shot due to the mutual-surprise-mechanic despite them having enough reaction-score.

Note that the AI is also aware of mutual-surprise on it's own turn. It'll not consider a path as dangerous in regards to reaction-fire unless it passes through two watched over tiles in a row.

[panzer]

When it comes to reserving TUs for reaction fire it is important to realize that there's a game-mechanic called "mutual surprise". It means that when two units see each other at the same time, the one who's turn it is always gets to decide on what to do first regardless of reaction-score. You basically need to see the other unit spending TUs in order to react, not just spot them. When the player knows about that, there's quite a bunch of ways of avoiding reaction-fire. Like avoiding night-missions where the aliens have a sight-range-advantage, always covering your path with smoke-grenades and using the 45° cornering-technique. Therefore having the AI rely on reaction-fire would make it overall weaker. I took some measures to make it preserve TUs if spending them would serve no purpose so there should be cases where reaction-fire happens. But also cases where you'll get the first shot due to the mutual-surprise-mechanic despite them having enough reaction-score.

Note that the AI is also aware of mutual-surprise on it's own turn. It'll not consider a path as dangerous in regards to reaction-fire unless it passes through two watched over tiles in a row.

Thanks for the answer and clarification. But gosh, it's a lot harder than I thought.)
The fact is that I play The X-Com Files, there are a lot of corridor maps. Enemies, knowing/guessing that my agent is around the corner, instead of entering the line of fire and reserve AP, they prefer maximum rapprochement and eventually die from a point-blank shot without action points.

[Xilmi]

The fact is that I play The X-Com Files, there are a lot of corridor maps. Enemies, knowing/guessing that my agent is around the corner, instead of entering the line of fire and reserve AP, they prefer maximum rapprochement and eventually die from a point-blank shot without action points.

I'm not quite sure I fully understand the situation. Would really be nice to see some footage of what is happening so I can analyze it better. Or maybe a savegame that showcases what you mean.
The enemies get into line of fire and then get close to you but don't attack anymore so you can then shoot them on their turn?
Are those melee-enemies? Or enemies with very short-ranged weapons? Or enemies with impaired vision?
I'm not quite sure why they wouldn't either attack you immediately when they have line of fire or at least only get close enough until they get a better chance to hit.
Note that enemies, that are flagged as "isLeeroyJenkins" in the mod will alway try to get into melee-range, no matter what.

But I'd definitely need more information to determine whether what you are talking about is intentional and or how it should be imroved.

[SIMON BAILIE]

Is there a problem with v 4.2.3 as when I equip a craft it is showing equipment on the troop screen at base that according to the first picture isn't aboard. Doesn't happen with v 4.2.2-see attached screenshots.