An Orcish Incursion, Revised [Mainline Campaign]

[Dalas120]

Six and a half years after being de-mainlined, a classic campaign returns...
---- An Orcish Incursion, Revised ----

Orcs once again trouble the Great Continent, and this time they don't need Lich-Lord Jevyan to order them around. Marshal the elves of Wesmere and lead them to victory during their first encounter against this alien new foe.

Available on the ***1.19 add-on server ONLY***. This LoW prequel is slated for mainline in 1.21, and features an experimental new 'regroup' [micro_ai] - please send a replay if you find an AI exploit or encounter strange AI behavior. Credit to @Inferno8 and @Durzi for some visuals.

There's a couple AOI add-ons floating around on the 1.19 add-on server. This announcement is about the ***add-on by Dalas***.

.

[Dalas120]

v1.0.1:
S02: enemy AIs are now less likely to ignore the player
S05: (hopefully) fixed a bug where the enemy leader randomly healed
adjusted some AI parameters

v1.0.2:
S03: increased turn limit by 2, added a couple more villages
S06: fixed a serious issue where your invisible ambushers would sometimes get attacked

[redbeard2]

Playing through it, mostly done, I'm in the swamp at the moment. It is, of course, very similar to the original, at least up to this point, which is fine. I guess the biggest thing I would comment on this point is the AI is quite...interesting. I am very used to the computer rushing you and almost never falling back, there have been multiple moments where I expected an all-out assault, or even then to just walk out and pick off a lone unit that I sent ahead, and they often wouldn't do it. Maybe related to time of day? it was a bit hard to tell.

Scenario 2 for example I noticed the northern enemy, once I got close, turtled up in the castle, and prettymuch forced me to wade into the moat to storm in, which is not something elves are great at doing. Something the typical AI almost never does.
In contrast scenario 4 that was where there were the most instances of the orcs falling back when i wasn't expecting it, even when they had a fairly easy opportunity to pick off a unit if they'd just commit. After a bit of back and forth (me based on time of day, them doing whatever they were doing (they did score some kills, they generally weren't just wandering around let me be clear)), they ended up leaving their camp far more open than they should've, a couple of their units did end up just wandering into the western half of the camp, away from where I was attacking. Maybe because it was day, but it certainly made that final assault and striking their leader far easier than it could have been, especially given how they behaved in scenario 2.

In a way it's a good thing to have a contrast and not be able to read the AI so easily, I'm having a hard time fully understanding what the AI's directives here are, it seems somewhat day/night based, but I feel if they'd just rushed me in a few cases, I'd have been hard put to stop them without losing a bunch, especially given how few forests there are in scenario 4.

Playing on Hard by the way

[Dalas120]

Thanks for testing!

I guess the biggest thing I would comment on this point is the AI is quite...interesting. I am very used to the computer rushing you and almost never falling back, there have been multiple moments where I expected an all-out assault, or even then to just walk out and pick off a lone unit that I sent ahead, and they often wouldn't do it. Maybe related to time of day? it was a bit hard to tell.
[...]
In a way it's a good thing to have a contrast and not be able to read the AI so easily, I'm having a hard time fully understanding what the AI's directives here are, it seems somewhat day/night based, but I feel if they'd just rushed me in a few cases, I'd have been hard put to stop them without losing a bunch, especially given how few forests there are in scenario 4.

Very broadly speaking, the new micro_ai tries to avoid moving into areas where you greatly outclass it (as determined by unit cost, with various modifiers for injury, ToD, and scout vs frontline units), and will instead retreat/regroup towards villages, allies, or its leader. At poor times of day the AI is drastically more cautious, not necessarily because I think it's stronger that way, but because I think a bit of push-and-pull is fun to play with.

So even if the AI can possibly get the kill on one or two of your units, it might rate your likely counterattack on the AI's own exposed units as too dangerous (especially during the day) and still decide to retreat - at least that's the way it's supposed to work.


One of my aims with this micro_ai is to help the AI defend its leader better, not worse. So this is what I'm trying to make it do more of (when it realizes it's outnumbered and losing):

Scenario 2 for example I noticed the northern enemy, once I got close, turtled up in the castle, and prettymuch forced me to wade into the moat to storm in, which is not something elves are great at doing. Something the typical AI almost never does.

And this is what I'm trying to make it do less of :

they ended up leaving their camp far more open than they should've, a couple of their units did end up just wandering into the western half of the camp, away from where I was attacking.

If you can send me replays showing (and/or saves just before the turn) when the AI does something stupid, I can turn on debugging and pick its brain to see why it did that stupid thing, then hopefully improve it. For example, maybe I need to weigh expensive units as less valuable - a Royal Guard costs more than 5 Spearmen, but is it actually superior in combat strength? Or maybe I need to set up new goals/targets for enemies near our leader; that sort of thing.

[redbeard2]

I sent a message with replay and my commentary, hopefully it came through and is usable.

The AI certainly creates a fascinating back and forth, perhaps my biggest worry is it might be possible to exploit by making some reckless moves to inflict damage right as an enemy retreats based on time of day, and they will fall back even if you leave a bunch of vulnerable units in the open that they could go after and potentially really devastate you if they were more willing to persist.

[Dalas120]

Sorry for the delayed response! I did see your message, but was waiting until I'd worked on each point before responding.

Your notes have been very helpful, thank you. Having both a replay and a turn-by-turn commentary is especially nice - that gives me specific areas to target, and lets me re-test the AI in each situation from the replay until I'm happy with its behavior. Really, I can't overstate how helpful this is, and I really appreciate it.

I've made a lot of AI adjustments to try and improve the issues you mentioned. Some I think I've fixed completely, but others I have less control over (or disagree with). Of course, every adjustment I make to the AI in one situation will affect it in others, so it's possible I've introduced more issues than I fixed - hopefully not.

The new add-on version will include all the fixes I was able to make. Here's my analysis and response on each of your points:

turn 4: the second recruit group of enemies seems too hesitant, they could probably reach me and do major damage if they just went forward, they seem to be prepping to defend...

Should be better now (although maybe not 100% fixed) - thanks. The issue was very similar to the one with the group in the northwest on turn 8.

Turn 6: They retreat! I have a lot of vulnerable units out here, they could probably do a fair amount of damage, eat through my scout and then the marksmen and/or mages, they have the manpower to do it, I think. Yes, it's day, but I was terrified of the counter they were going to bring, and nothing! The crossbowman going into the oasis in front of me is pretty funny.

In this case, I think I agree with the AI's decision to retreat. Even if the orcs dogpiled onto your Scout, they only have around a 50-50 chance of killing it, and even less chance of killing either of your Marksman. And then basically all the orcs would die to your counterattack - that's a pretty bad trade for an 18 gold recruit. By retreating, enough orcs survive that they can still pose a meaningful threat once the sun sets.

The crossbowman in the oasis is pretty sad, though, I have to agree . That particular AI mistake should now be fixed. When I retest this the Orcish Warrior on 14,14 still moves to the oasis and attacks, but that's because he's slowed and has no chance of escaping no matter what he does.

Turn 7: I felt for sure that scout was dead meat, trying to engage anyone killing the scout would have put my heavy hitters in the open, they probably would have won that engagement. And now they have left the SE encampments undefended, they could have at least put some units there. And now they're kind of out of position for a night-time push

On this one I agree with you. If the ToD was Morning, I think retreating would've still been best. But since it was Afternoon, the orcs could safely kill your Scout and you wouldn't be able to counterattack until Dusk. I've updated the AI to consider both the current ToD and the next ToD (partially), and it now correctly kills your Scout here.

Turn 8: A couple units venture forth while others retreat, I'm not sure what is going on with that group in the northwest? If they committed with that whole group I'd be in at least some trouble.
Turn 9: Now the NW enemy group sallies forth, too late, I can just fall back and they've lost the night-time initiative because they fell back excessively earlier.

This is very weird, and definitely not correct behavior. So far all the other issues you found I could at least guess as to the cause, but this one is wholly mysterious.

Upon further investigation, I see the issue. The western units on turn 8 were too scared to attack (which is reasonable; their eastern allies were far away), but instead of waiting for more allies they decided to flee instead, which left them totally out of position. This should now be fixed (at least in this particular case).

Turn 10: Just kidding! They again, don't seem to know if they want to attack or not, they split forces for no obvious reason.

I can improve this one a little bit, but I can't fix it completely - even when I disable my micro_ai completely, the default AI still splits its forces here (although in a different way than before). I still appreciate you pointing it out, though, and I can definitely improve it a little.

11: They run away, but again, leave their perimeter encampment undefended.

I actually agree with the AI's decision here, or at least think there's room to argue. Any units they left in that perimeter encampment would've been 100% dead, but by giving it up that Warrior managed to survive and help in the final battle near his leader.

12. Again they fall back, but the 2 western Warriors are badly out of position to attack me when needed.
13. They realize, too late, that I can attack their leader (or is it just that night is coming?), but by this time, I have too many units with accuracy abilities with a target on their leader, and those two warriors can't do anything.

Yeah, I think those 2 western Warriors falling back was good, but they went way too far west and not near enough to their leader. After some adjustments they now move to 3,7 and 3,8 instead of 1,8 and 1,9 - I think that's a more reasonable balance between avoiding your units and supporting their leader.

That scout should be uber-dead with that Warrior just standing there...
The eastern warrior in his little castle did absolutely nothing and dies to an archer's counterattack.

Ah, it looks like that Warrior is on a Zone Guardian micro_ai; not controlled by either the default AI nor my new micro_ai. I forgot I had that in there. I'll remove that and put him back on the normal AI / micro_ai, so he can participate in attacks and retreats like normal.

Also not sure why the Warlord moved NE when he attacked when that just opened him to more of my fighters getting him, he could have just stayed put and probably lived at least 1 more turn.

That's the default AI, unfortunately. I agree it's really stupid, but there's not much I can do about it unless I reduce leader moves to 0 (which has its own issues) or rewrite the entire default leader AI logic (which would be a pretty serious undertaking). I don't know why the default AI does that, but it really likes to move its leader for no reason.

Although... I could have the micro_ai limit its leader's move to 0 if he's on his keep and any enemy unit is adjacent? That would make him a little less deadly whenever any enemy is adjacent (he couldn't run off 3 hexes to kill a weak elf, for example), but it would solve this particular issue.

[Dalas120]

v1.0.4:
Erlornas's auto-recalled companion is now usually a Scout-line elf

v1.0.5:
Changed campaign preprocessor ID to no longer interfere with other AOI add-ons. THIS WILL BREAK EXISTING SAVES!
S02: gotos now adjust once the map changes.

v1.0.6:
AI: chaotic factions will now be more aggressive just before daytime ends, and vice-versa for lawful factions.
AI: the AI is now less impressed by high-level units and/or badly injured units (both allies and enemies).
AI: when many slow allies are approaching but haven't yet reached the battle, fast scouts are now more likely to wait and less likely to retreat.
AI: units should now be better at avoiding very dangerous healing locations, even if there're no other reachable options.
AI: units are now less likely to retreat towards / huddle up around their leader when their leader isn't seriously threatened.
AI: fewer units will now stay behind when ignoring a retreat to try and kill a low-hp enemy.
AI: leaders will no longer move off their keep to attack an enemy they could've reached without moving.

[Dalas120]

v1.0.7:
AI: THIS WILL BREAK EXISTING SAVES! Restructured the way the AI is loaded for better compatibility with existing micro_ais. THIS WILL BREAK EXISTING SAVES!
AI: now supports guard units. Guards loiter at a stationed hexes until enemies approach, then switch to using the same attack/retreat logic as other micro_ai units. Guards return to their station and/or heal when no enemies are near.
AI: when ignoring a retreat to try and kill a low-hp enemy, the AI will now consider combinations of multiple units attacking in concert.
AI: when ignoring a retreat to try and kill a low-hp enemy, units who weren't able to retreat anyway are now considered cost-free attackers.
AI: when ignoring a retreat to try and kill a low-hp enemy, the AI will be much more aggressive if killing the enemy would remove the need to retreat.
AI: allied sides will now respect each-others' retreat goals.
AI: advancing units are now less likely to advance past allies who've already retreated.
AI: frustrated leaders will now return to normal behavior if the situation drastically improves.
AI: when the default AI moves a badly injured unit next to an enemy but doesn't attack, the micro_ai will now force a suicidal attack to make it harder for the player to distribute XP.
AI: improved performance when many allies are present on the battlefield.

[ZIM]

Hello, I want to share my feedback for this campaign.

I noticed this campaign is rewritten to explain how the orcs came to exist in the North, while adding more lore about silver magic. I feel this is a great attempt at it, but I still have some questions.

Spoiler:

Like, wouldn't the gate be heavily guarded considering its significance? Also, Rualsha's army feels a bit lackluster, compared to similar orcish sovereign we have faced in another campaign like Rakshas. And he's supposedly managed to easily defeat Lich-Lord Selmar? I feel like a Lich-Lord who have the capability to open a gate to the orc's homeland would be better prepared should the orcs they command would turn on them. My proposed solution for this is that we should scrap Rualsha altogether and just made Selmar the final enemy we should vanquish. Then we could have Selmar dialogue's about how Jevyan is a failure for failing to retrieve the Ruby and they would be the one who gets it. And we could just made the orcish leaders from before simply say that they're probing elvish land for more land to conquer, without any particular sovereign telling them to do so.

Also, I know this is supposed to be small campaign, but I think we can try to have more variations. Like in the Valley of the Trolls, I don't really get the feel of being flanked by trolls, since they're so slow. I think it would be better to make this scenario into a "Move Erlonas to the objective" and making the scenario a bit longer, then the troll leaders could either placed strategically along the route or we could forgo them and instead make the trolls spawn sporadically. And in the Detour in the Swamp, instead of straight to defeating the enemy leader, I think we can instead have map covered in shroud and have the player to look for the source of the undeads, maybe make it a monolith instead of an enemy leader and have Linaera move to it to win the scenario.

About the AI, I feel it's smart enough, perhaps even too smart, because during the day, even if we managed to close in to the leaders, they would simply move away. I think it'd be more logical for the enemy to defend their leader regardless of time advantage.

[redbeard2]

Just finished, I like the assassination mechanism, I was a bit afraid the leader would have some lich-like magic attack, glad he didn’t. The portal animations were way cooler than mine.

Maybe have an epilogue in which Erlornas advocates for the full militarization of Wesmere Aethenwood and Lintanir, with the Kalian, over-confident in their success, over-ruling him. Maybe they could cite recovery efforts from the drake invasion to justify their inaction.

[Dalas120]

Like, wouldn't the gate be heavily guarded considering its significance? Also, Rualsha's army feels a bit lackluster, compared to similar orcish sovereign we have faced in another campaign like Rakshas.

It is heavily guarded! Erlornas had to go through the swamp to bypass most of Rualsha's orcs.

...but I get what you mean. Which difficulty did you play on? I'd dramatically increased Rualsha's army size compared to the original AOI, but if you played on Hard/Deadly and still found the scenario fairly easy I'm happy to turn the difficulty up further.

And he's supposedly managed to easily defeat Lich-Lord Selmar? I feel like a Lich-Lord who have the capability to open a gate to the orc's homeland would be better prepared should the orcs they command would turn on them. My proposed solution for this is that we should scrap Rualsha altogether and just made Selmar the final enemy we should vanquish. Then we could have Selmar dialogue's about how Jevyan is a failure for failing to retrieve the Ruby and they would be the one who gets it. And we could just made the orcish leaders from before simply say that they're probing elvish land for more land to conquer, without any particular sovereign telling them to do so.

Featuring Selmar directly a good idea, and is something I went back and forth on during development. My concern is: my biggest motivation for restoring AOI is for nostalgia - as a recently-removed mainline campaign, there've been a lot of people who're sad it's gone and want it back.

Because of that, I've tried very hard to keep the campaign similar to the original, adding and polishing while still keeping all the original characters and battles. I worry that adding Selmar to the final scenario would be too major a deviation from the original, defeating the purpose of re-adding the campaign.

I also think that showing orcs breaking free from their undead masters is a unique addition to the lore, and bridges the gap between TRoW (where orcs serve the Lich-Lords) and LoW (where orcs decide for themselves).

That said, I hadn't meant to imply that Rualsha *easily* defeated Selmar - it certainly might've been a huge epic battle. We see at least one little piece of that battle in S2, with the orcish infighting. Maybe it would help if I added more mentions or allusions to a major battle in the past? Orcish and undead corpses surrounding the portal? That kind of thing.

Also, I know this is supposed to be small campaign, but I think we can try to have more variations. Like in the Valley of the Trolls, I don't really get the feel of being flanked by trolls, since they're so slow. I think it would be better to make this scenario into a "Move Erlonas to the objective" and making the scenario a bit longer, then the troll leaders could either placed strategically along the route or we could forgo them and instead make the trolls spawn sporadically. And in the Detour in the Swamp, instead of straight to defeating the enemy leader, I think we can instead have map covered in shroud and have the player to look for the source of the undeads, maybe make it a monolith instead of an enemy leader and have Linaera move to it to win the scenario.

I'll think about these suggestions, thanks! I probably can't rework Detour in the Swamp so drastically (for the reasons mentioned earlier), but I can certainly try to adjust Valley of Trolls. Do you have a replay?

About the AI, I feel it's smart enough, perhaps even too smart, because during the day, even if we managed to close in to the leaders, they would simply move away. I think it'd be more logical for the enemy to defend their leader regardless of time advantage.

Hmm, that's not good - the AI should prefer retreating towards its leader, not away from its leader (except for weird cases, like your whole army being between the fleeing unit and their leader).

If you have a replay showing weird retreat behavior, can you send it? I'll take a look at why the AI is doing the wrong thing.

Just finished, I like the assassination mechanism, I was a bit afraid the leader would have some lich-like magic attack, glad he didn’t. The portal animations were way cooler than mine.

Maybe have an epilogue in which Erlornas advocates for the full militarization of Wesmere Aethenwood and Lintanir, with the Kalian, over-confident in their success, over-ruling him. Maybe they could cite recovery efforts from the drake invasion to justify their inaction.

Thanks for playing; I'm glad you enjoyed the last scenario! Credit to Inferno8 and Durzi for the animations.

Regarding the epilogue: your idea makes a lot of sense, but this is actually meant to segue into Mechanical's upcoming Legend of Wesmere revision, in which Erlornas is captured before he can return home (before LoW starts). So I need to end it before he returns to Wesmere, or else the two campaigns wouldn't match.

[redbeard2]

I hadn't gotten the impression that LoW was RIGHT after Orcish Incursion, the timeline is written kind of weird and disjointed regarding this time period, it officially says LoW starts in 20, whereas Incursion ends in 9, but in between it describes some of the events in LoW, so I'd always gotten the idea there was some separation between them

[Dalas120]

Yeah, I believe the way Mech has it set up is that Erlornas is held in prison between AOI and the start of LoW. Then during LoW you free him.

[lhybrideur]

That said, I hadn't meant to imply that Rualsha *easily* defeated Selmar - it certainly might've been a huge epic battle. We see at least one little piece of that battle in S2, with the orcish infighting. Maybe it would help if I added more mentions or allusions to a major battle in the past? Orcish and undead corpses surrounding the portal? That kind of thing.

That would be a nice touch indeed.
Maybe he could even have some physical characteristics because of the fight if he fought him himself, like missing an arm or smth.

[Dalas120]

That would be a nice touch indeed.
Maybe he could even have some physical characteristics because of the fight if he fought him himself, like missing an arm or smth.

Okay, will do.

Also, I know this is supposed to be small campaign, but I think we can try to have more variations. Like in the Valley of the Trolls, I don't really get the feel of being flanked by trolls, since they're so slow. I think it would be better to make this scenario into a "Move Erlonas to the objective" and making the scenario a bit longer, then the troll leaders could either placed strategically along the route or we could forgo them and instead make the trolls spawn sporadically. And in the Detour in the Swamp, instead of straight to defeating the enemy leader, I think we can instead have map covered in shroud and have the player to look for the source of the undeads, maybe make it a monolith instead of an enemy leader and have Linaera move to it to win the scenario.

Copying a response I wrote in discord regarding a similar suggestion:

I'm usually more than happy to include different objectives in campaigns, but since my main motivation for re-adding AOI is filling players' nostalgia, I'm worried about straying too far from the original. I don't want people to be excited their favorite campaign got re-added, only to be disappointed it now plays very differently than it used to.

So I'd tried to keep all the literal objective text unchanged, but to add gameplay variety by changing up the maps and dynamics (S2 is a FFA, S3 is "move your army to the northwest corner of the map", S6 is "survive until turns run out", etc). I don't have any mainline criteria condemning "defeat enemy leader" specifically, just that each campaign meet a certain level of gameplay variety / strategic complexity.

I've added a lot of new plot compared to the original as well, of course, but similarly I've tried to keep the core story and characters intact (that's a part of why you fight Rualsha in the final scenario instead of Selmar, for example).

That said, perhaps I'm being a bit too strict about pandering to nostalgia. While I'm not sure about the drastic redesigns you suggest for S5, I could certainly change S3 to "move Erlornas to the west edge of the map" or such, and rebalance the scenario accordingly.