Testing Siege AI in Campaign...

[revan.be]

If you have a relatively new nvidia card , you can use shadowplay to record.
Oh by the way, please be masochistic enough to not use any mods

[shireknight]

Being as you have a general thread for patch 14 sieges I may as well stick my Athens replay in here as well Humble Warrior if that's ok with you mate

https://www.youtube.com/watch?v=6CnDY_1_lx4 Why does my video start playing from halfway through rather than at the beginning?

Sparta and Athens teamed up to wipe me out shortly after this battle so in total my campaign lasted under 30 turns which is the fastest I can ever remember losing a campaign in pretty much any TW game.


I've not been able to play anymore since Friday due to work but I've now got two days off starting today and I'm really looking forward to getting my teeth back into the new siege AI, in fact I've been thinking about nothing else since Saturday -well done CA you have finally made a patch that's sparked my interest back in the game and put you back in my good books

[Humble Warrior]

Being as you have a general thread for patch 14 sieges I may as well stick my Athens replay in here as well Humble Warrior if that's ok with you mate

https://www.youtube.com/watch?v=6CnDY_1_lx4 Why does my video start playing from halfway through rather than at the beginning?

Sparta and Athens teamed up to wipe me out shortly after this battle so in total my campaign lasted under 30 turns which is the fastest I can ever remember losing a campaign in pretty much any TW game.


I've not been able to play anymore since Friday due to work but I've now got two days off starting today and I'm really looking forward to getting my teeth back into the new siege AI, in fact I've been thinking about nothing else since Saturday -well done CA you have finally made a patch that's sparked my interest back in the game and put you back in my good books

Interesting video and your participation is welcome.

While the rest of the attack was good, I notice that the first thing the attacking AI did was burn the gate with torches @ 2:25. This should not happen now. That is bad, sorry, but it is. The other AI has ladders, and the reinforcements should have joined them or brought their own ladders or wait. For sieges to be as good as RTW or MTW2 (and authentic to history) the torching of gates really needs to be erased completely. The AI needs to find another way or join its friends.

Strange because in all my sieges so far the AI has never burned down my gates yet, although it would try sometimes, then give up when it noticed the other side had the ladders up. I suspect the reinforcement in that vid would have given up if the AI had made it to the ladders before the gates had been burned down.

I believe the siege AI is good enough to do away with any torch burning of gates with just a little more tweaking. I pray CA do this.

[Humble Warrior]

Ok, had a naval invasion from Thracia on Roman Brundisium. He had a fleet of transports that attacked my Navy ships. Of course his transport ships attacked my military naval ships.
Anyway, he had too many, so even when I ran around him and made a direct ram hit with my two ships he just tractor beamed them and boarded.

I wondered how it would tackle the siege? A couple of his transports landed at the harbour that led into the city, but my men easily saw them off. the rest landed on the coast outside and made for the gate. Yep, Firestarter time! However, I had quite a few skirmishers and they were pelted all the way. By the time they had smoked the gate they all routed- they would`ve fared better had they bought just one ram or ladder. Torching needs to be deleted out of naval invasion battles and alternative options included as has been mentioned (using a ballista ship to bust walls or give a free ram or ladder on beaching).

Was attacked further north at Medhlan again 1396 barbs versus 730 romans.

This time my Romans had a catapult which smoked one of the 4 ladders, but not much else. The ladders made it to the twisty walls of the barb town quite well. However, while they did well at first, even getting over the walls without me seeing since they plonked a ladder quite far off, they seemed to get somewhat confused and my velites did a lot of damage to them. Perhaps it was because they were inexperienced why they ran so easily.

It was hard to get my troops to target barbs on these walls, the collision detection was very iffy, so I had to just wait until finally it `caught`. the General on foot, seemed confused on what to do next, but that could been due to one reinforcement cavalry I had waiting outside just for the right time to strike, which I did and they routed.

Not bad, but still needs tweaking on its movement on those walls, especially the more Northern Gaulish types. No torches were used here, thankfully.

Well it`s beta. Hard to know exactly in what direction CA are going until they produce patch 14. Are they totally doing away with torches? I hope so since it would be a shame to get a sieg AI that`s nearly getting as good as RTW\MTW2 just to have it torch the gates at the last moment, ruining potentially epic sieges.

[Slaists]

One trick with naval invasions, if you have a fleet and do not want to deal with the mess of fighting a battle at the sea, is to deny them landing real estate (such as the harbor inside Brundissium with 3 landing spots). Instead of arranging your fleet to meet the invaders at the sea, arrange it to beach anywhere where you do not want the AI's ships to beach. That way you can channel the ones who manage to find space exactly where you want them. You can deal with those from the walls as they'll have to burn the gates since a transport fleet has no artillery ships. The rest of their stack will just mill around the sea harmlessly...

[IlluminatiRex]

I don't thin Patch 14 is the last major patch, as there are still a lot of things "In Process B".

[Butan]

I don't thin Patch 14 is the last major patch, as there are still a lot of things "In Process B".

In Process B?

Can you develop and also (just a guess) give us a lit of whats in here?

[IlluminatiRex]

In Process B?

Can you develop and also (just a guess) give us a lit of whats in here?

"In Process B" means a report has been forwarded to the Development Team. Here's a few of the things recently sent to them: A bug I reported about units mouths not opening during speech, the unit fade thing, having heavy units push lighter ones back, illyrian marines not firing javelins, graphical artifacts on the campaign map for some players, incorrect morale effects, issues with multi-territory faction CAI, recruitment bonuses not being applied to Roman cities, along with many more bugs and issues are currently listed as "In Process B" on the official forums.

[Humble Warrior]

CA could have made it more clear that the beta was just there to see if we liked it or not. They made no impression that whatever we said was not going to be in patch 14 complete.
Also some clarity and honesty in what they're actuallyfixing in patch 15 considering our findings would be appreciated.
CLARITY, TRANSPARENCY, HONESTY and not ABUSING IT.

Seriously, do you guys in CA want a good relationship with your fans or not

Sent in a hurry on mobile. So forgive errors!

[IlluminatiRex]

CA could have made it more clear that the beta was just there to see if we liked it or not. They made no impression that whatever we said was not going to be in patch 14 complete.
Also some clarity and honesty in what they're actuallyfixing in patch 15 considering our findings would be appreciated.
CLARITY, TRANSPARENCY, HONESTY and not ABUSING IT.

Seriously, do you guys in CA want a good relationship with your fans or not

Sent in a hurry on mobile. So forgive errors!

Mentioned this in another thread, but I'm quite sure that the Beta Patches exist so that way modders can get their mods up to date and working. So that way there isn't widespread issues for everyone once the mod is out.

[Sint]

The siege:
Playing on Hard as Rome. Unlimited time. No flags or anything like that. I`m making it easy for the AI so as to induce sieges against my walls.

I caused a rebellion and they attacked one of my Italian Roman cities:

I had 2880 men comprising of Rorari and levies and skirmishers.
The rebels had 1920 comprising of Italian swordsmen, spears and 6 cavalry including the General.

I saw one Ladder infront of one of my gates held by one unit of Italian Swordsmen. The rebel army was quite far back, facing me.

I started the battle, placing some skirmishers and Levys on the walls while my Rorari stayed on the ground facing the gate. The ladder moved towards the wall.

This time I was going to make life difficult for the AI to see how it would react. Seeing that the main army had stayed well back, I sent out a 1 unit of Rorari to attack the ladder as they neared the wall. The Rorari slew the pushers leaving the ladder behind. A second unit of Italian swordsmen left the main rebel army and headed to the ladder, now guarded by my men. This would be an easy win if the AI just sends one unit each time.

As the one unit neared the ladders, 3 more units left the rebel army and marched forward. This was a good move, as I knew I could not hold that lot off with my one unit of Rorari and was forced to retreat. The extra units came quite close then and as then moved back. I was pleased with that.

The rebels pushed the ladders to the walls and boarded- then the rest of their army joined. I sent some Rorari up to defend the walls along with my levy and skirmishers.

This was when things actually started getting enjoyable. The rebels piled up that ladder and joined their friends on the battlements. Some actually made it off the wall before I could get all the Rorari up so I ended up having to fight an increasing number of Italian spearmen and swordmen on the ground just infront of the gate. It was becoming an actual battle to stop him- and he didn`t torch the gates at all.

After a long, tough fight, boosting moral with my general, the numbers of rebels finally broke and fled, leaving the enemy general and his five cavalry behind. This was the crucial bit... they couldn`t climb the ladders, what were they going to do. I stared in anticipation. Was I going to see bs torching?


They left! And the battle ended with a win for me! They never devolved to torches. He just took his cavalry and left!

Well that's better than before but.... only one Ladder ,no rams and all attack at the same ladder ?I mean just putting the damn roari on the wall and supporting them with javelins before the ladder touches the wall would have made it a masacre.

[M2TWRocks]

I almost never manually fight siege battles, so my meager experiences may be just on the fringes of probability. BUT, I fought two siege battles on my current Baktrian campaign and there were no torches used. The enemy attempted to scale walls with ladders both times. I haven't played this game since last year. I remember attempting siege battles then and it was a giant mess.

[Symphony]

I guess I didn't pay enough attention to other posts in the thread, or I didn't entirely believe it myself.

Roman campaign, moving up from Italia to hit Patavium. I'd been working on poisoning the defending armies for a couple of turns, so I knew that they had a fair amount of Light Horse (the ubiquitous early-game barbarian melee cav), but I wasn't too fussed; the AI would just reinforce to the center of town, right? So, I decided to bring ballistae, thinking they'd rack up their normal ~150 kills each, and I'd roll through it like normal.

Well, map spawns, and there's a slight rise up into the town, blocking LoS on the defenders. I could have deployed elsewhere, and not had the rise to deal with, but again, it's a settlement fight, right? Along those lines, I deploy my arty in front like a madman, thinking I'd have plenty of time to move them up for LoS and start the killing. I deployed skirmishers behind them, thinking to move them up ahead of the slower arty to provide a screen that's never really necessary anyway; all my infantry and cav was behind.

Before I can crest the rise, all those AI light cav were charging down that rise, and my infantry and cav were too far back to reinforce; my skirmishers and arty got shredded before I could get my (superior) cav into the fight and chase them off. Behind the AI's cav, I saw spears and skirmishers moving out of the town to support....with the faster skirms actually staying behind the spears, so I had to pull my cav back, rather than just continue pushing and get some easy skirmisher kills.

Early Roman infantry owns early barbarian infantry, so once I recovered, I ended up winning. But the lesson cost me all my arty and most of my skirmishers....units that would previously have simply hung back and racked up kills. It left me wishing three things:

I wish I'd deployed from a more advantageous angle
I wish I'd left the arty at home, and brought more cav (...and against an unwalled settlement, that's what I should have to think)
I wish I'd actually honored the AI, and deployed in a more sane fashion, without exposing my arty and skirms like it didn't matter

Basically, the AI is making me honor it in an offensive siege....a rude, but extremely welcome awakening. I know everybody has rose-colored glasses about previous TW games, but that might have been the most tactically correct AI siege defense I've ever seen in this series....there were WAY too many times in vanilla RTW, M2TW, and (especially) ETW where I'd use arty to kill towers, knock a few breaches in the wall, and walk my archers right up to those breaches to shred the cav sitting right on the other side at my leisure.

Good stuff.

[Humble Warrior]

I guess I didn't pay enough attention to other posts in the thread, or I didn't entirely believe it myself.

Roman campaign, moving up from Italia to hit Patavium. I'd been working on poisoning the defending armies for a couple of turns, so I knew that they had a fair amount of Light Horse (the ubiquitous early-game barbarian melee cav), but I wasn't too fussed; the AI would just reinforce to the center of town, right? So, I decided to bring ballistae, thinking they'd rack up their normal ~150 kills each, and I'd roll through it like normal.

Well, map spawns, and there's a slight rise up into the town, blocking LoS on the defenders. I could have deployed elsewhere, and not had the rise to deal with, but again, it's a settlement fight, right? Along those lines, I deploy my arty in front like a madman, thinking I'd have plenty of time to move them up for LoS and start the killing. I deployed skirmishers behind them, thinking to move them up ahead of the slower arty to provide a screen that's never really necessary anyway; all my infantry and cav was behind.

Before I can crest the rise, all those AI light cav were charging down that rise, and my infantry and cav were too far back to reinforce; my skirmishers and arty got shredded before I could get my (superior) cav into the fight and chase them off. Behind the AI's cav, I saw spears and skirmishers moving out of the town to support....with the faster skirms actually staying behind the spears, so I had to pull my cav back, rather than just continue pushing and get some easy skirmisher kills.

Early Roman infantry owns early barbarian infantry, so once I recovered, I ended up winning. But the lesson cost me all my arty and most of my skirmishers....units that would previously have simply hung back and racked up kills. It left me wishing three things:

I wish I'd deployed from a more advantageous angle
I wish I'd left the arty at home, and brought more cav (...and against an unwalled settlement, that's what I should have to think)
I wish I'd actually honored the AI, and deployed in a more sane fashion, without exposing my arty and skirms like it didn't matter

Basically, the AI is making me honor it in an offensive siege....a rude, but extremely welcome awakening. I know everybody has rose-colored glasses about previous TW games, but that might have been the most tactically correct AI siege defense I've ever seen in this series.

Good stuff.

Nice AAR.
Yep! Despite my jadedness, this AI is above and beyond the dreadful thing pre-patch 14. In fact, I would say it is as good mtw2's AI. Better? I can't say until I've completed a full campaign, something I could not do before.
I didn't believe it till I saw it. But this what we should have had at release. I'm afraid I won't buy CA games until 10 months later if this is how long we must wait.

[Symphony]

Nice AAR.
Yep! Despite my jadedness, this AI is above and beyond the dreadful thing pre-patch 14. In fact, I would say it is as good mtw2's AI. Better? I can't say until I've completed a full campaign, something I could not do before.
I didn't believe it till I saw it. But this what we should have had at release. I'm afraid I won't buy CA games until 10 months later if this is how long we must wait.

Regarding the bolded, I edited the above with:

"I know everybody has rose-colored glasses about previous TW games, but that might have been the most tactically correct AI siege defense I've ever seen in this series....there were WAY too many times in vanilla RTW, M2TW, and (especially) ETW where I'd use arty to kill towers, knock a few breaches in the wall, and walk my archers right up to those breaches to shred the cav sitting right on the other side at my leisure."

In my experience, previous TW games were just as easily exploitable (once you knew what to exploit) as this one, at least on settlement defense. Whether this one should have been better at release or not is a matter of opinion, obviously, and I'm not going to tell you yours is wrong, but I think it's telling that I've been trained to expect poor settlement defense AI, both from TW games and from other strategy games. Maybe responsive AI is just bloody hard to pull off?

Again, opinion is opinion. I'd agree wholeheartedly that the AI we had at release was sub-par. I'd simply suggest that you check your memory to make sure you're comparing vanilla to vanilla, and offer that I personally struggle to think of a similar situation in any game where I was encouraged to honor a settlement defense AI to the extent we're discussing; certainly I've never had to in a TW game.

It's fair to say that it's sad that my expectations are so low, but it wasn't just this game that made them so low.

[Humble Warrior]

Regarding the bolded, I edited the above with:

"I know everybody has rose-colored glasses about previous TW games, but that might have been the most tactically correct AI siege defense I've ever seen in this series....there were WAY too many times in vanilla RTW, M2TW, and (especially) ETW where I'd use arty to kill towers, knock a few breaches in the wall, and walk my archers right up to those breaches to shred the cav sitting right on the other side at my leisure."

In my experience, previous TW games were just as easily exploitable (once you knew what to exploit) as this one, at least on settlement defense. Whether this one should have been better at release or not is a matter of opinion, obviously, and I'm not going to tell you yours is wrong, but I think it's telling that I've been trained to expect poor settlement defense AI, both from TW games and from other strategy games. Maybe responsive AI is just bloody hard to pull off?

Again, opinion is opinion. I'd agree wholeheartedly that the AI we had at release was sub-par. I'd simply suggest that you check your memory to make sure you're comparing vanilla to vanilla, and offer that I personally struggle to think of a similar situation in any game where I was encouraged to honor a settlement defense AI to the extent we're discussing; certainly I've never had to in a TW game.

It's fair to say that it's sad that my expectations are so low, but it wasn't just this game that made them so low.

Indeed. Everyone has their personal experiences and views shaped by many factors. But I've only recently been playing vanilla MTW2 before deciding to try the beta. If you look above I mentioned how in one siege it did exactly what MtW2 SIEGE ai does on attack. And I have played Mtw2 a lot. I base Rtw2's performance along MTW2. If the AI can at least manage that, I'm happy. I believe you're using long ago memory on MTW2 rather than considering that it is 10 months of bad RTW2 that really got you into bad habits.
RTW2 was sold on release in worse state of every total war even Empire and their false advertising made it worst. 10 months to wait for an AI that is competent to others, ie a playable challenge, is not acceptable.

But I respect your point of view.

[Post Hoc]

Attack a town from the sea, land one unit on the edge of the battle field and have the rest of your ships waiting in the water. Every single enemy unit will chase that one unit that landed. This happens 100% of the time.

[the lightning legion]

FYI sometimes with rebel armies or displaced factions they will launch a suicide assault in which they will use cav to try and burn gates

[Mackles]

They cant remove the torches because a naval fleet assaulting a city has no way of breaking the gates. I think I'm right that naval fleets cant land and besiege.

Question time: Should navies, dedicated or transport, really be able to assault walled settlements? They can already fight at sea, blockade ports, raid trading routes and attack minor coastal towns, so from a gameplay design perspective they are already bloody useful. Do they really need to be able to instigate a siege battle from the water, where their only means of attack (unless there is an artillery ship present*) is the torches - which is usually a death sentence. I think it would be fine if a navy could still support a ground-based attack with reinforcements, but they should need a dedicated ground force (rather than transported ground forces) to actually start a siege battle.

*I wouldn't even bother making this an exceptional circumstance anyway, as naval artillery should neither be heavy nor accurate enough to cause a breach.

[Symphony]

Question time: Should navies, dedicated or transport, really be able to assault walled settlements? They can already fight at sea, blockade ports, raid trading routes and attack minor coastal towns, so from a gameplay design perspective they are already bloody useful. Do they really need to be able to instigate a siege battle from the water, where their only means of attack (unless there is an artillery ship present*) is the torches - which is usually a death sentence. I think it would be fine if a navy could still support a ground-based attack with reinforcements, but they should need a dedicated ground force (rather than transported ground forces) to actually start a siege battle.

*I wouldn't even bother making this an exceptional circumstance anyway, as naval artillery should neither be heavy nor accurate enough to cause a breach.

Logically, there's no reason they shouldn't be able to. At the end of the day, the primary differences between a navy and an army are the boats and training.

I wouldn't worry too much about the gameplay design perspective. Right now, armies can already fight on land, blockade cities, raid trading routes and the countryside....AND fight at sea relatively well and easily from no-cost, no-time transports. They have far more utility and flexibility than navies do already, and navies cost more.

I don't see a solid logical or design rationale for why you couldn't just allow navies light siege ladders. Armies already build ladders "instantly" upon laying siege; I assume this simulates them carrying the parts with them in the baggage train, and assembling them in the field. You couldn't fit the parts for at least a ladder or two in a trireme (at least enough to supplement from local foraging)?

Make it a two-turn process if it makes you more comfortable. On turn one, the navy blockades. On turn two, they could assault from land like any army; the intervening turn could simulate the time required to find a place to land in good order and build the ladders.

Sure, naval personnel would be trained in and used to a different sort of combat than their army brethren, but I feel like the reduced unit size of naval units already limits their combat effectiveness against an army to represent this.