Salamandre said:it would be nice if we could create a new "tactics" manual inspired from those exploits. For example, what is the ratio HP flying units to keep defender in town, what is the ration shooters HP needed to remain alive to keep flying units inside town and such things. Can you write down when you find such conclusions then post them somewhere? It would be great, thanks.
I guess this one is on you, better said on your amazing Trainer. Do you know Dimis? Maybe he could uncover the formula if enouch data is provided.
RoseCavalier said:I guess what I'm saying is that there's no clear-cut formula to figure these things out other than trial and error or figuring out where AI draws the line at which point we could use H3calc to get answer.
Nah, there is a "clear-cut formula", it's just how computers work. We just need to find it like someone already did with the one for Diplomacy and joinners, for instance.
yes, trainer is the perfect tool for testing such situations, as it can remove spells from hero, set skills, stats + add/remove creatures/artefacts. But yet someone needs to know in which direction to search.
zmudziak22 said:From Richard Chen learn that AI won't move from castle if you have weak shooters. Look at 2 his videos. Trick 5 and trick 6.
yes this is old news by now. The question is to find the exact cap, so during game play you don't waste your time to repeat the battle with different army values.
Then is not entirely true, because I attacked once AI with 500 titans and he has 2 groups of shooters which could annihilate my titans in one shot, therefore his blacks didn't get out. 500 is not exactly weak, so is a matter of ratio.
I started once creating such a tactics file (I think you already saw it, you can get it from my website) but quickly figured out that those are things that, atleast from my perspective, cannot be summarized in an easy way,and finally gave up, writing up only a tiny sliver of what there is to know. General rules of thumb that make the AI go out SEEM to include (but would need great testing to be 100% certain, which I have never done)
*Enemy placement inside castle is not safe
*Potential damage too high on player side
*Stack inside castle cannot be removed in time
Also, there are factors like spells, which are generally much in the players favor in terms of castle siege (as seen on some battles, AI tends to prefer buffs over damage spells and teleport) which seem easy to understand, but fall under some amount of RNG and are influenced by many factors.
One would have to do a ridiculous amount of testing to figure this out empirically, if even possible to.
zmudziak22 said:[...]AI won't move from castle if you have weaker shooters.
The difference is important here, so long as AI thinks its shooters are stronger than yours, it sees absolutely no reason to exit the garrison. I've used this tactic quite a bit, see 243 where I'm using 140 Titans to defeat a huge army which includes 4000 Cyclops. In that fight, AI did not bother to exit for as long as it had over ~500 Cyclops.
Yes of course AI uses clear formula, although sometimes there are RNG elements which may surprise. I just meant that there probably isn't an easy formula to apply which would allow us to decide where 46 or 47 Ghost Dragons are too much.
I'm might be scrapping the 311&312 update and do them all over. I may have found a way to speed up death of 1 key enemy by 1 whole day which would speed up Orb of Vulnerability by 1 day as well.
The idea is defeating Resus with a hero other than LLLB, who was short on movement to defeat the key hero. Previously I couldn't find enough SP to defeat Resus with Friend but I believe I've found a way now. One consequence would be that 'Sphere of Permanence' is pushed back as well as the 150 Archangels again!
EDIT: Doesn't work... Forgot I don't have Azure Dragon before 314 so extra movement on LLLB is useless.
Two more days with a lot of movement but not many intense fights. The hardest thing at this point is planning several days ahead, anticipating AI's movement and so on.
Brewster's day summarized in 1 screenshot. You need level 42 to have enough movement, she is 46 in my game at this point so I get an extra tile of movement.
Back in 'Sharp Corner' where Sepello and Vertigo are standing by, I hire one slow Stronghold creature which can be used by Sepello to save movement on the nearby 'rough' terrain and later by Vertigo to board a boat.
I trade some artifacts to Sepello, which were previously on LLLB.
Both heroes walk towards the Seer Hut (already completed) because there is a hidden passage nearby which can only be accessed using 'Waterwalk' - but first you have to remove the floatsam in the water with a boat.
Fastest way for Vertigo to leave is to retreat.
I re-hire him, naturally.
The previous operation allows Sepello through.
This is another network of monoliths, you can check in video to see other destination. For now I'm heading to the bottom of the underground.
While there is an easily accessible green monolith, it's not accessible because of a disguised border gate.
There are traps here which do not vanish after visiting...
Of course, I've planned for this. Take the piles of resources before Sepello runs out of movement.
After visiting stat boosters, Friend may go to Acre5 for its nearby monolith.
Once more, head to this area (feels like I've gone here way too many times) but for the stat boosters.
LLLB passes through the Oasis then to 'Gummy Gums' up until the quest guard requiring 1 Azure Dragon.
Re-hired Vertigo follows, he's part of the chain for the next day.
During the night, Red heroes activate. Because I've killed 2 of them, they are now able to open the prisons which contain heroes with generally high primary stats and many troops. There are 4 Swan Ponds+Peasants which add a 4-day buffer (247->313 release). This will change the tempo of the game a lot because now I am becoming invaded by the AI! As per a previous comment, I might try to ignore them as much as possible while I make my way to the 'Orb of Vulnerability'. Of course, if they place themselves within striking distance I will not hesitate to get rid of them. After all, I need to kill the heroes in the last 2 prisons to complete the map as they unlock quest guards later. [Yes, I used reveal map cheat to show this, but only for screenshot]
The released heroes have several options on where to go, it seems random but there appears to be some ways to influence where they go.
SLOB decides to come at me.
Hard fight, no? 3 MGenies vs. 5 Titans.. oh. Anti-Magic.
Try killing these guys! Resurrect my units while he's busy fighting Elementals.
This hero had 99 of all stats and would sometimes meet with the other released red hero and would take all army, making for a much tougher fight. I spent movement on Friend (not much else to do anyway) to make sure SLOB never had a chance to become strong. The other released hero is much weaker and can probably be killed in the open, I have not decided yet.
Brewster is recharged with 5280 more movement, time to capture 1 Azure Dragon! The 2 Seer Huts hold the key to the second quest guard.
In exchange for 2 'Orb of Silt', we get both vision artifacts.
There is the 'Helm of Heavenly Enlightenment' here which is a whooping +6 stats to all, unfortunately it's out of reach for now relative to my plans so I will come back later. Instead, Brewster refreshes at the Well then fights Azure Dragons.
Fight setup: as much +Attack and at least +3 Spd.
This fight is not 100% safe because I have no way to prevent 'Fear' - no undead, no elementals. Fortunately, the terrain is very friendly as it allows you to hide 3 stacks + ammo cart using 1 forcefield.
Cast 'Prayer'+'Haste' on Titans, this way I now have 3 creatures that are faster than Azure Dragons, basically odds of 3 'Fear' in a row is 0.1% but there are around 50 rounds in the battle. Probabilistically speaking, that corresponds to approximately 5% chance of 3 fears in a row during the length of the fight, which is very reasonable. If OTOH I only had 2 faster stacks, the odds would be ~40% which could prove quite frustrating.
Cast buffs and shoot away. Resurrect Rust Dragon before killing the last enemy!
Experience is good for 2 levels 47-48
Ignore everything but this Frozen Cliff, movement will be just enough as is.
Azure Dragons will try to damage shooters or block them, if you leave Titans next to Ammo Cart they will kill it, so place Titans last in order.
At long last, we get 1 Azure Dragon... there aren't that many things we don't have access to now.
Make your way to the east to another monolith network. This time we're heading to the top of the underground.
The quest guard here asks for our Rust Dragon, which explains why I even cared to go get it.
Make your way to 'Nidarcok', which is barely guarded.
It's nothing incredible, but capturing this city means more gold, more Behemoths but especially an easy way to go back and get the +6 Helm. Brewster gets here with ~300 movement left.
Back to the bottom of the underground, Sepello opens quest guard and claims all these resources. The top quest guard cannot be unlocked yet, few more days. The first reason to come here was for the Wizard's Well, the second was because I have some spare movement and it will make my second going through here a bit cheaper movement-wise.
There are 700 of each resources here, you might understand why I sold almost all I had on 311
Being done with the greed mission, Sepello grabs some Black Dragons.
And then the Behemoths from the new town.
Plus the Azure Dragon from Brewster, as well as some artifacts.
And go back to Sharp Corner where the bulk of the army was left.
FINALLY, we're ready to chain troops toward LLLB. Friend heads to Gummy Gums and pick up the slow troops left in the garrison the previous day.
Sepello follows and the chain begins. Spello > Friend > Vertigo > LLLB.
This chain brought the Azure Dragon from the other side of the map right up to the front of the main quest path within a single day!
Last portion of the chain...
Finally we can move forward. Sadly, I wasted quite a bit of movement in the last few days because of the quest guard requiring Azure Dragon here.
Remaining quest guards here are shown below.
LLLB captures the Frozen Cliffs but don't hire the Dragons yet...
...because of this garrison.
Speed tie [17+4] vs  but nothing that can't be permanently fixed. 'Slow' and later 'Prayer'.
It's a very boring fight like most wandering creatures now.
At least it's good experience, worth 1 level.
Vertigo follows LLLB with the remaining of the army, hiring the Azure Dragons while getting through.
For the next fight, it's very easy if you set up right.
The hero has 3 stacks of 9000 Medusa... and the Tactics skill!
If you remember a fight at the start of month 2 where I fought 3x500 Faerie Dragons with LLLB, enemy with Tactics will shift its troop formation before the fights starts, here combining the 3 stacks into 1 as Nemesis had done earlier.
Destroy the Ammo Cart and then just wait things out.
4 Shots drain by very fast, leaving us invulnerable.
For exhaustion phase, just Dispel the enemy's buffs.
Doesn't take long, and spell points are not an issue with the Wizard's Well anyway.
Resurrect Archdevils and other troops to kill 'Athmos' easily.
R.I.P. Black Dragons, your purpose has been accomplished.
While there is not much to say about experience gained this fight, @wow@ at those artifacts.
As much as the Armageddon's Blade is a nice prize, this is just godly.
We must share it quickly! Such a powerful spell we all know.
Split army and artifacts, LLLB will go forward with Archangels and Behemoths + Wizard's Well.
Vertigo brings 'Power of the Dragonfather' to Friend who will refresh at Uruk.
And then Friend brings it back to Brewster... going full circle in the trades.
Each hero goes his way, Friend to the trees and Brewster to the +6 helm next day.
Lastly, Sepello can visit some stat boosters as well. Not much free time left for her...
As we have so many details,do we know what part of the story is revealed by now?I remember in the begining there was some list of heroes and gods,including Metataxer.
"I heard the latest HD version disables playing Heroes. Please reconsider."-Salamandre
There isn't that much more story revealed at this point compared to the start. Most enemies are introduced by events shortly before we meet them, in most cases we know them from a previous time so there is short dialog between them. We're still running after Metataxer by defeating his guards and such.
The spellbook is from the Peasant at (138,76,1) in the Hotspot at the start. It is normally meant for 'Tan Horse', the first enemy hero I killed. In the sense that, if you don't kill him, he will get the spellbook and become quite powerful from a few events. I have no uses for it so it's just decoration!
Resus: yes he could still be beaten but it would be a lot more difficult.
1-hit and run to kill Faerie Dragons (tough because of Shackles, but feasible)
2- hit and run to drain SP
3- come in with Enchanters, 1 Archdevil [speed initiative] and 1 Vampire Lord [immune to fear]. Fight would need LLLB to use high mana pool and Expert Haste+Frenzy.
Round 1 (and every odd round) Cast Forcefield to protect Archdevils and Vampire Lord.
Round 2&4 Haste+Prayer, now Vampire Lords are faster than Azure Dragons so we're safe. Can get rid of Archdevil now.
Round 6 Resurrect Enchanters on Wait Phase of Vampires. Because they had not moved on round 1 where they died, they should be able to move right away and hide behind forcefield.
Then it's Frenzy-killing.
You could also try it wearing 'Power of the Dragonfather' but you'd probably run out of spell points because Defense on enemy is really high
Hi! Once more excellent work! I watched your last clip and the fight against Athmos seemed so easy! When I fought against him, I used A Friend and many shooters so he didn't combine his Medusas and he gave me a hard time! He cast sacrifice on any stack damaged or left without shots, haste when he wanted to reach me and even dispel when I filled the battlefield with landmines. The bow of the sharpshooter prevented any blocking and his +3 morale made things worse. He was so powerful that he could resurrect about 200 Medusas, then sacrifice them to resurrect 1000 ones and then sacrifice these to resurrect 5000 Medusas (the ones I killed first so they had full shots left)! Of course I didn't set up my army well and gave him the chance to make a fool out of myself but Athmos showed me that an AI hero with strong primary skills, the right artifacts and many spells can play really smart.
That gives you 11 targets before Black Dragon, assuming Athmos will cast 'Magic Mirror' 3 times before switching to Implosion, you should have enough time to resurrect powerstack so that 12 Medusa shots are wasted. For rest of fight, use 'Quicksand' and 'Sphere of Permanence' then resurrect powerstack to exhaust SP. I'd say use 'Protection from Earth' but odds are enemy will switch to 'Armageddon'.
Definitely expensive on spell points... using LLLB just works best in my game as he is going straight to 'Tongue' and makes for a casual Athmos
About 60% done testing through 317, pretty satisfied with results so far... expect an update or two over the weekend.
Thanks for your reply!
The problem was that Athmos, when he was left with no shots, moved to the centre of the battlefield and then cast expert haste and reached me (basic speed 5 + 4 from all speed artifacts + 4 from haste = 13). Even with quicksand -not land mine as I wrote in my previous post -, thanks to +3 morale he could get me for sure. It was like playing cat and mouse but I started as the former and became the latter! This battle was fun!
Anyway I cannot guess why an AI hero would prefer to merge his troops and therefore miss the opportunity to cast more spells (fairy dragons) or fire more shots (medusas). Does it make them feel stronger perhaps?
Wandering creatures never merge their troops to a single stack. Instead they will divide them into more stacks if they face many shooters.
I think we should track a former New World Computing developer down and make them spill the beans!
This is another thing for which I do not have a detailed explanation, I just know it exists at a certain threshold where the AI will split. I suppose the exact reasons are similar to why AI will not exit the garrison. It's an assessment of how much damage you can deal to AI as shown by these screenshots which can be figured out by testing a bit.
The moment you bring 17+ Archdevils, Medusae merge.
Any less and they split into 7 stacks.
A proof that it isn't simply the number of troops you use: 17 Archdevils but I unequipped 'Sword of Judgment' so -5 all stats comparatively.
An interesting use for this is shown by Richard Chen in his firewall video. Knowing that AI splits shooters if you use weak force, he took advantage of reduced numbers to kill them with firewall.
Do you have the Dragon Scale Armour? Maybe it only considers attack and deffense.
I saw the point when Sal filled several pages to find the information for neutrals split mod but if we succed in creating such a detailed enciclopedia of exploits mods will follow and I'm guessing the gap between the people that play the game without too much shared information and the puzzlers that can make and play maps such as this one will grow even more. Don't get me wrong, Metataxer looks amazing in the editor and this walktrough proves how sophisticated it really is but I'm not sure how to look at this trend, I want my game to be challenging enouch but I also want to have fun and that doesn't happen if I can't play the map.
Addressing this topic, I also can only guess, but from my experience, power and knowledge are never considered by anything. The only passive check that exists in regards to magic is if there is a probability to cast certain spells (AoE damage for example - does not influence troop split however I think). How much medusae were you able to kill in the Screenshots above? It would be possible that the AI splits it's troops for damage mitigation reasons... also I'm not sure if bow of the sharpshooter influences troop split.