Heroes of Might and Magic Community
visiting hero! Register | Today's Posts | Games | Search! | FAQ/Rules | AvatarList | MemberList | Profile


Age of Heroes Headlines:  
5 Oct 2016: Heroes VII development comes to an end.. - read more
6 Aug 2016: Troubled Heroes VII Expansion Release - read more
26 Apr 2016: Heroes VII XPack - Trial by Fire - Coming out in June! - read more
17 Apr 2016: Global Alternative Creatures MOD for H7 after 1.8 Patch! - read more
7 Mar 2016: Romero launches a Piano Sonata Album Kickstarter! - read more
19 Feb 2016: Heroes 5.5 RC6, Heroes VII patch 1.7 are out! - read more
13 Jan 2016: Horn of the Abyss 1.4 Available for Download! - read more
17 Dec 2015: Heroes 5.5 update, 1.6 out for H7 - read more
23 Nov 2015: H7 1.4 & 1.5 patches Released - read more
31 Oct 2015: First H7 patches are out, End of DoC development - read more
5 Oct 2016: Heroes VII development comes to an end.. - read more
[X] Remove Ads
LOGIN:     Username:     Password:         [ Register ]
HOMM1: info forum | HOMM2: info forum | HOMM3: info mods forum | HOMM4: info CTG forum | HOMM5: info mods forum | MMH6: wiki forum | MMH7: wiki forum
Heroes Community > Heroes 3.5 - WoG and Beyond > Thread: Official VCMI Thread
Thread: Official VCMI Thread This Popular Thread is 116 pages long: 1 10 20 30 40 50 60 70 ... 79 80 81 82 83 ... 90 100 110 116 · «PREV / NEXT»
avatar
avatar


Promising
Supreme Hero
posted November 17, 2016 09:35 AM

stachnie: to make game much harder, try trick described in this topic (one before last post)
http://forum.vcmi.eu/viewtopic.php?t=1157
____________

 Send Instant Message | Send E-Mail | View Profile | Quote Reply | Link
stachnie
stachnie


Known Hero
posted November 17, 2016 10:46 AM

Warmonger said:
How would you "correct it"? Describe "correct" AI.


No AI would match a skilled human player, but it does not need to be extremely dumb. Let's have a look at H5: the AI is not very smart, so at higher levels it gets significant bonuses. Quantomas is developing a new AI, so in his H5 mod AI does not need "unfair" bonuses, sometimes it can make things surprising to the player (like casting a few Armageddons during a battle, if this hurts the opponent more than AI itself). This AI is present in a big H5 mod called Might and Magic Heroes 5.5 or MMH55.

Original H3 AI is not very wise (e.g. it fails to properly use combo artifacts, especially the Necro one producing Liches) but present VCMI AI is much worse. E.g. it buys a lot of weak heroes and sends into enemy lands while the strongest units are left at home unpurchased. I am not sure if there is a simple way to use Q's AI in VCMI but this would be a HUGE steep ahead.

S.

 Send Instant Message | Send E-Mail | View Profile | Quote Reply | Link
Warmonger
Warmonger


Promising
Legendary Hero
fallen artist
posted November 17, 2016 11:20 AM
Edited by Warmonger at 11:22, 17 Nov 2016.

Quote:
E.g. it buys a lot of weak heroes and sends into enemy lands while the strongest units are left at home unpurchased

I didn't ask what AI does wrong (we all know it), but how it should act right. That's quite a difference. If you can describe correct algorithm for hero purchase that works in all cicrumsances, then we can talk.
Quote:
I am not sure if there is a simple way to use Q's AI in VCMI

Well, I'm sure there is not.
____________
The future of Heroes 3 is here!

 Send Instant Message | Send E-Mail | View Profile | Quote Reply | Link
Galaad
Galaad

Hero of Order
Li mort as morz, li vif as vis
posted November 17, 2016 12:10 PM

Yep, like Avatar says Macron's trick of giving extra resources by editing startres.json makes it better.
____________

 Send Instant Message | Send E-Mail | View Profile | Quote Reply | Link
stachnie
stachnie


Known Hero
posted November 17, 2016 12:49 PM
Edited by stachnie at 18:59, 17 Nov 2016.

Warmonger said:
If you can describe correct algorithm for hero purchase that works in all cicrumsances, then we can talk.


I am not a programmer, so I cannot help there. But if you have some numerical problem to solve, I may try to do that.

Quote:
Quote:
I am not sure if there is a simple way to use Q's AI in VCMI

Well, I'm sure there is not.


... or not a very difficult way (I am sure Copy&Paste would not work) to implement it.

Giving AI more resources may be a good workaround, but temporary only. At least this is more honest than giving AI extra creatures.

S.

 Send Instant Message | Send E-Mail | View Profile | Quote Reply | Link
revolut1oN
revolut1oN


Famous Hero
posted November 17, 2016 02:33 PM

Warmonger said:

I didn't ask what AI does wrong (we all know it), but how it should act right. That's quite a difference. If you can describe correct algorithm for hero purchase that works in all cicrumsances, then we can talk.


Sorry for offtop, but why are you so bitter? People give suggestions because they care about your project so in my opinion its no need for a grumpy old man speeches and big headed responses.

 Send Instant Message | Send E-Mail | View Profile | Quote Reply | Link
LizardWarrior
LizardWarrior


Honorable
Legendary Hero
the reckoning is at hand
posted November 17, 2016 04:12 PM
Edited by LizardWarrior at 10:07, 18 Nov 2016.

stachnie said:

I am not a programmer, so I cannot help there.


I will fire a shoot in the dark and take the bait! It may be totally unreliable, but whatever Make a rank-paired heap as a priority queue for each AI to , so you get O(log n) time for calculating top priority. Assign each map location an AI value for guidance, or better assign an union for each AI value so it can be dynamic and change according to game time, as well as occupying minimal space, resource piles would have a high value early and this value gets lower by passing week/month. Towns and artifacts(depending on their type) will have highest priority. Now as most stuff on the map is guarded there must be some genetic algorithm that will have a fitness function based on game time and AI army/needs, this will be basis for calculating priorities, it needs to take into account distance, guards and the AI value of the adventure object. A Monte-Carlo tree could be used too.

First compute the distances between AI hero and places of interests (for starting there can be used only Manhattan distance since it's O(1)) that are located within hero's movement, use a function to evaluate the object according to both relative distance and AI value, sort them, chose the best one then try to compute the actual path with A*(treat guards as high cost nodes, so it tries to avoid them), check if the path collides with any guards, check guards AI power, compare it with AI's one (AI should already have a summed power as a field in the class), use all those factors: movement cost, guard power, AI army power, AI object value, game time as arguments for the fitness function if the AI should go for it. If it's over a certain threshold, take the next most promising object and repeat. If the AI has marked/visited all viable or has no objects in sight, then it goes exploring or back to town (to replenish army or spells). There should be hero only flags and AI-wide flags for map objects, hero flags for 1-time per hero visiting stuff and AI-wide flags for 1-visit only objects.

Now AI should always have full hero roster, so there needs to be another function that checks if the AI is short on cash and needs to save up to buy an important building or recruit creatures, but in general is if there aren't 8 heroes, the AI should recruit. Number of heroes should have a priority that diminishes the more heroes he has, so the AI would hire more heroes only if the AI Power is over a certain threshold or there are not creatures available to buy, as well as no building outclasses its priority. If there are more towns where the AI can recruit heroes, then the town's LOS should be searched with flood-fill for important map objects, the function would also take into consideration the heroes that the AI possess inside the LOS, so it doesn't hire more heroes in an over-saturated area. The function could take as arguments the sum of AI value of map objects and divide it (or a more complex formula) to the number of existing heroes inside OR the number of map objects, so you don't hire 4 heroes for only 1 Utopia or such .There should also be flags for main and secondary heroes, the main hero will not go after objects that are under a certain level of priority (you won't send your main off path just to waste movement picking a pile of ore during month 3). A simple approach for dividing armies, is that the main should get a certain percent of the total AI power, while the rest would be given to secondaries. Chaining would be harder to program, but making the towns and main hero fluctuate as priorities for secondary heroes. If the main hero falls behind a minimum AI power, then it goes back to town or towards a secondary hero to replenish the army. Now as for engaging the enemy, if there's an enemy hero within AI's hero line of sight, then it compares army power/primary skills/levels, if it considers it's a good fight, then it engages, else flees.

As for town building, there need to be built some custom trees for each faction, taking into consideration costs, stockpiled resources and building priority. If the AI doesn't have resources to build the desired building then it chooses the next available top priority building that it's in a certain range priority-wise from the intended one, else it waits and increases priority for certain resources on adventure map, or it may be even programmed to use the market. But market is another monster of its own.

Yeah, I know it's also more about formulas and coming up with a good algorithm, but for a rough sketch I think that's how it should behave. Please note that I'm not an expert by any means.
____________

 Send Instant Message | Send E-Mail | View Profile | Quote Reply | Link
Warmonger
Warmonger


Promising
Legendary Hero
fallen artist
posted November 18, 2016 09:13 AM
Edited by Warmonger at 09:14, 18 Nov 2016.

Quote:
why are you so bitter? People give suggestions because they care about your project so in my opinion its no need for a grumpy old man speeches and big headed responses

It has been explained many times that AI is ambitiou and half from finished. Yet people still come up and talk about "bugs" and "fixes" as it could be patched overnight. No it can't be patched overnight. It's a lot of work and design from ground-up.

Quote:
First compute the distances between AI hero and places of interests (for starting there can be used only Manhattan distance since it's O(1)) that are located within hero's movement, use a function to evaluate the object according to both relative distance and AI value, sort them, chose the best one then try to compute the actual path with A*(treat guards as high cost nodes, so it tries to avoid them), check if the path collides with any guards, check guards AI power, compare it with AI's one (AI should already have a summed power as a field in the class)

This is already done, but individual object evaluation is missing. There are some objects important to AI (by default: all objects it can capture)

Quote:
use all those factors: movement cost, guard power, AI army power, AI object value, game time as arguments for the fitness function if the AI should go for it. If it's over a certain threshold, take the next most promising object and repeat.

This is exactly how it works now. I used fuzzy logic for aggregation.

However, I don't see how you wnat to use "game time" as an argument. What does game time change in logic?

Quote:
Now AI should always have full hero rooster

Well it does. And that's what people are complaining about - sometimes it doesn't make sense on a small map with nowhere to go.

Quote:
needs to save up to buy an important building or recruit creatures

Now that's a serious problem - how to distribute cash between recruiting creatures and construting buildings? How to save resources for buildings over many turns? When should AI recruit creatures and in which towns? (currently it spends all but 10K at the end of the turn). Keep in mind you can also get creatures from dwellings, allied heroes and even banks.

Quote:
if there's an enemy hero within AI's hero line of sight, then it compares army power/primary skills/levels, if it considers it's a good fight, then it engages, else flees

AI can't flee now - it was not programmed. Also AI can't defend towns. Also the simple question - where should AI flee and what should it do when it can't? What if AI can quickly get army form nearby dwelling or allied hero to be strong enough not to flee?
____________
The future of Heroes 3 is here!

 Send Instant Message | Send E-Mail | View Profile | Quote Reply | Link
LizardWarrior
LizardWarrior


Honorable
Legendary Hero
the reckoning is at hand
posted November 18, 2016 10:01 AM

Warmonger said:

However, I don't see how you wnat to use "game time" as an argument. What does game time change in logic?


Priorities changes over time, in month 3 you won't have the main hero pick up lowly piles of resources instead of rushing for an enemy town for example. It's more of an assumption, as AI should have enough resources as time passes by, so it doesn't waste movement with its main. Let's call this a "decay" factor, when you got 40k gold stockpiled and there's an enemy town to attack nearby, would you get distracted by a gold pile? Or a crypt?

Warmonger said:

Well it does. And that's what people are complaining about - sometimes it doesn't make sense on a small map with nowhere to go.


First if the total AI power is over a certain amount(this should be relative to the game time) and if the "hero recruit" priority is higher than the current building priority or if it already build in the town, then recruit hero. It may also be limited at one per day per town and as I suggested, flood-fill could be applied for each town's LOS that the AI controls, then sum up the AI value of individual unvisited objects found, divide this value by the number of heroes present in town's LOS and the number of objects (as you don't need more heroes for only 1 important object with high AI value). If it's below a certain number, then it recruits an hero in the said town, otherwise it doesn't.

Quote:

Now that's a serious problem - how to distribute cash between recruiting creatures and construting buildings? How to save resources for buildings over many turns? When should AI recruit creatures and in which towns? (currently it spends all but 10K at the end of the turn). Keep in mind you can also get creatures from dwellings, allied heroes and even banks.


Something like this I have in mind, if the AI is over a certain army power that changes with game time, it will check the town with the highest building priority, if it doesn't have resources, then it will try to build the next most prioritized building, whose priority is in close range to the desired one, else it goes exploring and rechecks if it got enough resources at the end of the turn. Disposable income can be calculated too, subtracting the cost of creatures needed to be recruited to go over that certain threshold and the cost of highest priority buildings in each town, which will be either saved, or used to buy the remaining creatures left to recruit.

Quote:

AI can't flee now - it was not programmed. Also AI can't defend towns. Also the simple question - where should AI flee and what should it do when it can't? What if AI can quickly get army form nearby dwelling or allied hero to be strong enough not to flee?


AI could run towards the nearest secondary hero or make the secondary hero come towards the main, if that hero has an army value that summed with his main could overcome the threat level of the enemy hero, or run towards the nearest town. It could also check it's movement area for dwellings and sanctuaries. Of course the AI would also check the enemy's LOS/area of movement to don't transit it, squares that would be reachable by the enemy hero in 1 could be treated as high cost nodes, so the AI won't chose this path unless it's the only option.

____________

 Send Instant Message | Send E-Mail | View Profile | Quote Reply | Link
Macron1
Macron1


Supreme Hero
posted November 18, 2016 11:28 AM

AI heroes do not flee from battle.
Maybe do it this way:
If player summary fighting power of army is bigger than ai's and as summary fighting army power is lower than threshold from defaultMods.json value, than try to flee from battle or surrender if AI has sufficient money

 Send Instant Message | Send E-Mail | View Profile | Quote Reply | Link
AlfWithCake
AlfWithCake


Known Hero
posted November 18, 2016 10:01 PM

LizardWarrior said:
Warmonger said:

However, I don't see how you wnat to use "game time" as an argument. What does game time change in logic?


Priorities changes over time, in month 3 you won't have the main hero pick up lowly piles of resources instead of rushing for an enemy town for example. It's more of an assumption, as AI should have enough resources as time passes by, so it doesn't waste movement with its main. Let's call this a "decay" factor, when you got 40k gold stockpiled and there's an enemy town to attack nearby, would you get distracted by a gold pile? Or a crypt?



 Not quite sure what's going on here, but I'd like to give my 50 cents. AI shouldn't consider game time, he must consider the situation. Thus he must attack enemy town not because it's high time for this, but because he's got high leveled hero and huge army and this would make much troubles for his enemy. So I want to say, that AI must act accordingly to the gathered information, and game time can't say anything besides "enemy could be strong" and "tomorrow will be new week", and first expression is useless, as you can't evaluate precisely how strong enemy could be.

 Send Instant Message | Send E-Mail | View Profile | Quote Reply | Link
stachnie
stachnie


Known Hero
posted November 21, 2016 07:58 AM

My remarks after some testing.

VCMI is quite stable and has a great potential. Crashes or freezes every (let's say) dozen of turns are manageable - If I did not like that, i would not play it.

Obvious bug: it is impossible to build a middle tier shooters (some Druids?) upgrade building in Preserve, even if there are enough resources. Neither AI nor me.

Unavoidable crashes:

- when equipping a combo artifact (like the Archery one), do not touch the bottom left slot (but it may be filled - just touch e.g. the bottom right slot)

- I wanted to prepare a new hero, gave his Commander some archery-boosting artifacts and a Flaming Arrow; a shot results in a crash (is there a way to take these artifacts back?).

@Warmonger: please do not take any statement critisizing AI personally. For me this is the weakest point of VCMI but I am not about to blame anyone for it. I guess skilled AI programmers are not growing on trees, but maybe there is a way to at least learn anything from successful projects like H5 AI mod and projects which use it (e.g. MMH55). But even if you manage to get the source code, implementation may be very difficult (especially if it is written in a different language - I guess VCMI is written in some clone of Pascal, AI mod may be written in C).

Things that (perhaps?) may be quite easily corrected: with enough resources AI builds the town and buys all creatures but seems to leave most of them at home. It may take neutral towns, but much later than it could be done. My suggestions:

- AI should have one-two strong heroes, maybe a troop transport hero and some weak collectors.

- If a hero attacks a creature bank, his army should be a few times stronger than expected guards.

- If this bank gives creatures, at least one creature slot should be empty.

- Unwanted troops (e.g. from Diplomacy, given by a bank or a seer) should be given to weaker heroes who possibly could upgrade their initial troops and give them to the garrison.

Another suggestion: an option to regenerate creature banks (after e.g. 2 months). It could be done in a way similar to MMH55: the strength of monsters may slowly increase with time (after a few months one may face Dragon Utopias with 4 groups of dozens of Dragons and some other creatures like Unicorns, Elementals or even Phoenixes). It should not be similar to H4, where both monsters and treasures increased with time and when a bank was regenerated, one could defeat a very poor garrison (like 1 Pikeman and 3 Crossbowmen or 1 Black Dragon) and get a loot as if the bank had been not visited at all.

S.

 Send Instant Message | Send E-Mail | View Profile | Quote Reply | Link
Warmonger
Warmonger


Promising
Legendary Hero
fallen artist
posted November 21, 2016 08:13 AM
Edited by Warmonger at 08:23, 21 Nov 2016.

Quote:
AI heroes do not flee from battle.

This belongs to BattyleAI.

Quote:
Unwanted troops (e.g. from Diplomacy, given by a bank or a seer) should be given to weaker heroes

How to determine what are "unwanted troops"? What if AI gets 1000 Devils from Seer?

Quote:
AI should have one-two strong heroes, maybe a troop transport hero and some weak collectors.

What if AI is limited to one wandering hero? What if AI starts with 8 heroes locked in separate locations?

Quote:
If a hero attacks a creature bank, his army should be a few times stronger than expected guards

It already does exactly that. "Expected guards" are evaluated by fuzzy logic, which in fatc means weighted sum.

Guys, I'm afraid you miss one thing: You have no idea how VCAI currently works. You just make some random statements on how it could possibly work, but don't get the bigger picture. You talk a lot, but never ask a question.
Did you bother reading Wiki and follow AI logs at least?
____________
The future of Heroes 3 is here!

 Send Instant Message | Send E-Mail | View Profile | Quote Reply | Link
avatar
avatar


Promising
Supreme Hero
posted November 21, 2016 08:48 AM
Edited by avatar at 08:49, 21 Nov 2016.

stachnie said:

Obvious bug: it is impossible to build a middle tier shooters (some Druids?) upgrade building in Preserve, even if there are enough resources.

You play very very outdated version. Preserve's newest version comes with new townscreen: https://www.mediafire.com/?s4et7hzwx16i6cv
So, nobody cares with fixing old H4-style buggy version (this version doesn't have external dwellings as well).

Quote:
- I wanted to prepare a new hero, gave his Commander some archery-boosting artifacts and a Flaming Arrow; a shot results in a crash (is there a way to take these artifacts back?).


Bugs with commander's artifacts have been fixed after 0.99 release, so you can play this with latest daily builds. Besides WoG mod is buggy, and no longer support. It it wise to play vcmi without this mod ;P

____________

 Send Instant Message | Send E-Mail | View Profile | Quote Reply | Link
stachnie
stachnie


Known Hero
posted November 21, 2016 10:46 AM
Edited by stachnie at 12:36, 21 Nov 2016.

Warmonger said:
Quote:
Unwanted troops (e.g. from Diplomacy, given by a bank or a seer) should be given to weaker heroes

How to determine what are "unwanted troops"? What if AI gets 1000 Devils from Seer?


One may define unwanted troops as e.g. a stack from a different alignment weaker than any other stacks in the hero army. Heroes should tend to have troops from only one alignment, maybe two or three.

Edit: if there is no room for new troops, it may be evaluated as follows: compare effective strength of the original army with all possibilities with one slot replaced with new troops. I would suggest the following formula:

strength := sum [n_i*XP_i*(1+p_i)]

where n_i is the number of creatures in slot i, XP_i is XP or relative strength of a creature in slot i and p_i is probability of a positive morale (positive number) or a negative morale (negative number).

p_i may depend on the slot if the army is a mixture of. e.g. living, mechanical and undead creatures. I would calculate, which combination gives the greatest strength. In most cases the result would be the original army (refuse to take the stack) or the weakest slot replaced with new troops.

Quote:
Quote:
AI should have one-two strong heroes, maybe a troop transport hero and some weak collectors.

What if AI is limited to one wandering hero?


Perhaps this hero should have a strong army. And a hero with one weak stack of creatures and lots of artifacts should not wander in proximity of a hostile player (I have defeated such a hero recently).

Quote:
What if AI starts with 8 heroes locked in separate locations?/quote]

AI should be able to decide to dismiss at least one of them (if some of them may meet with each other, they may combine their troops).

Quote:
Quote:
If a hero attacks a creature bank, his army should be a few times stronger than expected guards

It already does exactly that. "Expected guards" are evaluated by fuzzy logic, which in fatc means weighted sum.


Perhaps it should be something like "3x stronger than the strongest possible guards". I have seen some banks partially cleared (with some slots missing or weaker than expected), so I guess AI tried to clear them but it was defeated.

Quote:
Guys, I'm afraid you miss one thing: You have no idea how VCAI currently works.


Yes, I do not. But I have no idea about AI programming (my only programming experiences are some scientific calculations and small utility programs) so I may only hope that my observations, remarks, suggestions etc. may help to improve AI.

@Avatar: thanks.

S.

 Send Instant Message | Send E-Mail | View Profile | Quote Reply | Link
Warmonger
Warmonger


Promising
Legendary Hero
fallen artist
posted November 21, 2016 01:47 PM
Edited by Warmonger at 13:52, 21 Nov 2016.

Quote:

Edit: if there is no room for new troops, it may be evaluated as follows: compare effective strength of the original army with all possibilities with one slot replaced with new troops.

Okay, but this is already done. AI is exchanging troops between heroes whenever it can. Doesn't it?

Quote:
I would suggest the following formula:

strength := sum [n_i*XP_i*(1+p_i)]

This is the target formula. But how to maximize it? Imagine you have 7 stacks in your army (3 alignments) and other hero has 3 other stacks of different alignments. How many combination for morale are there? When will morale penalty overshadow army strength? How much is p_i, anyway?

Stack exchange is already more complicated than it may seem, and we would take much more factors into considerations than just morale.

Quote:
I have no idea about AI programming

There is nothing special in AI progrmaming - all basis is already done. But AI needs to be able to handle any situation, at any time. That's what makes it different from fixed scripts and simple reactions to events.
____________
The future of Heroes 3 is here!

 Send Instant Message | Send E-Mail | View Profile | Quote Reply | Link
stachnie
stachnie


Known Hero
posted November 21, 2016 03:50 PM
Edited by stachnie at 16:03, 21 Nov 2016.

Let's say we have a Dungeon hero with a mixed Dungeon and Rampart army with no Leadership and no other morale modifiers. IIRC morale for most of creatures will be 0 (for Minotaurs it may be 1 if their morale is always positive, I do not remember). All slots are occupied. The hero is offered a stack of Devils.

p_i is the probability of an extra move (positive) or to lose a move (negative). I do not remember the numbers - IIRC for H4 and H5 this is just morale/10, for H3 this is different (something like 0.125 for maximum morale, capped at 3, lower values for morale 1 or 2, similar for negative morale), I have no idea about VCMI.

sum 0: add n_i*XP_i*(1+p_i) for all 7 stacks, p_i is equal to 0 (not Minotaurs) or appriopriate number for morale 1 (Minotaurs).

sum 1: you replace the first stack with Devils. As far as I understand, the morale for most stacks is -1 (3 different towns). If Minotaurs have been not replaced, their morale is 1.

sum 2, 3... 7: like sum 1, but you replace nth stack with Devils.

Now you have 8 numbers: sum 0, sum 1... sum 7. You check, which sum is the greatest. If sum 0, you refuse the stack. If sum 1, you replace the first stack with Devils etc. You may do it e.g. like this:

nmax:=0;
max:=sum[0];
for n:=1 to 7 do if sum[n]>max then
begin
 nmax:=n;
 max:=sum[n];
end;

If nmax=0, no stack is replaced. If nmax>0, stack nmax is replaced.

If Commanders are enabled, you may need to include them as 8th stack (I do not know if they may change morale if e.g. Dungeon hero has a mixed Rampart and Inferno army).


S.

 Send Instant Message | Send E-Mail | View Profile | Quote Reply | Link
zeryss
zeryss


Promising
Famous Hero
http://i12.pixs.ru/storage/6/4
posted November 23, 2016 09:55 PM

PR?

 Send Instant Message | Send E-Mail | View Profile | Quote Reply | Link
stachnie
stachnie


Known Hero
posted November 26, 2016 07:41 PM

avatar said:
Bugs with commander's artifacts have been fixed after 0.99 release, so you can play this with latest daily builds.


It works, thanks. BTW: the crashes had been caused not by Flaming Arrow but by the fact that I had given an artifact boosting ranged damage to a commander who cannot shoot (from the Egyptian town - IIRC Bastion).

Another reason to crash is if you put off a combo artifact and place some other artifact to one of the emptied slots (maybe not the one marked by the artifact). In order to avoid crash one may put the combo artifact off, exit the hero screen, enter it again and fill empty slot(s).


S.

 Send Instant Message | Send E-Mail | View Profile | Quote Reply | Link
Warmonger
Warmonger


Promising
Legendary Hero
fallen artist
posted November 26, 2016 08:11 PM
Edited by Warmonger at 22:04, 26 Nov 2016.

Quote:
BTW: the crashes had been caused not by Flaming Arrow but by the fact that I had given an artifact boosting ranged damage to a commander who cannot shoot (from the Egyptian town - IIRC Bastion).

The crash was caused by moving ANY artifacts around.

Of course you can give artifact triggering effects on shoot to a stack that cannot shoot at all. I suggest you should focus on reporting bugs rather than making your own speculations
____________
The future of Heroes 3 is here!

 Send Instant Message | Send E-Mail | View Profile | Quote Reply | Link
Jump To: « Prev Thread . . . Next Thread » This Popular Thread is 116 pages long: 1 10 20 30 40 50 60 70 ... 79 80 81 82 83 ... 90 100 110 116 · «PREV / NEXT»
Post New Poll    Post New Topic    Post New Reply

Page compiled in 0.2682 seconds