|
| Thread: A prespective from the Ubisoft generation |  This thread is pages long: 1 2 3 4 5 6 7 8 · «PREV / NEXT» |
|
B0rsuk
Promising
Famous Hero
DooM prophet
|
posted August 09, 2011 10:48 PM |
|
|
For me Heroes 4's biggest flaw wasn't balance, wasn't "too many changes" or "heroes in combat drop like flies". It was the AI.
I blame AI for poor campaigns. They had good story I actually enjoyed reading, but mechanically they were weak. There were tents/gates everywhere and AI was barely able to move on its own. Single scenarios are just not enjoyable to play because AI can't navigate a map and come close to being a threat.
Numbers can be tweaked. Tweaking AI is much harder. If AI was good, scenarios would be more open-ended, and the game could be eventually patched by players.
The game was complex and that likely contributed to difficulties AI had. Skills differed vastly. "Might" heroes could be anything - a stealthy thief who spies on enemy and makes army move faster, a lord who increases creature growth, a knight who boosted creature stats, or a conan-style barbarian.
They could have made Potion of Immortality work like this: if your hero is killed in combat, he comes back after combat. That would prevent ridiculous barbarian rushes AND allow players to continue with their offensive.
____________
http://www.youtube.com/watch?v=u5um8QWWRvo RSA Animate - Smile or die
|
|
Corribus
Hero of Order
The Abyss Staring Back at You
|
|
posted August 09, 2011 10:53 PM |
|
|
|
I've always said that H4 could have been the greatest HOMM game, if only it had had a better AI, a more inspired artistic design, and an editor with any documentation whatsoever. I think the design struck a nice balance between trying new things and retaining some core HOMM elements. Unfortunately the broken AI just caused it to be stillborn.
|
|
alcibiades
Honorable
Undefeatable Hero
of Gold Dragons
|
|
posted August 10, 2011 08:46 AM |
|
|
I think H4 could have been better than H3 if it had had a better AI, different graphics, different factions, different unit systems, had included unit upgrades, had had a balanced heroes-in-combat system, had had a better magic system, and ... well, been a different game. 
____________
What will happen now?
|
|
JollyJoker
Honorable
Undefeatable Hero
|
|
posted August 10, 2011 08:57 AM |
|
|
*Sigh*
Let me repeat for the umptieth time, that it's a design flaw in the difficulty level concept, that kills H IV's AI - no matter any behaviour problems like in sieges. The following is a short sketch why no amount of AI tweaking would have been enough to give the game a good AI:
H IV has a brilliant difficulty model. Depending on difficulty level, wandering stack numbers are multiplied with a factor (2/3, 3/2, 2, 3), while retaining the overall XP value of a stack (which means that creature XP is divided by that number), so that you will have to fight three times the numbers of creatures as in normal for the same overall XP.
Unfortunately the AI faces the same multipliers.
That means, it's basically impossible for the AI to play a satisfactory game on high difficulty levels. The game would have needed a different model, so that for the AI players stacks would have had to look different.
To give an example, if a stack would consist of 30 Minotaurs originally (that is on normal difficulty and the way the map was set up), the human players would face 20, 30, 45, 60 and 90 Minotaurs for Easy, Normal, Advanced, Expert and Impossible diff (giving all the same XP, so that heroes are not levelling faster on higher diffs)). However, the AI would have to play under the same rules which is of course killing the game. In H IV you can beat those wild amounts of opposition only in tricky battles making use of certain spells and creatures, and the AI makes use of Quick Combat which is of course calibrated differently. Consequently the AI simply cannot handle the amount of creatures it faces on higher diff. (For those who cannot imagine that, think about HoMM 5, and now imagine that stacks are TRIPLED in numbers: you play a regular map and the Sawmill is guarded by 170 Skeletons. Great. Now imagine, what must happen, that the AI will win such a fight with acceptable losses - that is, when the AI is actually attacking the Sawmill.)
INSTEAD, the AI would have to face stuff THE OTHER WAY ROUND, that is 60, 45 30, 30, 20 (the AI would have been unable to handle highest diff anyway). Which means, on Impossible difficulty a human player would face 90 Minotaurs, but an AI player only 20 - for the same XP.
So that's a basic design flaw that makes it impossible to play a game on high difficulty levels without heavily scripted AI, which simply means, that in H4 you cannot play any MP map Single Player on more than Advanced Difficulty, and considering that the Expansions had somewhat tough map set-ups (these expansions were for veteran players, not the lukewarm stuff you get today), an expansion map would be no fun above Normal.
|
|
Zenofex
Responsible
Legendary Hero
Kreegan-atheist
|
|
posted August 10, 2011 09:36 AM |
|
|
|
There is no point in listing the hundreds flaws of Heroes IV, they are - by far - not related only to the AI or the the balance. And I strongly disagree that there was anything wrong with the editor - it was actually even better than the Heroes III one and only lacked large group landscape objects to make the work with it really easy. For me everything about Heroes IV can be summarized in: underdeveloped. Everything could have been fixed (with A LOT of work) but it remained unfixed and that's that.
|
|
Raelag84
Famous Hero
|
|
posted August 10, 2011 02:52 PM |
|
|
To claify what I said earlier I don't think Heroes of might and magic III is a bad game. I have even gone back to playing it...More or less. All that I am really saying was that Heroes V was not judged on it own terms by a lot of people.
Heroes III really does have something good about it that you can't put your finger on. There is just something beatiful about it. When it comes to the heroes games I find Ubisoft is better at science than at art. 3DO on the other hand had some real artist no doubt.
All that said I still find Heroes V to be my prefred game.
As a side note there is one other minor issue that fasinates me that I have seen discussed.
Dungeon III VS Dungeon V.
|
|
Maurice
Hero of Order
Part of the furniture
|
|
posted August 10, 2011 03:17 PM |
|
Edited by Maurice at 15:18, 10 Aug 2011.
|
Quote:
For me Heroes 4's biggest flaw wasn't balance, wasn't "too many changes" or "heroes in combat drop like flies". It was the AI.
I blame AI for poor campaigns. They had good story I actually enjoyed reading, but mechanically they were weak. There were tents/gates everywhere and AI was barely able to move on its own. Single scenarios are just not enjoyable to play because AI can't navigate a map and come close to being a threat.
Actually, I was quickly running up against the wall with regards to the implementation of the battle maps. The view angle was horrible, the center of your line was not directly facing the center of the enemy line, most creatures acted in absurd gestures and movement was awkward in the best of cases. The fact that Heroes also participated in battles - and could be killed! - was also something I seriously resented. Early game your Heroes had to be babysit to prevent them from dieing, late game they were one-man armies. Siege battles may have had the pro of actually having to place units in defense towers (something I really want to see in future Heroes titles!), but the central defense tower was in my opinion out of whack.
I agree with you, however, that some scenario's were frustratingly stacked against the player, with one-way teleports; the entrance of course next to the enemy's main castle, the exit next to yours. While I played it (which was only briefly), I can't count the number of times I had to return my main hero to my main castle to either recapture my castle or to smack around an enemy hero who popped up, instead of covering ground towards the enemy castle(s). Basically, you needed two strong Heroes before you could advance.
|
|
Corribus
Hero of Order
The Abyss Staring Back at You
|
|
posted August 10, 2011 03:36 PM |
|
Edited by Corribus at 15:37, 10 Aug 2011.
|
@JJ
To me, that seems like an easy enough problem to fix. Just fix it so that the AI doesn't experience the stack multiplier.
But AI suffered from other problems. I distinctly remembered it would just walk by empty mines, empty castles, etc. without flagging them, suggesting it had a much deeper problem prioritizing what it was doing.
____________
I'm sick of following my dreams. I'm just going to ask them where they're goin', and hook up with them later. -Mitch Hedberg
|
|
War-overlord
Responsible
Undefeatable Hero
Presidente of Isla del Tropico
|
|
posted August 10, 2011 03:42 PM |
|
|
Quote:
As a side note there is one other minor issue that fasinates me that I have seen discussed.
Dungeon III VS Dungeon V.
On this topic, I'd say they both had their pros and cons. Most of them aesthetic.
III Pros: Great atmosphere, gave a realy brooding evil feel.
III Cons: Lack of cohesion, just a bunch of evil and semi-evil things joined together on a lackluster pretense.
V Pros: Good, believable cohesion & Architecturally very pleasing.
V Cons: Sexuall sell-out & Overly Warhammer inspired.
Playing Dungeon in III: It was my 3rd favorite faction, but it suffered the overall sameness that I found everything suffered in III
Playing Dungeon in V: More Unique, but it doesn't suit my style of playing. Which relies a great deal on smashing, i.e. doing a lot of damage and being able to take damage in return. Dungeon's High Damage, Low survivability, Low Growth does not suit that at all. (Stronghold and Fortress suited it a lot better.)
I am a great advocate of making Minotaurs the Main Race of Dungeon in Ashan. I'm not a great fan of Dark Elves. That said, I am confident that it is doable to make an enjoyable Dark Elf Dungeon.
____________
Vote El Presidente! Or Else!
|
|
JollyJoker
Honorable
Undefeatable Hero
|
|
posted August 10, 2011 04:20 PM |
|
|
Quote:
@JJ
To me, that seems like an easy enough problem to fix. Just fix it so that the AI doesn't experience the stack multiplier.
But AI suffered from other problems. I distinctly remembered it would just walk by empty mines, empty castles, etc. without flagging them, suggesting it had a much deeper problem prioritizing what it was doing.
If it IS an easy enough problem to fix - why didn't anyone fix it?
Admittedly, the AI indeed was desoriented, to put it mildly. The thing is, however, as long as this basic problem isn't fixed, no amount of classy prioritizing will help, since the big stacks that come with higher diff levels are too high a hurdle for "QuickCombat".
That said, the AI in Heroes 3 has one big flaw as well, and that's hero building. I remember having made a lot of tests with the AI and what skills it picks when building a hero, and the result was abysmal. The AI builds absolutely crappy heroes, and that's the main flaw - you hardly ever see it cast mass spells, sce the AI doesn't pick magc schools.
The combat AI ain't bd however. I made the Battleground maps with the specific purpose in mind to let the player fight against well-developed enemy heroes, and the result is pretty good. They ve you a good run for your money.
|
|
shyranis
Promising
Supreme Hero
|
|
posted August 10, 2011 04:34 PM |
|
|
Quote:
Here's "my experience", part of the oft-recorded history concerning MM. (I still don't know if HoMM-1 had an Editor, so zero exp. there)
Homm1 had an editor, but only the Win95 version =)
____________
Youtube has terminated my account without reason.
Please express why it should be reinstated on
Twitter.
|
|
alcibiades
Honorable
Undefeatable Hero
of Gold Dragons
|
|
posted August 10, 2011 05:13 PM |
|
|
Quote:
That said, the AI in Heroes 3 has one big flaw as well, and that's hero building. I remember having made a lot of tests with the AI and what skills it picks when building a hero, and the result was abysmal. The AI builds absolutely crappy heroes, and that's the main flaw - you hardly ever see it cast mass spells, sce the AI doesn't pick magc schools.
Didn't H3 AI just pick whatever skills it was offered? At least that was the impression I got, it didn't seem to value skills higher than others. And I guess in a way that makes sense, if developers felt some skills were worth less than others, they were directly admitting there were balance issues among the skills, so conversely, that means if they felt skills were ballanced, it made sense to just let AI pick whatever it was offered, and use the distribution rates to control which skills were offered on which classes.
____________
What will happen now?
|
|
B0rsuk
Promising
Famous Hero
DooM prophet
|
|
posted August 10, 2011 06:11 PM |
|
|
In all Heroes games I played, the AI seems to have no concept of building heroes. Not just skills - it assigns whatever scraps it can find to random heroes. There are guys with 1 or 2 weeks worth of units running everywhere. Then when you eventually meet a bigger fish, it's only because map designers gave him 4 castles compared to your 2.
____________
http://www.youtube.com/watch?v=u5um8QWWRvo RSA Animate - Smile or die
|
|
JollyJoker
Honorable
Undefeatable Hero
|
|
posted August 10, 2011 06:24 PM |
|
|
No, Alcibiades, it's not random.
There's some crappy prioritizing going one. There is a way to test that as well. You have to make a 2-player-map (rudimentary) with a spot that offers enough XP for the number of level-ups you want.
If you start the game (with you on first position), and then save it, the "development tree" is fix for the game.
You can load it as hotseat and check things first, forming a picture of what is possible.
You then can let the AI play a turn and check afterwards what it picked.
|
|
alcibiades
Honorable
Undefeatable Hero
of Gold Dragons
|
|
posted August 10, 2011 06:33 PM |
|
|
Ok, I never really game much thought to it, nor did I do anything serious to test it, but it seemed to me that whenever you bought a high-level enemy hero from the tavern, they had an uncanny nack for having learned Eagle Eye, Mysticism and First Aid, so I just assumed they took whatever got offered to them.
EDIT > Btw. JJ, how do you know that it DID not just take what was offered? I reckon you'd have to do a rather large number of test in order to statistically proove that the choices deviated from the statistical distribution you'd expect based on the particular hero class.
____________
What will happen now?
|
|
Warmonger
Promising
Legendary Hero
fallen artist
|
|
posted August 10, 2011 06:38 PM |
|
|
Obviously. How could AI figure out that some of skills are useless, if even game designer didn't expect it to be so?
Still, these skills may be very handful in ceratin situations and which certain strategy, but not in a generic free-for-all scenario.
Keep in mind that we have over 10 years of experience with this game, can't blame Ai that it didn't improve during that time on its own.
____________
The future of Heroes 3 is here!
|
|
Corribus
Hero of Order
The Abyss Staring Back at You
|
|
posted August 10, 2011 06:44 PM |
|
|
@Borsuk,
That was a change from H2. In H2, the AI tended to accumulate massive armies in one hero, and all other heroes had nothing. Of course, the H2 AI was susceptible to the divide and conquer tactic - it would always leave half its army in a conquered castle for defense.
They changed that in H3, and more evenly distributed (although not always sensibly) trooped over multiple heroes. Which of course is not the way humans usually play. So H3 could usually be beaten fairly evenly without a massive troop advantage.
As a mapmaker, it took me some time to adjust from making H2 maps to H3 maps at high difficulty level.
Of course, with scripting capability of WoG, these sorts of issues can easily be compensated for.
____________
I'm sick of following my dreams. I'm just going to ask them where they're goin', and hook up with them later. -Mitch Hedberg
|
|
JollyJoker
Honorable
Undefeatable Hero
|
|
posted August 10, 2011 06:48 PM |
|
|
Their first priority is to to keep true to their orientation: A might hero will pick Wisdom only when the other option is "worse".
Also, if I remember right, the AI is keen on filling all slots as soon as possible.
That means a might hero will have Scouting, Artillery, Pathfinding and Ballistics as a matter of fact.
With Magic heroes it's the same. Eagle Eye, Mysticism, Learning, Sorcery - the magic schools are at the end of the food chain, but magic heroes WILL pick them, provided the alternative is a silly might skill like Attack for example. 
|
|
Salamandre
Admirable
Omnipresent Hero
Wog refugee
|
|
posted August 10, 2011 07:51 PM |
|
|
You can make AI learn exactly what skills you want for him, simply by unchecking others in the SOD editor options. With WoG this is even automatic, a simple line script and AI has all goodies. Or you still dream about a random map giving you a fair challenge? We know it doesn't, so now the question is what to do, not what wasn't done and blame UBI- not productive.
Those things must be planned. What is different from AI and human in battle? Not only bad evaluation of targets, but also short vision in which spell to use. With the last WoG version (ERA 1.81) we have now access to code its behavior, increase the value of spells as blind, force field and berserk. I am fairly certain that we can code it to fight much better, not only by giving him more troops or stats.
There is another problem which has never been mentioned: the human cheats even more than AI, by reloading as soon as he does not like the outcome. Then blame the AI for poor challenge. Try once to play a map without any reload, you will see how it goes.
____________
Era II mods and utilities
|
|
alcibiades
Honorable
Undefeatable Hero
of Gold Dragons
|
|
posted August 10, 2011 07:54 PM |
|
|
Quote:
Their first priority is to to keep true to their orientation: A might hero will pick Wisdom only when the other option is "worse".
Also, if I remember right, the AI is keen on filling all slots as soon as possible.
That means a might hero will have Scouting, Artillery, Pathfinding and Ballistics as a matter of fact.
With Magic heroes it's the same. Eagle Eye, Mysticism, Learning, Sorcery - the magic schools are at the end of the food chain, but magic heroes WILL pick them, provided the alternative is a silly might skill like Attack for example.
That IS indeed weird prioritizing, but that would explain why it was so rair to find AI heroes with mastery in one for not to say several magic schools.
____________
What will happen now?
|
| |
|
|
