Talk:AI value

From Heroes 3 wiki
Revision as of 18:56, 3 January 2024 by Imahero (talk | contribs)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigation Jump to search

Does anyone know what this does? The other "hidden" characteristic is Fight Value. –Kapteeni Ruoska (talk) 10:08, 3 December 2015 (CET)

I'm not sure, but my guess (mostly from the name) is that it affects how the creature will behave when you fight it in wandering stacks. I have noticed that higher level creatures tend to wait on the first turn, for example, or move to just outside a ranged creature's range (while lower level creatures do not), although I've never tested combat behaviour rigorously (I can't really imagine how one might, to be honest). Ibraesil (talk) 14:33, 3 December 2015 (CET)
It seems to me no one is entirely sure. I remember that I have tested AI and Fight values by changing them to CRTRAITS.TXT, but it did not seem to affect combat. The only real thing it changes is the experience Altar of Sacrifice gives the hero when creatures are sacrificed. –Kapteeni Ruoska (talk) 16:02, 3 December 2015 (CET)
Oh, and [1] and [2] have some info about it. –Kapteeni Ruoska (talk) 16:03, 3 December 2015 (CET)
There does seem to be some degree of consensus in those threads you linked that AI Value is used to determine the sizes of wandering stacks, based what they're guarding. Also, it might affect how the computer players cast spells. Also also, it seems as though Fight Value might be used to work out which creature to destroy when a hero goes through a whirlpool. Ibraesil (talk) 01:19, 4 December 2015 (CET)

How to see AI value?[edit]

There is a necessity to see AI value in hidden files in order to find AI value and Fight value for new Factory units.

See [3]. –imahero 18:56, 3 January 2024 (UTC)

I cannot understand why...[edit]

  • Gold Dragon > Titan & Ancient Behemoth
  • Phoenix > Ancient Behemoth
  • Giant & Bone Dragon > Behemoth
  • Manticore > Cyclops King (in fact basic Cyclops is more efficient)
  • Monk and Dendroid Guard are so low, behind Basilisk and Pegasus
  • Royal Griffin & Cerberus > Evil Eye
  • Obsidian Gargoyle > Orc & Marksman
  • Infernal Troglodyte & Pikeman > Master Gremlin

I admit in most cases it's harder to fight against wandering monsters on the left (therefore, this list is based on wandering monsters' potential), but why then are ranged Monks or Orcs so low? Unless having Archangels, advanced/expert Resurrection, Animate Dead, expert Forgetfulness or someone extremely resilient, you are almost confined to have casualities against ranged monsters, unlike Pegasi, Griffins, or Gargoyles who are really fast but can be just slowed down.--Dread Knight (talk) 07:29, 28 August 2021 (UTC)


 Q: Let’s talk about balance.  What methods did you use to make decisions about creature statistics, AI value, faction balance and schools of magic?  Was every single value chosen on a whim and then verified or rather were there more complicated mathematical calculations involved?
 A: Balance was determined using a process of opinion and verification, via abstract simulation.  JVC and I hashed out the statistics for the various troops, based on our personal opinion of which was more or less powerful, in relation to one another.
 Using these preliminary numbers, I put each Troop through a Battle Simulator I constructed in Microsoft Excel.  This Battle Simulator produced a relative number, which I then used to verify our assumptions.  If something appeared to be out of whack... an adjustment was made.
 Using these base ai values, Gus Smedstad further modified the numbers, to account for special abilities, creating adjusted ai values, which were again verified using the Battle Simulator.  If something appeared to be out of whack... an adjustment was made.
 Faction balance was less complicated.  I did an analysis of Troop Power versus costs over time. If something appeared to be out of whack... an adjustment was made.
 Spells and spell schools were largely opinion.
 Q: Were there dedicated testers and balance designers in the dev team of Heroes III?
 A: Short answer... no.  Please keep in mind, at the time, network play was a relatively new thing, and the thought of competitive HoMM3 was inconceivable.  Also, dedicated balance designers and testers didn’t really become a thing until Starcraft became an esport in South Korea.  Today, it is standard practice for any pure multiplayer game.

Greg Fulton, the lead designer of HoMM3 (RoE and AB)

Behemoth Cave [4] Interview (2020.11.20)

Fanstratics Newsletter #10 (June 2021)