Wildmana won't be a modmod of FFH in the future. To prevent too much confusion there will be a namechange to Master of Mana, a homage to the classic fantasy strategy game Master of Magic. Given that lots of the great game mechanics like a Global Enchantments, Mana Node Guardians, Spell Research or a global Mana pool are also implemented in Master of Mana, it might embody some of the glory from Master of Magic (and ofcourse the many game mechanics implemented in Civ4 or FFH help too).
The Release date is 15. December 2010. I'll try to create a short manual in the next days, that gives an overview about the new game mechanics.
I could write days about the new game features, I will just focus on the feature that from my point of view will make the biggest difference in the long run: a completly rewritten AI. The AI is vastly better and I will explain below why.
When I started modding the AI in FFH it was in a really bad shape. Mostly because it had little to no idea of the features introduced by FFH (like magic, non linear tech tree, lots of mechanics written purely in python, etc.). The first step was to get the economy of the AI right, which mostly involved tech selection and building selection. The functions in Civ4BTS didn't really work well for the AI, so I wrote some new functions the quick and dirty way because basically anything was better. There were a couple other AI mods who used this approach, so gladly AI improvements could be shared fairly easy and progress was fast. The much more difficult thing was to teach the AI to use it's units a bit more intelligent. I tried to add in some tools to help the AI coordinate its units better. This was very important for the AI to use magic.
However success here was quite limited. In the case of magic one huge barrier was that magic was implemented very AI unfriendly. It used a lot of python code for example and that is something the AI has problems to understand. So for the AI in Master of Mana I changed some of the underlying mechanics to make it easier for the AI to understand. There are typically two approaches. The first one is to allow the AI to calculate the use of something. And the AI in Master of Mana has some really impressive abilities. It can calculate the actual value of any building. So if the building gives 2 gold, but the city has gold modifier +100%, it would understand the building in this city is twice as good than in a city without the modifier. While this sounds pretty trivial it is something the AI in Civ4 can only dream of. Same for terraforming rituals. The AI calculates how much food/production/etc. it will gain from the terrain changes and uses that to decide, if it uses the ritual or not. For the tech trading the AI mostly uses some AI flavors. The Hippus for example have some mounted flavor, so they will sooner or later research all techs that are tagged as mounted. If they run short on cash or research time is long, they will research techs that are flagged as gold or science. The important thing here was to streamline the techtree so that the flavor formula works better.
The real heart of the new AI however is the new AI for units. I recently found this article which gives a good idea about the problems the AI has to coordinate its units.
http://www.gamedev.net/reference/articles/article545.asp
Using the descriptions in the article the BTS AI probably operates on a level between 4 and 5. Most units act on their own, but sometimes they coordinate their actions (stack of doom, naval transports). The Wildmana AI is a good deal closer to 5 as the AI uses stacks to coordinate units a lot more often. The new Master of Mana AI is somewhere between 6 and 7. It creates little projects like "Take City A" and then adds units to those tasks as needed "Some normal units, a few arcane units, a few siege units" and if can't find the necessary units it will order cities to produce these units. This approach allows for so much more coordination between units. This video shows a good example what difference coordination between units can make
http://www.youtube.com/watch?v=WVaDD_ydHFQ
It's basically a fight between two unit squads in Starcraft:Broodwar. The key to success here is not to value actions for each individual unit and then pick the best, but to use a strategy for the whole squad (like focus fire on one target).
Everything comes with a cost and the more complex AI needs quite a bit time to get finetuned. So it will probably take month or years to use most of the potential the new AI logic has to offer. But from my expierence the new AI is way more fun to play against, because Stack of Doom is not the only tactic it can use.
The Release date is 15. December 2010. I'll try to create a short manual in the next days, that gives an overview about the new game mechanics.
I could write days about the new game features, I will just focus on the feature that from my point of view will make the biggest difference in the long run: a completly rewritten AI. The AI is vastly better and I will explain below why.
When I started modding the AI in FFH it was in a really bad shape. Mostly because it had little to no idea of the features introduced by FFH (like magic, non linear tech tree, lots of mechanics written purely in python, etc.). The first step was to get the economy of the AI right, which mostly involved tech selection and building selection. The functions in Civ4BTS didn't really work well for the AI, so I wrote some new functions the quick and dirty way because basically anything was better. There were a couple other AI mods who used this approach, so gladly AI improvements could be shared fairly easy and progress was fast. The much more difficult thing was to teach the AI to use it's units a bit more intelligent. I tried to add in some tools to help the AI coordinate its units better. This was very important for the AI to use magic.
However success here was quite limited. In the case of magic one huge barrier was that magic was implemented very AI unfriendly. It used a lot of python code for example and that is something the AI has problems to understand. So for the AI in Master of Mana I changed some of the underlying mechanics to make it easier for the AI to understand. There are typically two approaches. The first one is to allow the AI to calculate the use of something. And the AI in Master of Mana has some really impressive abilities. It can calculate the actual value of any building. So if the building gives 2 gold, but the city has gold modifier +100%, it would understand the building in this city is twice as good than in a city without the modifier. While this sounds pretty trivial it is something the AI in Civ4 can only dream of. Same for terraforming rituals. The AI calculates how much food/production/etc. it will gain from the terrain changes and uses that to decide, if it uses the ritual or not. For the tech trading the AI mostly uses some AI flavors. The Hippus for example have some mounted flavor, so they will sooner or later research all techs that are tagged as mounted. If they run short on cash or research time is long, they will research techs that are flagged as gold or science. The important thing here was to streamline the techtree so that the flavor formula works better.
The real heart of the new AI however is the new AI for units. I recently found this article which gives a good idea about the problems the AI has to coordinate its units.
http://www.gamedev.net/reference/articles/article545.asp
Using the descriptions in the article the BTS AI probably operates on a level between 4 and 5. Most units act on their own, but sometimes they coordinate their actions (stack of doom, naval transports). The Wildmana AI is a good deal closer to 5 as the AI uses stacks to coordinate units a lot more often. The new Master of Mana AI is somewhere between 6 and 7. It creates little projects like "Take City A" and then adds units to those tasks as needed "Some normal units, a few arcane units, a few siege units" and if can't find the necessary units it will order cities to produce these units. This approach allows for so much more coordination between units. This video shows a good example what difference coordination between units can make
http://www.youtube.com/watch?v=WVaDD_ydHFQ
It's basically a fight between two unit squads in Starcraft:Broodwar. The key to success here is not to value actions for each individual unit and then pick the best, but to use a strategy for the whole squad (like focus fire on one target).
Everything comes with a cost and the more complex AI needs quite a bit time to get finetuned. So it will probably take month or years to use most of the potential the new AI logic has to offer. But from my expierence the new AI is way more fun to play against, because Stack of Doom is not the only tactic it can use.