商店页面
全面战争：幕府将军2
全部 讨论 截图 艺术作品 实况直播 视频 创意工坊 新闻 指南 评测

全面战争：幕府将军2

商店页面
Total War: SHOGUN 2 >  创意工坊 > WalrusJones 的创意工坊 > UAI Revived > 讨论
UAI Revived
描述 讨论1 留言70 改动说明
WalrusJones  [开发者] 2014 年 10 月 9 日 下午 4:42
AI Issue Reporting.
This is just a centralized place to put complaints regarding any issues one sees with the AI, generally, things it does poorly.

As an example, I will use the vanilla AI:
- Invests very little in economic growth.
- Does not specialize settlements.
- Does not centralize unit production, making it difficult to place good modifiers on units.
- Uses Yari samurai as its primary means of anti-infantry.
- Does not use seige weapons.

Now, my one major issue, and a minor one with the current generation AI in the mod:
- Attacks forts with insufficient force.
- Slow to develop core areas of its strategy: As an example, second level military buildings seldom appear before turn 20.

Things such as these are things I wish to minimize, to further optimize an otherwise great AI.
最后由 WalrusJones 编辑于; 2014 年 10 月 12 日 下午 2:36
< >
正在显示第 1 - 7 条，共 7 条留言
Destin Faroda 2014 年 10 月 12 日 下午 2:33 
引用自 MagicalSmallrusWalrusJones：
- Does not use seige weapons.
Have a look at cdir unit qualities table.

引用自 MagicalSmallrusWalrusJones：
- Attacks forts with insufficient force.
Try campaign variables and campaign ai behaviour junctions.
This one sounds especially interesting: MINIMUM_CHANCE_OF_SUCCESS
I've increased it from 55 to 60. I presume it tells the AI to attack when the chance of a victory is equal to or higher than the set value.
#1
WalrusJones  [开发者] 2014 年 10 月 12 日 下午 2:36 
I gave an example list using the Vanilla AI, albiet, the second suggestion is very much valid, thank you.

Looks like UAI lacks a value to override minimum chance of success. I am a little sad that it doesn't look like there is a seige specific variable for minimum chance of success.
最后由 WalrusJones 编辑于; 2014 年 10 月 12 日 下午 2:38
#2
nobubblegums 2020 年 8 月 6 日 上午 2:20 
Hi, thanks for keeping this alive. I know you're not working on this anymore, but maybe other people still playing can chime in and we can work towards some improvements. My two issues so far are:
AI never tries to starve the defenders in a siege and attacks instantly (more or less what you said)
AI is prone to launch overseas invasions against the player at the cost of leaving its home provinces undefended. At the same time, they are reluctant to do this against other AI factions. Ex: Playing as Shimazu, I get factions invading me from other side of the map, while Chokasabe sits with a few full stacks in their island, not attempting to expand. Aggression against me is probably a cause of difficulty though (VH).

#3
nobubblegums 2020 年 8 月 30 日 下午 2:42 
So I think I found a possible typo while going through the mod in PFM. In campaign ai personality junctions,the entry
" INVASION_OBTAINING_LOCAL_MOBILES_ARMY_STRENGTH_MULTIPLIER_FOR_BALANCE_TYPE_DEFAULT " has the value of 9. The original value is 0.8 and the other similar entries (invasion_obtaining_x) have values between 0.8 - 1.3. Also, the same value for navy (INVASION_OBTAINING_LOCAL_MOBILES_NAVY_STRENGTH_MULTIPLIER_FOR_BALANCE_TYPE_DEFAULT) has the value of 0.9. None of the other changes made by UAI are that large, so I think that could have been 0.9. Not sure what this entry does, but I'll try to reach the hedge night on TWC to see whether this was intentional.
#4
Bubbleezzz 2020 年 12 月 17 日 上午 2:48 
引用自 nobubblegums：
So I think I found a possible typo while going through the mod in PFM. In campaign ai personality junctions,the entry
" INVASION_OBTAINING_LOCAL_MOBILES_ARMY_STRENGTH_MULTIPLIER_FOR_BALANCE_TYPE_DEFAULT " has the value of 9. The original value is 0.8 and the other similar entries (invasion_obtaining_x) have values between 0.8 - 1.3. Also, the same value for navy (INVASION_OBTAINING_LOCAL_MOBILES_NAVY_STRENGTH_MULTIPLIER_FOR_BALANCE_TYPE_DEFAULT) has the value of 0.9. None of the other changes made by UAI are that large, so I think that could have been 0.9. Not sure what this entry does, but I'll try to reach the hedge night on TWC to see whether this was intentional.

If you looked it up, did you find out if this is intentional or not?
#5
nobubblegums 2020 年 12 月 17 日 上午 8:01 
I haven't heard from him, but I'm pretty sure that was a typo. I corrected it in my version
#6
WalrusJones  [开发者] 2020 年 12 月 17 日 上午 8:55 
It might not be, what I understood about how the original UAI (Which I just made work with the workshop) was made, a lot of these changes are made doing case studies with the AI and gradually tweaking the values until the actually makes the right decision in a specific case, then playing a campaign to make sure that the AI's decisionmaking isn't ruined by making the AI make one right decision.

I can't say for sure whether that is a typo or not as a certain thing, but it probably is.

I can try and see if the game launcher stops denying me the ability to touch any old workshop file.

This may actually be related to one of the key issues I saw years back.
最后由 WalrusJones 编辑于; 2020 年 12 月 17 日 上午 8:55
#7
< >
正在显示第 1 - 7 条，共 7 条留言
每页显示数： 1530 50