Store Page
Total War: WARHAMMER II
All Discussions Screenshots Artwork Broadcasts Videos Workshop News Guides Reviews

Total War: WARHAMMER II

Store Page
Total War: WARHAMMER II >  Workshop > Reithan's Workshop > Advanced Autoresolver > Discussions
Advanced Autoresolver
Description Discussions3 Comments227 Change Notes
 This topic has been pinned, so it's probably important
Reithan  [developer] 1 Jan, 2020 @ 12:20pm
Annotated Change Notes
Change Notes
  • Bugfixes
    • Fixed Doubled Missile Unit Modifier in Encampment Battles by Changing one into a Melee Modifier that was Missing
      Encampment battles currently give the defending army a small (7.5%) bonus to KPS, but this was double to give the bonus to missile units twice, unlike other bonuses like this that apply to missile and melee units. So this was updated to make one of the 2 bonuses apply to melee units instead.
  • General Improvements
    • Increased Unit Complexity to Consider up to 100 Units (from 10)
      The AR seems to have an in-built limit on how many units it 'considers' at one time, rather resolving units in small batches. Raising this limit from 10 units up to 100 (5 full 20-stack armies) seems to yield much nicer results overall. This DOES change the balance to less favor armies with a small set of elite units and tons of chaff.
    • Lowered Battle Timestep from 5 to 1 for More Granular Results
      The AR seems to resolve each 'group' of units in chunks of time based on this value. Testing here with a very large value led to each unit group completely mutually wiping each other out, so it seems the AR doesn't re-target units or check for death until after each timestep. Lowering this to 1 made results look much more 'smooth' across an army.
    • Removed Modifiers for Naval vs Non-Naval Units, as Naval Battles aren't REALLY a Thing
      There were some modifiers defined for naval units fighting non-naval units in naval battles. Hopefully these weren't doing anything at the moment, as naval battles are currently fought on land anyway, but we removed them just to be sure.
    • Increased Displayed AR Prediction Uncertainty, as AR is Always Somewhat RNG
      Since the result of the AR is never 100% certain, we added a small extra bit of 'fuzz' the balance of power bar so that it would never appear 100% locked in.
  • Protected High-Priority Units
    • Removed 'Ensured Unit Loss' Damage Multipler Bonus
      There was a modifier defined for 'ensure unit loss' set to 1.5x, which we assume the AR was using to sometimes trigger a unit to kill it's opponent in some situations. Removing this (set it to 1x) seems to have made the AR less likely to randomly destroy units in battle results.
    • Increased 'Ensured Unit Loss' Combat Potential Weighting
      The modifier described above also had a CP weight defined, which we increase, to hopefully steer it away from higher CP units more, just in case it's still having some effect.
    • Lowered Unit Wipe-Out Damage Threshold from 20% to 10%
      The AR would 'kill' any unit brought to <20% hp before. We changed this to only kill units dropped below 10% instead.
    • Added KPS Penalty Against Characters
      All units now take a KPS penalty when targeting a character (lords and heros) which, combined with their higher CP, does an excellent job of protecting them from dying in most AR battles, as well as allowing them to avoid getting 'pulled' from the battle as early in most cases, allowing squishier characters like wizards to actually get some work done in most battles.
    • Added KPS Penalty Against Artillery
      Basically the same change as was made for characters in the previous line.
    • Adjusted Combat Potential Modifiers Based on Units' Existing Base CP and Player Tendencies w.r.t. Protecting Units
      Original CP modifiers were applied in very broad strokes: +1000 CP to all units marked "High Value", small multiplier negatives to certain frontline units. We went through the base values in land_units, analyzed which units were getting the best consideration, and whether that lined up with how a competent player would likely want to use and protect that unit, and rebalanced the modifiers to better put final CP value in a good array for more units.
  • Adjusted Chariot Effectiveness vs Infantry
      Added KPS tweaks for chariots to make them more durable and potent in AR battles against infantry-heavy armies as they weren't really living up to their potention in the AR.
    • Added Melee KPS Bonus for Chariot vs Infantry[REMOVED]
    • Added Melee KPS Penalty to Infantry vs Chariots
  • Adjusted Cavalry Effectiveness vs Infantry
      Mirrored KPS tweaks for chariots to cavalry to maintain their potency in AR battles against infantry-heavy armies without racking up crazy kill counts in strange scenarios.
    • Removed Melee KPS Bonus for Cavalry against all Targets [NEW]
    • Added Melee KPS Penalty to Infantry vs Cavalry [NEW]
Last edited by Reithan; 3 Jan, 2020 @ 12:16pm
< >
Showing 1-6 of 6 comments
Reithan  [developer] 2 Jan, 2020 @ 11:27pm 
Test results that let to the Jan 2nd patch:

https://imgur.com/a/R7ZGDRm
#1
Reithan  [developer] 2 Jan, 2020 @ 11:28pm 
So, in investigating this, it turns out all the lizardmen monstrous infantry and dinosaurs, for the AR kps calculations, are classified as melee infantry.

So, trying to adjust these chariots against them, while also making them as effective as they should be in ACTUALLY chariot vs infantry scenarios is a little...well...kind of impossible.

I'm working on tweaking the numbers to as good a middle-ground as I can right now, but it turns out, even with 0 changes to the AR, that Malus vs Lizardmen scenario results in those scourgerunner chariots pulling like 250+ kills.

Without this mod, melee cav has a +50% bonus to KPS.

Swapping all the bonuses (original and modded) to offensive KPS for cav and chariots to defensive bonuses for them produces a bit more reasonable result. This also makes battles involving cavalry contingents a bit less blood-bath-y

The update pushed out should keep from making chariots and cavalry any WORSE than it is normally, while improving its durability against melee infantry in the more problematic cases.

Unfortunately, there's a lot of mis-classed units for some armies that cause really bad AR results in cases involving cavalry and chariots. Not much we can do about that without a full overhaul mod.
#2
V 1 Jul, 2023 @ 5:10am 
In your large amounts of testing, no doubt you have came across these two variables.

• autoresolver_alliance_wipe_out_loser_strength_of_max_threshold

Default value 0.3

• autoresolver_wipe_out_army_force_unary_lost

Default value 0.92

Do you guys happen to know the meaning of each, and the difference between them?
#3
Reithan  [developer] 1 Jul, 2023 @ 5:14pm 
IIRC (which I may not), the first one is the % the loser needs to be under to be wiped out in cases where they might not be. So, if the enemy army could retreat, but end up under 30% strength at the end of autoresolve, they're wiped out instead.

The second one, IIRC (again, I may not), that's the per-unit wipe out. So if a unit takes 92% of more losses, that unit is wiped out.
#4
V 3 Jul, 2023 @ 3:42am 
Ok after some testing, this is the right conclusion I believe. This is all for AUTO-RESOLVE.

• autoresolver_alliance_wipe_out_loser_strength_of_max_threshold (default value 0.3)

If the entire losing army's strength is under 30% of its original pre-battle strength, it will get wiped out. So if you want entire losing armies to be wiped out less, lower this value.

• autoresolver_unit_wipe_out_threshold (default value 0.2)

For a multi-model unit, if less than 20% of its models survive, the unit is wiped out. So if you want units (from both winners and losers) to be wiped out less, lower this value.

• autoresolver_wipe_out_army_force_unary_lost (default value 0.92)

For a single-model unit, like heroes/lords/dragons, if more than 92% of its HP is gone, it will be wiped out. So if you want single important units to be wiped out less, increase this value.
Last edited by V; 3 Jul, 2023 @ 3:44am
#5
Reithan  [developer] 3 Jul, 2023 @ 4:48am 
Sick. Thank you for testing these! :D
#6
< >
Showing 1-6 of 6 comments
Per page: 1530 50