[New experimental stuff]More challenging, BAI friendly battles - 13.07.2015 update

[rjacko10]

hmmm I tried restarting a game with Arveni but I don't seem to see the 45% disband effect.

Remember that the 45% disbanding only refers to the event of a lost battle. The victor has a much more timid rate.

[RollingWave]

Quote Originally Posted by Ritterlichvon86 View Post
Remember that the 45% disbanding only refers to the event of a lost battle. The victor has a much more timid rate.

Oh, well that seems ... odd, I mean most of the time if you lose your going to lose your whole army anyway. I would think it make more sense to have it be for the winner so there's much less of a steam rolling effect for the player.

[FlashHeart07]

I kindda fear that this will remove the effect of troops hiding meaning that many units will lose their advantage. Iberian troops with their guerilla deployment and awesome hiding skills will be useless if the AI is able to spot them from far away being able to counter . That being said I ofc hope that this is not the case and that the changes made by Krixux will show an AI being better at countering our flanking manouvers when they should be countered and not counter regular well set and sprung in battle ambushes

[rjacko10]

Iberian troops with their guerilla deployment and awesome hiding skills will be useless if the AI is able to spot them from far away being able to counter

Their might a special "hiding" variable that can be buffed to negate the increased spotting range..

[KAM 2150]

Could someone make a video or something about that spotting rage? Because on closer deployment mods where units stat pretty much next to eachother it did not change how AI uses his units eventhough all my units are visible to his.

Anyway, I won't be able to mod or even play Rome 2 until Wednesday

[Kurisu Paifuaa]

Could someone make a video or something about that spotting rage? Because on closer deployment mods where units stat pretty much next to eachother it did not change how AI uses his units eventhough all my units are visible to his.

Anyway, I won't be able to mod or even play Rome 2 until Wednesday

I've tried taking the minimum distance up near the maximum (1500) and didn't see too much difference, but need to test different arrangements and map setups.

In older TW games (going back to Empire) having the spotting distance too high would cause hyper-reactive behavior (premature cavalry charges, formation breaks etc) if I recall correctly.


For anyone editing these or other values across the whole table, please try the apply expression function available by right-clicking the column header in PFM. I see some comments in the thread that lead me to think some may be manually editing individual cells, which would be madness

[Butan]

Anyway the hiding is completely under-powered as is, especially against automatic-all-seeing AI (not so in multiplayer).

If we wanted to make hiding really important, we would need to buff hiding by decreasing spotting distance so that hidden units in forests/scrubs can be seen only if you're basically walking on their damned toes. + some League of Legends invisibility system and stealth would be more natural and impactful.


Because its not impactful today, I'm perfectly okay to nerf hiding capacity if it means improved AI combat.

[Krixux]

Guys !

I am changing and testing BLIND (!) here !

I really have NO clue what those values mean
I only suppose they affect "detection"

modifications in spoiler:

Spoiler Alert, click show to read: 

what I see however is that , in custom battles , flanking becomes increasingly difficult for me.
(I truly hope to have something here, and not only my imagination lol )

It seams that one of the values is about "priority"

- one of my Socii Equites although ordered to, and already engaged with an AI Socii Equites, it breaks regularly the attack and is going (continuously) for the AI general. !!!
maybe because "it has a priority for this" ????


I am still looking for this
in data_Rome2
there is ton of stuff there and little or (mostly) no info...
keep you all posted

here is the modification for Rome auxiliary:

Rome_land_units
with
visibility_spotting _range _minimum = 200
tree =200
scrub = 300

try a costom battle with this try mod (and DeI only)___dei_try_01.zip:


rome vs rome
each

4 socii hastatii
3 accensi
2 socii equites

???????
change values as you see fit...

edit

in
data_rome2
db
battle_entities_tables
battle_entities

found:
charge_distance_adopt_charge_pose
and:

charge_distance_pick_target <------- ( pick_target ) ?

priority or distance of picking target ?
might solve AI unit lingering "un aware " of my Cav just in their left?
...extracting and changing values
testing

entities changed to 50 (from 35 and 42 ) =
entities.zip

[Shaggy1973]

Guys !

I am changing and testing BLIND (!) here !

I really have NO clue what those values mean
I only suppose they affect "detection"

modifications in spoiler:

Spoiler Alert, click show to read: 

what I see however is that , in custom battles , flanking becomes increasingly difficult for me.
(I truly hope to have something here, and not only my imagination lol )

It seams that one of the values is about "priority"

- one of my Socii Equites although ordered to, and already engaged with an AI Socii Equites, it breaks regularly the attack and is going (continuously) for the AI general. !!!
maybe because "it has a priority for this" ????


I am still looking for this
in data_Rome2
there is ton of stuff there and little or (mostly) no info...
keep you all posted

here is the modification for Rome auxiliary:

Rome_land_units
with
visibility_spotting _range _minimum = 200
tree =200
scrub = 300

try a costom battle with this try mod (and DeI only)___dei_try_01.zip:


rome vs rome
each

4 socii hastatii
3 accensi
2 socii equites

???????
change values as you see fit...

edit

in
data_rome2
db
battle_entities_tables
battle_entities

found:
charge_distance_adopt_charge_pose
and:

charge_distance_pick_target <------- ( pick_target ) ?

priority or distance of picking target ?
might solve AI unit lingering "un aware " of my Cav just in their left?
...extracting and changing values
testing

entities changed to 50 (from 35 and 42 ) =
entities.zip

OK I gave this a test and have to say that the AI army does seem to act more coherently as a full group. There were cavalry skirmishes on the flanks. Missiles protected behind a heavy infantry line.

As my skirmishers began contact, their whole army reacted and engaged.

Looks pretty damn good.

Won a close victory on hard difficulty.

[rjacko10]

Because its not impactful today, I'm perfectly okay to nerf hiding capacity if it means improved AI combat.

Good point

[Ritterlichvon86]

Anyway the hiding is completely under-powered as is, especially against automatic-all-seeing AI (not so in multiplayer).

If we wanted to make hiding really important, we would need to buff hiding by decreasing spotting distance so that hidden units in forests/scrubs can be seen only if you're basically walking on their damned toes. + some League of Legends invisibility system and stealth would be more natural and impactful.


Because its not impactful today, I'm perfectly okay to nerf hiding capacity if it means improved AI combat.

Yeah makes sense

[zonks40]

More often than not I find the hiding a pain in the rear, marching a line towards the enemy and a quarter of my army hunker down in the scrub to hide while advancing, causing a broken line, even when men are running they attempt to hide, would have been much better to have a button that enabled "Attempt to hide"

I don't need to hide my men to win against the AI, if I just place 4 units forward and the rest back, the whole lot of them will converge on the 4 units, leaving the rest of my army to flank, I'd prefer a smarter AI than the ability to hide from them.

Having the AI being able to spot from much further may also prevent the AI changing formation just before you clash with their line.

[zonks40]

For anyone editing these or other values across the whole table, please try the apply expression function available by right-clicking the column header in PFM. I see some comments in the thread that lead me to think some may be manually editing individual cells, which would be madness

It doesn't seem any different to mass selection and paste, how is it used?

[Kurisu Paifuaa]

It doesn't seem any different to mass selection and paste, how is it used?

You can apply a mathematical expression instead of just a flat value... same as a spreadsheet function.

Let's say you wanted to decrease a value by 10% across the board, while maintaining relative differences between different units... you'd enter x = (x * -.10). Of course, you can still just enter a constant as well (same as copy/paste).

[zonks40]

You can apply a mathematical expression instead of just a flat value... same as a spreadsheet function.

Let's say you wanted to decrease a value by 10% across the board, while maintaining relative differences between different units... you'd enter x = (x * -.10). Of course, you can still just enter a constant as well (same as copy/paste).

Thanks for the help mate, will come in very handy

[FlashHeart07]

Im totally with you on that one Butan. But it stil kindda bugs me that we might have to come to this. Because what is the meaning of unit abilities like Snip, Guerilla deployment, Stealth etc if the AI is set to detect everyone of your troops from miles away. I for one have used units with these abilities to spring ambushes on the enemy on the battlefield especially when trying to destroy one enemy army before they get reinforced and greatly outnumber me.
Even though they have been far apart and few there have been battles where the AI sends troops to deal with my main line while sending spear wielding units of to deal with my cav trying to flank their main force.
Perhaps we will know more when this gets proper testing.

[Butan]

I love stealth/spotting system and abilities too but its really under exploited

The sniping ability which is overpowerful on the paper (firing and staying hidden), is almost useless except on ranged units which fires from a very long distance, because to fire you have to get close: except slingers and long range archers, you will almost never exploit snipe because of the detecting distance values.
Guerilla deployment and basic stealth is good for ambushes but only if you place them far away from the enemies: sometimes the time you use getting your ambush units into action is so long that you miss the surprise attack (and if you are too greedy you will get too close and bam detected). Not to add that there is not really "surprise" melee attacks because melee units are always detected way before they get into melee contact and the AI/human often have all the time in the world to about face a few units and counter (if every one of his units arent already fighting).
Its too hard to know when/where you will be hidden and when/where not relative to distance and spotting levels.


I think what works atm are ambush battles, because you can deploy your units close enough to fire/charge right after the battle begins.
The line of sight system also works marvel (when not fighting on open flat plains ) because it overrides the too far spotting distance value by forcing the game to hide units if they are on the other side of a cliff even if they are 10 meters away from you. One problem thought is how to "anticipate" line of sight radius when you move your units, sometimes its not intuitive and you get detected trying to place your units closer.



What would really bug me is for human vs human battles (especially with few/no battle interface help), because even under exploited hiding system is very fun there.
I would love to pimp the hiding system instead of making it further underwhelming

[MagicCuboid]

Okay, I've just run two test battles. Both used KAM's unit stats overhauls, and both were a Polybian army vs a Polybian army on a flat grassy plain. I used the same exact formations in both battle with a similar strategy. The only difference was the addition of Krixux's first test pack for the second battle.

Just so everyone has an idea of the armies, here is their roster:

2x Equites
1x Equites Extraordinarii
1x General
2x Velites
2x Accensi (I wanted to test how the AI handled long range)
2x Hastati (late)
2x Hastati Samnitici (late)
2x Principes (late)
2x Principes Samnitici (late)
2x Triarii (late)
2x Pedites Extraordinarii

1st Battle (without Krixux's mod)
1. Both armies start opposing each other in a similar formation, 3 ranks deep (thanks DeI!). The opposing army has 2 equites on my right flank, and I have 2 equites on my left, thus, a mirror image.

2. I advance my army to a safe distance (1.5x range of their accensi), then send my skirmishers forward to engage. I have my accensi focus target the other nearest accensi, while trying to stay out of range of the farther accensi.

3. The AI waits for my maneuvers dumbly, and doesn't move at all until the first stones strike.

4. After the first stones strike, the AI sends its entire front line and its right flank (4x Hastati, 2x Principes, 1x Triarii) after my one accensi. Presumably this is because the accensi is the only unit the AI is currently aware of despite my entire army being in plain sight.

5. The enemy's remaining units advance straight.

6. Half of AI's units continue to chase the fleeing Accensi, who leads them to my left flank. 5 of them "discover" my leftmost Hastati Samnitici and dogpile on it. The remaining 3 units are engaged by my leftmost Principes Samnitici who advance from the second line.

7. The dogpile leaves their backs completely exposed to the remaining units in my army. My principes and hastati unload pila on them and charge, eventually routing half of the AI army without any cause for maneuver. Their only saving grace is that their Triarii have considerable staying power, which buys them some time.

8. Meanwhile, on my right flank, the 2 equites and couple of Hastati/Principes have chased my other skirmishers into my waiting right units and die.

9. Battle ends as my victorious right flank is able to envelop the dumb AI; battle over. Decisive victory, ~15 units left.

2nd Battle (with Krixux's mod)
1. Both armies start in the exact same formations and positions as the 1st battle.
2. I advance my army to that same safe distance (1.5x range), and advance my skirmishers.

3. As soon as my army is near enough, the AI advances its whole army forward, maintaining formation. They largely ignore my skirmishers who are forced to flee, and they send their cavalry around both flanks of my army in a textbook maneuver. I maneuver my Velites as a javelin shield against the enemy equites and position my 2nd line Principes to meet them.4. As the frontline Hastati engage, the AI sends its 2nd line Principes around the sides as a flanking force to deal with my own counter-flanking Principes, forcing me to maneuver my Triarii to meet this challenge.

5. Meanwhile, on my left flank, the enemy AI's 2x equites against my 1, forcing them to flee. They then mop up my skirmishers and begin flanking maneuvers as described above, but this time against a weaker force.

6. My left flank begins to waver as they are faced with superior number. Conversely, my right flank is successful at repelling the AI flankers and begins a counter flank maneuver.

7. I divert my remaining 2 Equites to deal with enemy skirmishers once and for all, while the enemy Equites instead try to come around my rear.

8. My Pedites Extraordinarii block the enemy Equites and try to protect my general, who suffers casualties from the attack.

9. Free of the enemy, my skirmishers on my right flank begin to move out and harrass the engaged enemy AI, forcing a route. They then kill the enemy general with equites help, and focus on killing til they run out of javelins.

10. At this point, half of my left flank has routed, and half of the AI's left flank has also routed. This is precisely the kind of phenomenon balanced ancient armies experienced. Their most experienced troops would block the right flank due to concerns about their shield coverage, relying on less experienced troops to guard the easier left flank (in this case, my Samnites held their traditional honor of guarding the left). Occasionally, the less experienced left flanks would break in the face of superior fighters, despite their tactical advantage.

11. From here on out, it is a cat and mouse game to the death. At one point my line is split in half (due to KAM's less forgiving unit stats), and enemy pila murder my general! Luckily Romans are disciplined.

12. I eventually win the battle, largely due my superior ability to manage my units' fatigue, and the fortuitous preservation of my velites, who help significantly in the flanking maneuvers. Costly victory, only ~6 units left.


Conclusion
Krixux is really on to something here. Although the AI notably did not change tactics much once engaged, their initial maneuvering was far superior showing intelligent flanking maneuvers and seizing an early advantage. They simply lacked the endurance of a human player once their card had been played and all their units began to tire. Therefore, I recommend increasing the LOS in land_units_tables of all units alongside KAM's balancing pack for a superior battle experience.

[krunsh]

Conclusion
Krixux is really on to something here. Although the AI notably did not change tactics much once engaged, their initial maneuvering was far superior showing intelligent flanking maneuvers and seizing an early advantage. They simply lacked the endurance of a human player once their card had been played and all their units began to tire. Therefore, I recommend increasing the LOS of all units alongside KAM's balancing pack for a superior battle experience.

This sounds very promising. Keep the tests coming, they bring me great joy!

[FlashHeart07]

Nice Magic. Hopefully this mean that we are perhaps one step closer to a more challenging BAI. But when that is said I have had nearly the same experience with the BAI without Krixux mod. Fighting a macedonian army they send 3 units of hoplites out on my left flank to counter my Thessalian cav while they engaged my other units and even held some in reserve that (mainly because I wasnt paying attention) actually hit my sword units, who was doing a flanking manoveur on the enemy, in the rear causing them to rout.
Have anyone tried to see how easily hidden units are detected compared to how DeI is now?
Unfortunately I wont be able to test this myself. Someone forgot to tell me that you actually have to study at the University. And therefore I have to live off of your campaigns and battle test.