Open Beta 0.72 for new difficulty level

[Nick Thomadis]

We deployed a special test branch with a few AI experimental changes + a new special difficulty mode for those who desire extreme challenge.

To access this beta branch: 
- Right click "Ultimate General: Civil War" in steam library and select "Properties"
- Go to Tab "Betas"
- Copy/Paste this code "ugcwPublicBeta" to the empty space and press "Check Code" button
- Your code should be confirmed and then you can switch to the beta mode from the list "Select the beta you would like to opt into".

Note:
The beta branch is not final so we need feedback to improve it or fix any bugs that you may find.

[Nox165]

awesome, cant wait to give this a try.

[anibal barca]

Does this new level of difficulty have bonuses for AI?,If so, I'm not interested,because an AI with bonuses is not a good AI.

also breaks the realism and historicity,see units of militia or recruits to win elite or veteran units is quite disappointing,because the moral levels and the stats of abilities of the units are with bonuses.

the AI must be without bonuses and prove that it is a challenge for the players.

[Salty Sails]

Its to difficult for me and my playstyle, i think!

I won the first battle with CSA. Captured fort and defend it against the union! No problems!

But after that... NEWPORT was a disaster for me... i lost my 2 brigades... they get captured by union forces because to much union brigades attack the town at same time... enemy fire and melee...no chance to hold it!

reinforcments could not help....

both cav. brigades are useless in this fight!

So... if i play a game,  i will have fun...not getting frustrated because its to difficult!

Sorry!

 

Edited December 27, 2016 by Sepp Stuart

[RobWheat61]

6 hours ago, anibal barca said:

the AI must be without bonuses and prove that it is a challenge for the players.

Would be a nice change, if players would adhere to this too, instead of beating the AI by gaming the mechanics and AI weaknesses.

6 hours ago, Sepp Stuart said:

Its to difficult for me and my playstyle, i think!

I won the first battle with CSA. Captured fort and defend it against the union! No problems!

But after that... NEWPORT was a disaster for me... i lost my 2 brigades... they get captured by union forces because to much union brigades attack the town at same time... enemy fire and melee...no chance to hold it!

reinforcments could not help....

both cav. brigades are useless in this fight!

So... if i play a game,  i will have fun...not getting frustrated because its to difficult!

Sorry!

haha, I can relate! The all-out attack on the town is pretty nasty :)! But you have always the option to withdraw from a battle, which would be prudent with overwhelming odds ;).

[Koro]

7 hours ago, anibal barca said:

Does this new level of difficulty have bonuses for AI?,If so, I'm not interested,because an AI with bonuses is not a good AI.

also breaks the realism and historicity,see units of militia or recruits to win elite or veteran units is quite disappointing,because the moral levels and the stats of abilities of the units are with bonuses.

the AI must be without bonuses and prove that it is a challenge for the players.

It doesn't have fake bonuses like total war, but it has more men and more veterancy than hard. This option for sure isn't for everyone but for those who can easily beat hard it is an extra challenge for them. 

[Aetius]

I did a couple of runs through 1st Bull Run on Normal (BG) difficulty to gauge the AI changes. Overall, I like it. The AI feels stronger in some respects. It's better at attacking, for sure. On my second game of Bull Run, the AI bypassed Henry Hill entirely and drove into the rear of my bridge position, overrunning it. At the same time, he threw a brigade across the northern ford, and two brigades across the bridge. It came fast, and one of my brigades was caught, routed, bounced back and forth a couple of times, and then surrendered. In other words, the weirdness with units routing through your lines seems to be fixed.

The AI also now deals correctly with units in the little woods north of the bridge. Instead of ignoring them and soaking up damage, both times the AI maneuvered a unit to attack them from the north. This seems to represent a more sensible way of dealing with units that can't be seen, but whose location can be inferred from their fire. This is a very, very good improvement, although it may result in problems down the line against sniper units - the AI might be tempted to send out units to attack them, and fall into a trap.

My notes:

1) This is a big one - lethality is increased. Playing essentially the same strategies I did before on Normal difficulty, everyone took roughly 50% more casualties. The two Bull Run games were 12k to 6k, and 11k to 6k. In the first game, I lost two units entirely - my baby snipers and my artillery. In the second game, I lost one infantry brigade entirely and took took nearly 2/3rds casualties on the other. The snipers made it out with just over 50% casualties, and my artillery with two guns (out of six). The two earlier scenarios, Potomac Fort and Newport News, had similar levels of casualties, so much so that I couldn't get Mississippis for both units.

I believe that this is because the changes made units more willing to move up close, stand and fight, and take more casualties before breaking. This works up to a point, and makes the AI a better attacker. Unfortunately, it also drains him of troops and units. In both Bull Run games, about 2/3rds of the way through the game (about when Johnston shows up) the Union was completely shattered and scattering across the map in an attempt to escape. (This behavior, by the way, is much better than before - it makes the kill box much harder to construct and close, which is good.)

What I suspect is going to happen is that in later scenarios with larger units and better weapons, attacks are just going to dissolve in a bloodbath. Standing and fighting in front of a couple of one-star 1500 man brigades with 1842s and Mississippis is one thing - doing the same to a couple of three star brigades with 1855s or Enfields, backed by 24lb howitzers, is going to result in immediate and massive casualties for the attacker.

What the AI needs is a better, more coordinated rush to melee - particularly for the Union, because they have superior numbers but inferior troops. And to be fair, the AI did better in these Bull Run scenarios than I've seen before - the coordinated attack on the bridge area was evidence of that.

2a) I saw several instances of serious cavalry stupidity. In my second game of Bull Run, the cavalry that tries to cross the ford north of the bridge made repeated frontal approaches against my skirmishers until the cavalry was destroyed. I think the AI, for some reason, didn't track the last known location of the skirmishers, and thus kept trying to path right by them instead of going around further north. I also saw a cavalry unit repeatedly try to rush Henry Hill right in the center - it hit the water and got blasted from ~200 to ~75, and then wiped out.

2b) I also saw two possible cavalry bugs. Stuart at Newport News got pushed over against the right side of the map, and when he routed, the unit froze in place. I suspect the routing logic was trying to go further right, but the map edge wouldn't let it, so it just sat there. This happened twice - he routed, froze, recovered, started to head south, got hit again, routed, froze, and then was destroyed.

The second possible bug was on Potomac Fort, during the initial attack. I charged the left skirmishers with my cavalry. They routed, and the cavalry started running them down. In the course of that, the cavalry actually got ahead of the skirmishers, and it looked like the skirmishers were chasing the cavalry. This apparently confused the AI, because the cavalry then gave up, and wandered off, leaving the routed skirmishers still alive to turn around and shoot them a few seconds later.

3) Also on Potomac Fort, I was forced to manually turn a couple of units in order to get them to fire on their target. Not sure what that was about.

4) The unit AI apparently got a dose of merpeople genes in this update. Everyone loves the water. Skirmishers falling back? They'll stop smack in the middle of the nearest creek, and sit there. Unit got routed? They'll run half a mile to the nearest river, and stop for a bath right in the middle of it. Unit maneuvering to attack? If being in the water is an option, they will choose it. Units attacking across a river will inevitably stop right in the middle to exchange fire with the defenders, instead of pushing across into open ground or cover. General needing to recover a routing unit? He'll do so from the river. (This is how McDowell died in both my Bull Run games.)

I suspect this is occurring because to the AI, the water is just a "slow spot" that consumes the equivalent of movement points. This means that as units are moving, and especially running away, the slow spots tend to snag them. The AI really needs to have a aversion to the water - it shouldn't stay in it unless specifically ordered to do so, and spend as little time as possible there when crossing, even if it means not shooting.

5) Still have to manually force skirmisher / sniper units to fire about half the time, even when it's the same target in the same position.

6) Unit AI appears to handle being out-ranged correctly about half the time. My Mississippi-equipped brigade was able to kill several hundred Yankees at the bridge without a single shot getting fired in return, most of them on a couple of units that just stood there and took it without responding. Other units would quickly withdraw about 50 yards back to avoid the fire, and artillery backed out of view to avoid counter-battery fire.

7) The increasing amount of firing puts an increased drain on logistics. My wagon had 6700 supply on it for the Bull Runs, and was cleaned out about halfway through both scenarios - even though in both cases what little artillery I had was destroyed or severely damaged, and in one case I had lost nearly 2/3rds of my personal troops. The attacking Union artillery at the bridge ran out of ammunition before Beauregard arrived at Matthew's Hill (and, as a likely bug, I knew this because I was able to see the red AMMO warning on enemy artillery). In Potomac Fort, the continuous fire of the two 4-gun batteries drained the 1000 supply in the fort in less than five minutes - it essentially required 1000-1500 supply to reduce the two ironclads and fire some canister at a couple charges.

8) The crazy wandering of the AI supply wagons seems better. Stuart almost caught one of them during my second Bull Run game, but it turned out to be too close to a skirmisher unit and he got a face full of bullets instead. We'll see how this goes in some of the known trouble scenarios, like Gaines' Mill.

Edited December 28, 2016 by Aetius

[Grognard_JC]

Thank you for this beta. I really will give it a go (but I'm really busy these days -Christmas !). Is it supposed to be plyaed on Normal or Hard ?

 

- Edit : I found the solution. Whatever !

Edited December 28, 2016 by Grognard_JC

[Slaithium]

I. I think there is a general agreement here that artificially buffing the AI is awful.
 Ia) using more troops and veterancy is the samething, just spliting hairs here, don't do it. Ib) anything that is given to AI that is not earned is not a increase in difficulty, that is called over coming odds. By artifically increasing numbers, vet, supplies, etc etc you not creating difficulty of a opponent you are creating a larger obstacle. Two distinctly different things.

II. The point and the best way to increase difficulty that is legit is making the player and the AI both earn what they have. IF you give AI or even a player something it is not worth anything and starts to loose any meaning.

III. Solution, make the AI increasing intelligent on each level increase of difficulty. Giving the AI bonuses of any type without earning them is poor game design. Also could arguably ruin any and all historical narrative that is in the game if it is blown too far out of proportion.

[Koro]

1 minute ago, Slaithium said:

I. I think there is a general agreement here that artificially buffing the AI is awful.
 Ia) using more troops and veterancy is the samething, just spliting hairs here, don't do it. Ib) anything that is given to AI that is not earned is not a increase in difficulty, that is called over coming odds. By artifically increasing numbers, vet, supplies, etc etc you not creating difficulty of a opponent you are creating a larger obstacle. Two distinctly different things.

II. The point and the best way to increase difficulty that is legit is making the player and the AI both earn what they have. IF you give AI or even a player something it is not worth anything and starts to loose any meaning.

III. Solution, make the AI increasing intelligent on each level increase of difficulty. Giving the AI bonuses of any type without earning them is poor game design. Also could arguably ruin any and all historical narrative that is in the game if it is blown too far out of proportion.

Don't forget that it's for legendary mode only and for those who want to be challenged this way. The other difficulties remain the same. 

The AI is the same for all levels. 

[Slaithium]

Still even if that is just for legendary mode, the points still stand no matter the lowest or maximum difficulty that can be had in the game.

[Koro]

1 minute ago, Slaithium said:

Still even if that is just for legendary mode, the points still stand no matter the lowest or maximum difficulty that can be had in the game.

Maybe but it's the same in almost every strategy game. Total war AI simply gets bonus stats and stack spam, CkIi the AI gets more money and army spam. 

Across the board it's like that. 

I think it's good that easy doesn't face a dumbed down version of the AI but just less numbers to make it more manageable. 

[Slaithium]

4 minutes ago, Koro said:

Maybe but it's the same in almost every strategy game. Total war AI simply gets bonus stats and stack spam, CkIi the AI gets more money and army spam. 

Across the board it's like that. 

I think it's good that easy doesn't face a dumbed down version of the AI but just less numbers to make it more manageable. 

So lets answer the simple question then, does Game Labs want to be like others, or do they want to be different. To do the same as others but slap it in a different era is not unique. If they want to keep doing the same as others, do we need to expect DLC which for many has ruined games arguably. ( which total war is notorious for, from you example). Is there going to be perpetual imbalance? (like company of heroes). When you buff things by numbers and not intelligence the game is going to be imbalanced in some way at every difficulty. Age of empires tried to just hyper inflating numbers like this and almost destroyed the franchise because it pissed off there player base. 

Also, how the current design of the game is set, it is meant as a historical game, a game that allows you to relive and re-write the history of the game as it goes through the progression. If you do what is offered, it will undermine the foundation of the game.

 

[Koro]

The normal version of the game gives you this. The historical setting and In theory at least, an AI that plays fair by numbers and veterancy. The goal, from what they've said, is to make the AI as good as possible against human players. 

The difficulty settings then exists for those who either can't compete with the AI and for those for whom an Ai will never truly challenge without bonuses, which won't happen for many years I suspect. Not until you make a learning AI anyway at which point we might all have to play on easy. 

No,  I don't want an endless stream of dlc, and I certainly don't want them to go total war style on us. 

[Slaithium]

increasing intelligence is not all that hard, I myself enjoy a good challenge, lets take Malvern Hill for example ( I think) the one you attack where the union is heavily fortified. Instead of increasing the numbers and the veterancy. Give the union more entrenchments, I can theorize a union general giving that order and them digging deeper in. Which is not a far stretch of a pre-lude to the Western Front of World War I. Which actually could introduce a new mechanic for the player could utilize, allow brigades to entrench, of course include a long set up time and unable to fire and take increase damage if caught. But things like that are in todays technology. That is just one example of what I mean.

[Aetius]

Updates to my previous post from play last night. I dropped back to 0.71 and did a new Normal difficulty Confederate run through to the end of First Bull Run, then reloaded 0.72 and did the same again for comparison. For my general I chose Logistician / Artillery / Politics, and put a point in Politics after each battle.

Notes:

1) 0.72 still feels better, terms of pressure from the AI during attacks. However, I was able to adjust my tactics a bit to compensate for the AI's increased willingness to roll up and exchange fire, which resulted in my casualty levels remaining about the same and sharply increasing for him (0.71 - 13k to 5k, 0.72 - 16k to 5k). My casualties also shifted from my personal units to the units I was given for the battle (in particular, Hampton's Legion got hammered pretty hard). The increased AI casualty rate meant that my 0.72 First Bull Run game ended with Johnston chasing down the last few Union brigades in the upper right corner of the map. The old AI was reluctant to get into it, but the new AI seems unable to close - he rarely charged, and apparently prefers to stand in the river and shoot.

2) The AI pushing a unit across just to the north of the bridge happened on my 0.71 run, as well as my 0.72 run, so that's eliminated as a possible change for 0.72.

3) I noticed that instead of starting 2500 supply, in 0.72 I started with 0 supply. This may be why I felt that supply was more of an issue in 0.72. However, during my 0.72 run I used up my wagon (~5k supply), a captured wagon (~2k supply I think), and half of Johnston's wagon ... with 8 guns, one 500-man sniper unit (Hunters), and two Mississippi-armed infantry brigades.

4) My ability to see AMMO warnings on enemy artillery was there in 0.71 as well, so appears to be an existing bug, rather than something added in 0.72.

5) I need to run some more tests, but I think fortifications may not be working, or aren't working as intended. In my 0.71 run I didn't use any fortified positions except for the artillery positions in Potomac Fort. In my 0.72 run, I did put a brigade in the fortifications in front of the bridge in First Bull Run. In the 0.71 run, the unit that was in the same position was strong enough to counter-attack across the bridge and destroy some artillery; the 0.72 unit was hurt significantly worse by the end of the battle, despite being in the fortification; however, he did endure a significant melee attack, so it's hard to tell. It seemed to me anecdotally that the unit's firepower was reduced, and this resulted in longer exposure to fire from attacking units.

6) In my 0.72 First Bull Run game, I had two Union infantry units across the bridge just ... stop. I sent my snipers across the river and killed the artillery behind them, and they didn't respond at all. I then sent my snipers in to shoot at them. The first unit routed, came back to the same position, routed again, came back again, and then actually shattered under the sniper fire. The other unit routed once, and then began maneuvering and fighting again.

7) I filed a bug on an issue with the Battle Map screen - when you first look at the Battle Map, the dollar values for the battles are calculated at that time, and cached. If you add a point of Politics and then go back to the Battle Map screen, the values are the same as before. If, however, you add the point of Politics first, the values that appear when you first view the Battle Map screen are increased.

8) The foolish cavalry behavior DID occur in 0.71 - two cavalry units suicided themselves on a 200-man skirmisher unit at the ford just north of the bridge in First Bull Run, so that behavior is not new to 0.72. The behavior happens like this: cavalry unit approaches a hidden unit. When they spot the unit, they stop. The unit shoots them, and then they run away. Then they return to the same spot, and the pattern repeats. Sometimes the cavalry will get a shot off, so I think they may be trying to shoot but it takes longer than it takes the skirmishers to fire.

9) In my 0.71 run, I had two enemy units rout through my lines, one on Potomac Fort and one in First Bull Run. In my 0.72 run, it did not happen, even during a large swirling melee on my side of the bridge in First Bull Run. Yay!

10) In my 0.72 run, McDowell died again standing in the river in front of Henry Hill.

Edited December 29, 2016 by Aetius

[Grognard_JC]

@Slaithium :

You are asking for a "better" AI because you do not like the idea of having challenge come from cheer "bonuses" given to the AI's troops. Fine.

 

But, for real, you cannot blame Darth for this, or maybe you do not know what he has achieved, even before he started working for Game Labs. He is a king of AI.

 

Most games have crap AIs. AIs are scripts mate. The more scripts you put in, the more weird the AI behaves. You have to be a very patient Magician to find an AI that works. And NO, it's not "easy" to create a "better" AI. It would cost tones of money of developpement that even SEGA is not willing to pay !!!

 

The best AIs I know on the video game markets are in CK2, EU4, Naval Action and UGCW. No wonder these are niche games.

 

If you want to ask the devs something they can ACTUALLY make real, so that you get a real challenge from a good "AI", just ask them to put the Multiplayer mode in. This way, the game will be much more challenging, without cheats.

 

[Slaithium]

10 hours ago, Grognard_JC said:

@Slaithium :

You are asking for a "better" AI because you do not like the idea of having challenge come from cheer "bonuses" given to the AI's troops. Fine.

 

But, for real, you cannot blame Darth for this, or maybe you do not know what he has achieved, even before he started working for Game Labs. He is a king of AI.

 

Most games have crap AIs. AIs are scripts mate. The more scripts you put in, the more weird the AI behaves. You have to be a very patient Magician to find an AI that works. And NO, it's not "easy" to create a "better" AI. It would cost tones of money of developpement that even SEGA is not willing to pay !!!

 

The best AIs I know on the video game markets are in CK2, EU4, Naval Action and UGCW. No wonder these are niche games.

 

If you want to ask the devs something they can ACTUALLY make real, so that you get a real challenge from a good "AI", just ask them to put the Multiplayer mode in. This way, the game will be much more challenging, without cheats.

 

You mis-understand me, I am opposed to any additional AI. Multiplayer would be better, in fact I think having the game where you can one player play south and the other play north on a campaign. replacing the AI all together would be the highlight. I never said better, I said dont artificially inflate it with number or things the AI did not earn during the campaign of the game. Two and totally different statements. There are ways to increase difficulty without doing so that would not require what you described and others.

To the accusation, I am not blaming anyone, I did not blame a single person or singled them out, all I did was a player/designer/theorist/joint owner of a smalling gaming comapny, my rejection to the proposition that was offered. I know all about scripts and the difficulties that go into programing.

I think what I have said has answered your entanglements.

[Karri]

Scaling remains a major issue, though I think at this point it is caused by a single factor: the AI scales to your whole army, not the units you bring into battle.

 

So if I have two corps loaded with spencers and 2500 men brigades, and a 3rd corps armed with farmers and 500 men per brigade the AI will still bring 3k men brigades with best weapons. So the only option is to either downsize your army(never an option) or always bring your strongest units. The scaling should apply to only units you bring into battle. However, that can cause some gamifying aspects, as such there needs to be some balancing factor still. like minimum strength.

[A. P. Hill]

5 minutes ago, Karri said:

Scaling remains a major issue, though I think at this point it is caused by a single factor: the AI scales to your whole army, not the units you bring into battle. ...

This may be the root cause.  After all, the program has to look somewhere to make a determination as to how big it's got to be, so rather than take time calculating from the battle selection menu pre battle, it examines your whole army database and calculates 1.5 percent or whatever the ratio is difference then throws that number into the fray.

I personally would rather the program looks at the battle you're fighting and then determines that historically, the opposing armies has X number of men in the battle and then limits the provision of men and material to those actual historical limits.   I also wish that the programming would take into account historical personality traits.  For example even though McClellan had superior force at Antietam, (hell any battle where he was a part of,) he husbanded his forces because he did not want to destroy the object of his creation.  Therefore in given battles where McClellan would have been known to be in overall command the AI should react accordingly, thereby sparing troops instead of sending every possible unit it possess into the battle.

But that's just me and my "Purist" attitude.

[Karri]

One issue is that once the troops are in the battle they are in the battle. There are no reserves as such(only troops that arrive later), and no units is ever "off battle".  I wouldn't mind some of the bigger battles taking more "turns" ie give me changes to go to camp and change corps positions and such. And instead of the AI attacking my one corps with first one, then two then three corps it should take turns deploying each corps by turn each a separate "battle". 

[Aetius]

I had the game freeze bug occur twice near the end of Shiloh. On my third run I saved every few minutes, but didn't experience the bug. I did reboot between the second and third runs.

[Aetius]

Units routing through other units isn't entirely fixed. I had it happen multiple times on Harper's Ferry over the weekend. In one case, I had two Union brigades rout into my front line, melee, slip through, continue running into my second line, melee, slip through, and continue running until they reached the river crossing ... where they promptly regrouped and attacked a brigade I had there who had been chasing another runner.

I have a theory about this, based on that behavior: I don't think fleeing units are subject to Condition limits. This is why your pursuing units can't keep up - they are still burning Condition, and slowing down as it drops, while the fleeing units continue moving at maximum speed.

[Aetius]

Started a Legendary campaign over the weekend as Confederates. Took me three tries to win Potomac Fort, and I'm not sure I could do it again. I ended up ignoring the gunboats and going for a forward active defense to buy enough time to win the scenario. Casualties were heavy.

[Nick Thomadis]

This beta is now obsolete. Please switch to official latest version 0.74. It has just been released!