Help needed testing 3.2.x Campaign games!

The projects speaking tube.
Add your two cents if you want to.
Bethrezen
Regular
Regular
Posts: 584
Joined: 25 Sep 2009, 02:05

Re: Help needed testing 3.2.x Campaign games!

Post by Bethrezen » 07 Jan 2018, 18:15

Pushed 7523d1ae737fa67179d9b9566599bdcaaa9931c7. I saw a few things that needed a change in Beta 4. Good news is the hover lancers will defend that one base instead of being shelled to death, but that means taking out the command center for disabling the VTOLs will be a bit tricky on insane.

At first I tried playing a conventional way (setting up on the escape route), and got crushed by 40+ hover lancers going to defend the escape route, but soon realized that the constant stream of lancer VTOLs are quite taxing on northward progress unless you build a bunch of walls and Hurricane AA sites every few tiles. So I opted for a rush tactic to intercept the commander before he gets to the western base and used a bunch of newly produced lancers tanks to absorb the initial wave of Collective lancers. Basically, the longer you wait on this mission, the tougher it gets.
by the sounds of it this might just be one of them situations where you simply need more numbers, because i know in the past that in order to do that level using artillery you need grater numbers because artillery is really really slow, although what i tent to do as soon as i land is move all the way to the west and wait for the commander and it's guards to cross the river and then i just ambush them and target all weapons on the commander, at which time I can either retreat and end the level or shell the collective back to the stone age for there impudence with my bombards since hover units will go down pretty quickly to massed bombard fire since they are lightly armoured but I'll have to wait and see how that plays out with the latest changes because at the moment the lancer vtols on beta 03 are causing me some real trouble since none of my units can survive against there lancer vtols and cannons are not a viable option because they don't inflict enough damage fast enough, or at least not when you are playing on insane, but even if they did that still doesn't solve the problem of there lancer VTOLs one shotting my mobile AAA and of course if i loose my AAA them I'm pretty well screwed because there vtols will just pick me apart which is pretty much what is happening to me on beta 03 at the moment.

User avatar
Berserk Cyborg
Code contributor
Code contributor
Posts: 615
Joined: 26 Sep 2016, 19:56

Re: Help needed testing 3.2.x Campaign games!

Post by Berserk Cyborg » 07 Jan 2018, 20:26

Take advantage of the AI almost always attacks the closest object to them (hint: place walls in front of anything you want to protect). I have a huge base setup where the first Collective base is on Beta 3 which is pretty much impenetrable to their VTOLs so Beta 3 was easy for me.

Pushed 1ca02b037b60f5c2e1f10f4b9d59d18035b6a96e. Just a few modifications to the script of Beta 5 to make it more interesting. The newly used sensor units help the ripple rockets (which are quite powerful if they can hit something). I will say that if you have a poorly designed base you will get wrecked by their VTOLs and those bunker buster units are annoying as ever.

alfred007
Trained
Trained
Posts: 434
Joined: 31 Jul 2016, 06:25
Location: Stuttgart, Germany

Re: Help needed testing 3.2.x Campaign games!

Post by alfred007 » 07 Jan 2018, 22:29

Berserk Cyborg wrote:Take advantage of the AI almost always attacks the closest object to them (hint: place walls in front of anything you want to protect)
That's what I always did in Beta mission. In Beta 03 I just needed 4 AA Tanks to handle the Collective VTOLs because I placed two sensor towers between the VTOLs and my tanks. The VTOLs attacked first the sensor towers with the lancers and later my tanks and repair facilities with their bombers. I couldn't destroy all VTOLs in every wave, but I had them under control that way. And for the VTOLs in Beta 04, I used 8 AA tanks placing them a few tiles south of their spawning point and build some MG bunkers in front of my AA tanks. The Collective VTOLs attacked the bunkers first and then they had not enough firepower left to destroy my AA tanks. For the western base in Beta 04, I used a group of 6-8 bombards attached to a sensor unit, destroying the western base carefully piece by piece. And for the eastern base, I used a combat group of 16 tracked mantis lancers attached to an experienced commander. And this combat group can also help to destroy the western base after the eastern base is destroyed. But always very carefully so that they don't "wake up" the hover lancers. I will see if this still works after the radius of the hover lancers are increased.

Bethrezen
Regular
Regular
Posts: 584
Joined: 25 Sep 2009, 02:05

Re: Help needed testing 3.2.x Campaign games!

Post by Bethrezen » 08 Jan 2018, 00:26

Berserk Cyborg wrote:Take advantage of the AI almost always attacks the closest object to them (hint: place walls in front of anything you want to protect). I have a huge base setup where the first Collective base is on Beta 3 which is pretty much impenetrable to their VTOLs so Beta 3 was easy for me.

Pushed 1ca02b037b60f5c2e1f10f4b9d59d18035b6a96e. Just a few modifications to the script of Beta 5 to make it more interesting. The newly used sensor units help the ripple rockets (which are quite powerful if they can hit something). I will say that if you have a poorly designed base you will get wrecked by their VTOLs and those bunker buster units are annoying as ever.
Unfortunately that doesn't work in my case because it's not my base they are attacking it's my units and I can't put walls around them because I'm on the move, even with the changed stats mod that Alfred uploaded earlier between them they are still doing over 3000 damage I know this because they attacked one of my heavy body tracked commanders that was at full health and they reduced it from 4748hp to a little over 1000 hp so I don’t think there can really be any argument that there lancer vtols are doing way to much damage and that’s after they where nurfed by the changed stats mod now either that mod wasn’t working right or they are even more overpowered in an unmodded game than they appear to be and to be perfectly honest relying AI exploits to evade an unfair advantage that the computer shouldn’t have in the first place isn’t a good way to go about things.

To be perfectly honest when they can one shot any of your units with perhaps the exception of heavy body tracked commanders and heavy body tracked heavy cannons then i really don't see how you are supposed to defend against that fairly because exploiting the AI like that is in my opinion cheating.

its why when we where going through Alpha campaign i made the point that really when you bomb enemies from beyond sight range with artillery they should come looking for where the fire is coming from instead of just sitting there like newbs getting shelled to death because again that sort of a thing could be considered an exploit, but obviously this sort of a thing is not necessarily easy to fix because if it was it probably would have been fixed already.

User avatar
Berserk Cyborg
Code contributor
Code contributor
Posts: 615
Joined: 26 Sep 2016, 19:56

Re: Help needed testing 3.2.x Campaign games!

Post by Berserk Cyborg » 08 Jan 2018, 01:52

I am curious to know why these VTOL lancers do that much damage. I just checked on Beta 6 and with all research that I have my VTOL lancers does about 1200 damage to a tracked tiger hpv cannon unit (have HESH rockets mk3). And on Beta 6 an enemy VTOL lancer took one of my half-track python lancers from 1512 to 412 HP (have dense mk2 alloys).

It could be an extreme balance side effect for being able to receive all research items from previous campaigns. Before I squished the research files into one file to interconnect everything for easy research dependency pulling across the campaigns, the research tree was more like Alpha all the time in terms of depth. And this is why Beta (and maybe Gamma) have these new research items popping up because they depend on research items that truly never existed until a few months ago for that campaign (research needs to be redone for Alpha -> Beta and Beta -> Gamma transitions which is why it takes longer to load those two missions than others).
Bethrezen wrote: its why when we where going through Alpha campaign i made the point that really when you bomb enemies from beyond sight range with artillery they should come looking for where the fire is coming from instead of just sitting there like newbs getting shelled to death because again that sort of a thing could be considered an exploit, but obviously this sort of a thing is not necessarily easy to fix because if it was it probably would have been fixed already.
I've had some ideas about this for a little while. Campaign has a good amount of droids over a bunch of missions that just sit and do nothing (and so far I mostly have let it stay that way).

My idea is to enumerate around the victim's position so many X tiles and create a new group dynamically, via eventAttacked, with droids that do not already belong to a group (think of those nine Alpha 12 tracked cannons near the LZ that do nothing, and then say you hit them with a mortar and then all of a sudden they go attack the player). Would be quite simple.

Bethrezen
Regular
Regular
Posts: 584
Joined: 25 Sep 2009, 02:05

Re: Help needed testing 3.2.x Campaign games!

Post by Bethrezen » 08 Jan 2018, 03:37

I am curious to know why these VTOL lancers do that much damage. I just checked on Beta 6 and with all research that I have my VTOL lancers does about 1200 damage to a tracked tiger hpv cannon unit (have HESH rockets mk3). And on Beta 6 an enemy VTOL lancer took one of my half-track python lancers from 1512 to 412 HP (have dense mk2 alloys).

It could be an extreme balance side effect for being able to receive all research items from previous campaigns. Before I squished the research files into one file to interconnect everything for easy research dependency pulling across the campaigns, the research tree was more like Alpha all the time in terms of depth. And this is why Beta (and maybe Gamma) have these new research items popping up because they depend on research items that truly never existed until a few months ago for that campaign (research needs to be redone for Alpha -> Beta and Beta -> Gamma transitions which is why it takes longer to load those two missions than others).
so am i because if you divide the damage there lancer vtols are doing to me on beta 3 by 4 as there are 4 lancer vtols then that comes out at around 800 or 900 damage each, which seems to be roughly double the damage that the ground base equivalents do as my lancer tanks do 400 damage with HESH rockets mk2.

To be honest I'm of the opinion that something screwy is going on with the lancer upgrades because at the start of Beta 01 on v1.10 my lancers have 160 damage but on master they have 304 damage which is almost double the damage that can't possibly be right surly ? How are your lancers getting so powerful so quickly ? Then of course if you factor in the fact that HESH rockets don't seem to be available on v1.10 or at least i haven't encountered them yet then that just makes the damage difference even grater.

Although come to think of it wasn't there a bug that made lancers to powerful ? so much so that at the point you got tank killers it's was pointless to use them as lancer where either more powerful or equal in power but had a faster fire rate did that one ever get fixed ?

Personally I'm of the opinion that the lancer upgrades probably need scaling back a bit because if you look at what each upgrade gives in v1.10 compared to what it does in master I'd bet you'd find that the upgrade in master give a bigger boost, then of course you have to factor in the difficulty scaling which never excised before, because if on insane the damage being done by enemies is increased by 10% for example then that can make a huge difference to a weapon that is already overpowered due to xyz issue.

For example 10% of 800 is 80 and given that HESH rocket upgrade increase damage by 48 i think it was, then 80 points of damage is almost equivalent to 2 more damage upgrades, which is like giving the computers vtols HESH rockets mk5 and when you think of it that way it very quickly becomes apparent why they are so hugely overpowered.

alfred007
Trained
Trained
Posts: 434
Joined: 31 Jul 2016, 06:25
Location: Stuttgart, Germany

Re: Help needed testing 3.2.x Campaign games!

Post by alfred007 » 08 Jan 2018, 03:47

Berserk Cyborg wrote:My idea is to enumerate around the victim's position so many X tiles and create a new group dynamically, via eventAttacked, with droids that do not already belong to a group (think of those nine Alpha 12 tracked cannons near the LZ that do nothing, and then say you hit them with a mortar and then all of a sudden they go attack the player). Would be quite simple.
I think this a good idea. This would make some levels more competitive and interesting. But for Beta 04 you should then give those lancer hovers the order to attack but not to defend. Because if they get accumulated to 15 or 20 you would have no chance against them no matter what difficulty you play.

Unfortunately, I can't help you with those levels at the moment because my beta saves of the beginning of Beta mission are from September and due to all these changes that were made since then, I can't reload these saves with the current master. I will move on with Gamma mission the next days so that we can push the next release latest end of January and then I will start a new campaign to help you with adjusting those levels.

For those annoying VTOL lancers in Beta 03, it would maybe help to use some cheap suicide units (wheeled bug sensor or mg tanks) and send them north so that they attract the fire of the VTOLs.

User avatar
Darkling
Rookie
Rookie
Posts: 18
Joined: 07 Feb 2016, 00:45

Re: Help needed testing 3.2.x Campaign games!

Post by Darkling » 08 Jan 2018, 03:54

My idea is to enumerate around the victim's position so many X tiles and create a new group dynamically, via eventAttacked, with droids that do not already belong to a group (think of those nine Alpha 12 tracked cannons near the LZ that do nothing, and then say you hit them with a mortar and then all of a sudden they go attack the player). Would be quite simple.
Wouldn't that have the effect of giving every AI unit a cb radar? Rather than setting it to attack something that it can't see perhaps have it (or its group) scout towards a point say halfway in between? That way it has a chance of discovering the attacker "naturally".

Bethrezen
Regular
Regular
Posts: 584
Joined: 25 Sep 2009, 02:05

Re: Help needed testing 3.2.x Campaign games!

Post by Bethrezen » 08 Jan 2018, 04:14

alfred007 wrote:
Berserk Cyborg wrote:My idea is to enumerate around the victim's position so many X tiles and create a new group dynamically, via eventAttacked, with droids that do not already belong to a group (think of those nine Alpha 12 tracked cannons near the LZ that do nothing, and then say you hit them with a mortar and then all of a sudden they go attack the player). Would be quite simple.
I think this a good idea. This would make some levels more competitive and interesting. But for Beta 04 you should then give those lancer hovers the order to attack but not to defend. Because if they get accumulated to 15 or 20 you would have no chance against them no matter what difficulty you play.

Unfortunately, I can't help you with those levels at the moment because my beta saves of the beginning of Beta mission are from September and due to all these changes that were made since then, I can't reload these saves with the current master. I will move on with Gamma mission the next days so that we can push the next release latest end of January and then I will start a new campaign to help you with adjusting those levels.

For those annoying VTOL lancers in Beta 03, it would maybe help to use some cheap suicide units (wheeled bug sensor or mg tanks) and send them north so that they attract the fire of the VTOLs.
it would certainly make life more interesting because i cant count the number of times i have breezed through a level using nothing more than mortars/bombards because the enemy units only react if they see me which means that i can bomb them from range and they will just sit there like retarded newbs instead of trying to figure out where the artillery fire is coming from, now figuring that out with short range artillery units like mortars/bombards is easy enough but trying to figure that out when it's long range artillery like ripple rockets that's a good deal more difficult but in that case you would simply go looking for the sensor rather that the actually artillery which is what a human player would do as they may or may not be able to actually get to the ripple batteries, beta 5 being a case and point you can't get to the ripple batteries initially as you need to fight your way through there defences first so its better to just knock off there scanners instead and simply cut the range of there ripples down.

having said that doing this could be problematic on some levels beta 4 comes to mind as you are facing off against a lot of rocket units which could actually leave you in a situation where you can't actually complete the level because you just get crushed because to many units come to quickly, so on those levels care would need to be taken but otherwise i think it would be a good idea if enemy units actually reacted to being shot and come looking for you if you are attacking them from out side there visual range.

Bethrezen
Regular
Regular
Posts: 584
Joined: 25 Sep 2009, 02:05

Re: Help needed testing 3.2.x Campaign games!

Post by Bethrezen » 08 Jan 2018, 04:24

Wouldn't that have the effect of giving every AI unit a cb radar? Rather than setting it to attack something that it can't see perhaps have it (or its group) scout towards a point say halfway in between? That way it has a chance of discovering the attacker "naturally".
short answer

no

long answer

at the moment if i attack a units with mortars for example then that unit wont react it will just sit there getting shelled to death why because it can't see me because scanners have a bigger range of sight, this is in my opinion an exploit.

The idea behind this is that if i bomb you with mortars then the units will react in the same way they would if they saw me aka they will come looking for where the fire is coming from and when they find my units units they will attack me, doing this stops players from exploiting design flaws in the AI and helps to make the game more of a challenge, this is not about giving the computer an unfair advantage its simply making the AI react more like a human would.

Think about it this way if you are siting there building your base and the computer starts shelling you with mortars for example what are you going to do you are going to send units in the general direction of the incoming fire looking for the enemy units, and then you are going to blow them to bits, the idea behind this is to make the computer do the same thing.

User avatar
Darkling
Rookie
Rookie
Posts: 18
Joined: 07 Feb 2016, 00:45

Re: Help needed testing 3.2.x Campaign games!

Post by Darkling » 08 Jan 2018, 04:46

I wasn't suggesting that it not respond. What I was suggesting was that it scout in the direction of the incomming fire. (DORDER_SCOUT vs DORDER_ATTACK). I think the AI should respond to an attack and that the response should be to scout towards the attack. By scouting to a point halfway in between it's moving close towards the attacker and if it's attacked again it will get close enough to spot and return fire. Players can tell the direction that fire is coming from, but not exactly where so we scout towards it. Setting the AI to scout to a point halfway inbetween is a compromise that gets it moving closer to the attacker but without setting it to attack a specific target.
I interpreted the initial sugestion as saying that the AI should respond by attacking which requires it to know where the attacker is (hence the CB reference).

User avatar
Berserk Cyborg
Code contributor
Code contributor
Posts: 615
Joined: 26 Sep 2016, 19:56

Re: Help needed testing 3.2.x Campaign games!

Post by Berserk Cyborg » 08 Jan 2018, 06:44

It is not that big of a deal since the droids that are not grouped to anything and are quite low in numbers (1-3 commonly next to each other). I was talking about droids that are already placed on map, with no group and have no resemblance of any form of intelligence at all. These droids are quite content taking punishment and only move if something comes within a very short range of them (and even then some are so lazy they prefer to sit at their position and do nothing unless they can hit it).

The most famous example would be at 8:03 in this video. Beta 2 is composed of probably 75% no group droids with one or two sitting next to each other. Overall, I think it would be more exciting for them to simply be given a generic campaign library attack order and seek out stuff close to them and repair if needed.

EDIT:
Here it is in 0b4012f09e0d0810440e068f4f11077919379592. Most missions have at least a few units that do nothing so hit one and it will attack and bring in nearby non-grouped allies belonging to the victim's player number. Beta 1 and 2 show how it works quite well. Only potential problem I see is that any group activated by some timer could lose a unique order. But I guess its better than them being destroyed or weakened in that case anyway.

Bethrezen wrote: To be honest I'm of the opinion that something screwy is going on with the lancer upgrades because at the start of Beta 01 on v1.10 my lancers have 160 damage but on master they have 304 damage which is almost double the damage that can't possibly be right surly ?
Each of those upgrades gives a 30% increase for damage so maybe doing 20-25% could make a difference? Compared to every current release, the player and the Collective now receive all research from Alpha unlike before (three more kinetic alloys, three more rocket damage upgrades, ...and so on). You only see 160 in past versions because Beta had no concept of the research in Alpha (hence why I said the research tree was pretty much the equivalent of Alpha in terms of upgrade quantity for each campaign). Look at the HP of a collective heavy cannon tank in 3.2.3 and compare it to master as an example.

Math checks out since the base damage of lancer is 160. Assuming it only uses the base value per upgrade, a 30% increase in lancer damage is 48. 48 x 3 = 144. 160 + 144 = 304.

EDIT:
Pushed 0ee481f258d93e737595a6f2b9d123652234ccc2 for Beta 6. Should be easier to do a total annihilation victory because of the new base (where the pit vtols are) and extending the main base to include the walls to the west of it.

Bethrezen
Regular
Regular
Posts: 584
Joined: 25 Sep 2009, 02:05

Re: Help needed testing 3.2.x Campaign games!

Post by Bethrezen » 09 Jan 2018, 21:11

Berserk Cyborg wrote:
Bethrezen wrote: To be honest I'm of the opinion that something screwy is going on with the lancer upgrades because at the start of Beta 01 on v1.10 my lancers have 160 damage but on master they have 304 damage which is almost double the damage that can't possibly be right surly ?
Each of those upgrades gives a 30% increase for damage so maybe doing 20-25% could make a difference? Compared to every current release, the player and the Collective now receive all research from Alpha unlike before (three more kinetic alloys, three more rocket damage upgrades, ...and so on). You only see 160 in past versions because Beta had no concept of the research in Alpha (hence why I said the research tree was pretty much the equivalent of Alpha in terms of upgrade quantity for each campaign). Look at the HP of a collective heavy cannon tank in 3.2.3 and compare it to master as an example.

Math checks out since the base damage of lancer is 160. Assuming it only uses the base value per upgrade, a 30% increase in lancer damage is 48. 48 x 3 = 144. 160 + 144 = 304.
That is worth a try to see what sort of a difference it makes, general I'm of the opinion that between the 4 of them the collective lancer VTOLs shouldn't do more that about 500 damage each or 2000 damage total because at 500 damage each or 2000 damage total that is just barley survivable for Heavy body tracked AAA and as for your lancers while a tracked heavy body lancer only has 1909 HP with experience and attached to an experienced commander there armour might be boosted just enough to survive that again just barley, now obviously this damage can maybe be pushed up a bit more after Beta 03 as you get the compost alloy armour boost but since you don't get that till near the end of the level then for the sake of fairness they probably shouldn't do more than that.

now obviously this is going to take a bit of tinkering to find exactly the right value but what i would be aiming at for there lancer vtols attacks to be just barley survivable on insane and then decreased in severity from there for the easer difficulties.

alfred007
Trained
Trained
Posts: 434
Joined: 31 Jul 2016, 06:25
Location: Stuttgart, Germany

Re: Help needed testing 3.2.x Campaign games!

Post by alfred007 » 09 Jan 2018, 21:21

I updated to master 0b4012f and started testing Gamma 04. Because of the changes, I asked for, Gamma 04 became way too easy. I was wrong. I could win the level in 3 minutes by rushing south, rescuing the alpha team and then rush north to my LZ. Makes no sense. For a comparison, I played this level with 3.2.3 and 3.1.5.

I found four issues in Gamma 04. First, I could command the alpha team after I reached the phantomFacTrigger. That's too soon. I should not able to command them until I reach the rescueTrigger like in 3.1.5 and 3.2.3. Second, the truck of the alpha team disappeared as soon as I reached the phantomFacTrigger. I guess because I already had 15 trucks. But this doesn't happen in 3.2.3 or 3.1.5. There, the alpha truck became my 16th truck. Third, the NE and SW factories should only produce cyborgs. Fourth, the cyborgs fight again till death instead of retreating for repair. Same problem we have with the cyborg's behaviour in Gamma 02.

And I made a short look into the beginning of Gamma 05 and there are two things that should be changed. At the beginning, my HQ is gone. In 3.2.3 and 3.1.5 Nexus took over my HQ and I had to destroy it. And the VTOLs should alternately appear like they do in Gamma 03.

In the logs, we have again the mapTile and assert infos and also path findings errors.

In 3.2.3 and 3.1.5 all three factories started their production as soon as I reached the rescueTrigger with sending the first production immediately on the map. I think we should do that again. Instead of making the level easy and then step by step tougher we should make it as tough as possible and only if we see it is too tough making it a bit easier. I was too guardedly.
Attachments
logs gamma 4.zip
(7.26 KiB) Downloaded 29 times

alfred007
Trained
Trained
Posts: 434
Joined: 31 Jul 2016, 06:25
Location: Stuttgart, Germany

Re: Help needed testing 3.2.x Campaign games!

Post by alfred007 » 10 Jan 2018, 01:26

@Berserk Cyborg
I visited your GitHub page and saw the changes for Gamma 04. That will be really as tough as possible. You coded the land reinforcements to spawn immediately after reaching the phantomFacTrigger. What I meant is immediate spawning after reaching the rescueTrigger. Is this an intended difference?

Post Reply