Help needed testing 3.2.x Campaign games!

[alfred007]

if we extrapolate further we can work out roughly what the game is doing when it calculates lancer damage

256 per shot x 2 shots x 10 shots per minute = total damage per minute 5120

This calculation is incorrect. In my test, I watched a full minute and with a shown ROF of 10 in the design screen the lancer shot 5 times per minute 2 rockets as I wrote in my previous post. The lancer is not shooting 10 times 2 rockets as you calculate. Watch a full minute and you will see. In sum 10 projectiles as shown. That means for lancer and cannons (and I think also for all other weapons though I didn't check it for all weapons), you get the correct damage per minute if you multiply the ROF with the damage in the design screen. Of course before modifiers.

Where I totally agree with you is that heavy cannons are underpowered. In the mod I used I set the base damage of heavy cannons to 90 and for lancers to 120. With the also made changes in the modifier files the results are the following damages per minute:

In alpha 12:

Against hover tanks: Lancer 2400; Heavy Cannon 2565
Against tracked tanks: Lancer 2400; Heavy Cannon 2257
Against Cyborgs: Lancer 1440; Heavy Cannon 1744

At the end of beta 1 because of a lot of upgrades:

Against hover tanks: Lancer 3000; Heavy Cannon 3656
Against tracked tanks: Lancer 3000; Heavy Cannon 3217
Against Cyborgs: Lancer 1800; Heavy Cannon 2486

I think now Heavy Cannons are a bit overpowered against hover tanks and cyborgs. I have to do some more calculations to see how I should adjust the modifiers. Maybe it could also be an idea to reduce the value of every upgrade from 30% to 25% for cannons like we did with lancers (from 30% to 20%) But this is another calculation that I have to do when I have the time. And I will also have a look at the other weapons as soon as I can.

With this mod, I played beta 01 with heavy cannon python hover tanks. They have as much HP as a lancer python tracked tank, but are way faster. In the beginning, it was harder than with lancers but the more cannon upgrades I researched it became easier and easier and I was able to win this level with this tanks. I don't like tracked heavy cannons tanks though they have much more HP because they are way too slow.

The problem with the vtol lancers should also be (nearly) solved with the decreased base damage of 180. This means (before upgrades) 5 VTOL lancers with 10 rockets are doing a total damage of 1800 if all rockets hit the same tank. The HP of my heavy cannon python hover tanks is 1900 so they should survive (?). I don't know what upgrades the collective got, also my tanks are ranked and you have to subtract the armor of my tanks. I can tell you more about this after I tested my mod in beta 03 and/or beta 04. But if I'm right they should survive an attack of up to 5 vtol lancers. I will tell you as soon as I tested it.

What I meant with only one rocket hitting my own tank is the situation when the lancer shot 2 rockets but only on hits the target. Even then the target received the full damage. I didn't mean the situation when the lancer shoots only one rocket. And if one the two shot rockets miss the target only half the calculated damage should be seen.

My confusion about my commander is gone as I wrote in my edit of the previous post. The commander had no rank and if you subtract the armor of the python body from the damage of every rocket the calculated and shown damage are the same.

[Berserk Cyborg]

I think finding better weapon/structure modifiers is a good start for balancing. Here is an example of what alfred007 has done. the ones on top are current master. I have not tested, but I would guess it is no where as easy to destroy a hard structure as fast with a MG.

Code: Select all

"ANTI PERSONNEL": {
- "BUNKER": 100,
- "HARD": 75,
- "MEDIUM": 100,
- "SOFT": 200

+ "BUNKER": 45,
+ "HARD": 35,
+ "MEDIUM": 65,
+ "SOFT": 125
}

It may be be necessary to use custom changeOnDiff() functions for some missions.

We can create unique customized versions of a component for the AI if we want. One such useful case would be for the Collective long range artillery (make ripple rockets something to be feared again).

It may be useful to compare a fully upgraded tank killer/lancer and heavy cannon unit in multiplayer. Generally, the cannon wins as long as it is in range of the tank-killer (I only tested in motionless combat). Because the rocket is lacking in HP and has much more reload time the way it wins is to get outside the range of the heavy-cannon and hit from afar using superior speed. In campaign the rocket wins regardless and, ironically, the cannon is much less accurate.

Edit:
Here is a version of libcampaign that does not crash when there are reinforcement transporters. I had to disable truck management in the mean time because it causes an infinite loop when loading a save with these newer Qt versions. Tested on Beta 1. It matches the version in my campaignJS branch, with the addition of what I stated above which may make playing campaign more tolerable at the moment. And I messed with offworld victory, again.

updated-campaign.wz

[Bethrezen]

This calculation is incorrect. In my test, I watched a full minute and with a shown ROF of 10 in the design screen the lancer shot 5 times per minute 2 rockets as I wrote in my previous post. The lancer is not shooting 10 times 2 rockets as you calculate. Watch a full minute and you will see. In sum 10 projectiles as shown. That means for lancer and cannons (and I think also for all other weapons though I didn't check it for all weapons), you get the correct damage per minute if you multiply the ROF with the damage in the design screen. Of course before modifiers.

Just double checked this and timed it using my stop watch and indeed you are correct it only fires 5 salvos per minute for a total of 10 rockets, however I still maintain that the info on the design screen for lancers and similar is incorrect see for every other weapon that fires only a singe round per salvo you can look at that and know what damage each shot does and how many shots that weapon is going to do per minute, assuming 100% accuracy of course, however this is not the case for lancers and similar.

While this is a minor technicality because 256 x 10 and 512 x 5 will give the same answer 2560 damage per minute before modifiers it is inconsistent with the way the weapons info is displayed for all the other weapons like the cannon, because for single shot weapons like the cannon the rate of fire is equivalent to salvos per minute, but the rate of fire for lancers and similar is not salvos per minute its rounds per minute and this is misleading and the fact that I got this wrong proves that.

So for the sake of consistency the rate of fire for lancers should be showing 5 on the design screen because it fires 5 salvoes per minute and the damage should be showing either 256 x 2 as it fires 2 shots per salvo or 512 as that’s the total damage per salvo.

The tooltips should also be updated accordingly so that the damage tool tip reads either
damage per round or damage per salvo

and the rate of fire tool tip should be updated to read either
rounds per minute or salvos per minute

because the difference while subtle is important as that is going to inform you as to how you actual calculate the damage of a given weapon.

With this mod, I played beta 01 with heavy cannon python hover tanks. They have as much HP as a lancer python tracked tank, but are way faster. In the beginning, it was harder than with lancers but the more cannon upgrades I researched it became easier and easier and I was able to win this level with this tanks. I don't like tracked heavy cannons tanks though they have much more HP because they are way too slow.

indeed tracks definitely need rebalancing because they are far to slow, what always puzzled me though is was why tracks add so much armour ? armour should come from the body not the propulsion so the armour that is added by the propulsion should be removed and the armour of the different body types should be adjusted accordingly.

like you i don't really like cannons so much but that's because they are underpowered, plus they don't do bugger all to targets other then vehicles which makes them useless for attacking enemy bases as they cant actually destroy the enemy turrets which is exactly what you are supposed to use heavily armoured units for, to brake through there defences so that the light vehicles can move up with out getting pulverised.

as far as using cannons against the collective even with upgrades i still found levels to be a real slog. Upgraded or not they are just to slow and can't come even remotely close to matching lancers the ironic thing about it is that the issue isn't even damage necessarily or at least not directly the issue is that they are completely ineffective against structures of all types, and speed is really really important particularly when playing on insane not only movement speed but the speed at which you can take out enemy targets.

Personally I'm of the opinion that as far as cannons go they should have there speciality changed from Vehicles to General and they should have there modifiers adjusted accordingly because cannons can fire different rounds for different targets, high explosive for soft targets like buildings, kinetic penetrators for armoured targets and flame rounds for bunkers and the like so it makes no sense that the speciality of cannons is Vehicles only

Edit:
Here is a version of libcampaign that does not crash when there are reinforcement transporters. I had to disable truck management in the mean time because it causes an infinite loop when loading a save with these newer Qt versions. Tested on Beta 1. It matches the version in my campaignJS branch, with the addition of what I stated above which may make playing campaign more tolerable at the moment. And I messed with offworld victory, again.

I'm still waiting on the compatibility issue being sorted out i still have warzone2100-master-20180204-051309-ab17b9b so I'm able to look at things i just can't test any of the changes since the latest masters still wont load for me, but by the sounds of it due to the new QT version you guys cant really test properly either because the change is fouling up your saves.

[alfred007]

The tooltips should also be updated accordingly so that the damage tool tip reads either
damage per round or damage per salvo

and the rate of fire tool tip should be updated to read either
rounds per minute or salvos per minute

I agree and prefer damage per round and rounds per minute. Talking about a salvo when a weapon shoots only one round could maybe confusing.

Personally I'm of the opinion that as far as cannons go they should have there speciality changed from Vehicles to General and they should have there modifiers adjusted accordingly because cannons can fire different rounds for different targets, high explosive for soft targets like buildings, kinetic penetrators for armoured targets and flame rounds for bunkers and the like so it makes no sense that the speciality of cannons is Vehicles only

That's what I did in my current balance mod that you can download in my next post.

as far as using cannons against the collective even with upgrades i still found levels to be a real slog. Upgraded or not they are just to slow and can't come even remotely close to matching lancers

That's why I used hover cannon tanks. As strong as tracked lancer tanks but way faster.

[alfred007]

I extracted the damage and ROF information of some weapons from weapons.json into an excel sheet and found some interesting things and also some issues. I will use the following two abbreviations
BDPM: base damage per minute
This is the damage a weapon is doing in one minute based on weapons.json without modifier or upgrades
DPM: damage per minute
This is the damage a weapon is doing in one minute to a specific type of target including modifier and upgrades

I think I found the difference between firePause and reloadTime
firePause is the gap between every shot of a salvo
reloadTime is the time you need to reload all barrels after the salvo
That means that firePause makes only sense to weapons that shoot salvos (for example the Ripple Rocket Battery). All other weapons should have reloadTime and not firePause.
If you give me an ok I'll do the work.

The issues

1) For Needle Gun, VTOL Needle Gun and VTOL Pulse Laser numRounds is coded. But this are one barrel weapons, numRounds makes no sense and should be deleted

2) The Ripple Rocket Battery has 8 missile launcher but numRounds is set to 6. It's illogical that a weapon with 8 missile launchers shoots salvos of 6 rockets. Same problem with Angel and Archangel Missile. 6 missile launchers and a salvo of four missiles.
While I think numRounds for Ripple Rockets can simply be increased to 8 because they are too weak, reloadTime for Angel Missiles should then be set to 450 and for Archangel Missiles to 600.
Also, the Angel Missile have a higher BDPM than the Archangel Missile. This should not be. So the damage of the Archangel Missile should be increased to 300.

3) The Hellstorm Howitzer have 3 barrels and no numRounds is coded. Also, the BDPM of 1500 is too high in comparison to the other artillery weapons. I think reloadTime was forgotten.
My suggestion: firePause 5; reloadTime 150; numRounds 3, giving a BDPM of about 1000.

4) Needle Gun has the same BDPM as Gauss Cannon and more than Rail Gun. That makes no sense. Gauss Cannon should be the strongest of the three weapons and Needle Gun the weakest.
My suggestion: Needle Gun firePause up to 45, Rail Gun firePause down to 50, Gauss Cannon firePause down to 60.

5) Scourge Missile is way too overpowered because the reloadTime is too short. Scourge Missile have a BDPM of 4800 at the current master. In comparison, the overpowered Assault Gun have a BDPM of 3000. I think it was forgotten that the Scourge Missile shoot two rockets with every salvo like Lancer and Tank Killer. A more realistic value for reloadTime would therefore be 150.

6) I think that the Pepperpot is also overpowered. The Mortar has a BDPM of 500, the Bombard of 533 and the Pepperpot of 1200. Not very consistent. I suggest setting the damage of the Bombard to 90, giving him a BDPM of 600 and the damage of the Pepperpot to 30, giving him a BDPM of 900.

For those that are interested I attach the excel sheet. The most interesting tab is „General“. I moved the cannons and gauss weapons to the new weaponEffect ALL ROUNDER that I use in my current balance mod, that is also attached. In the weapons.json file in my mod only the damage for the Lancer and the Heavy Cannon is changed. Lancer is reduced to 120 from 160 and Cannon is increased to 90 from 70. I think the modifier files are small enough that you can check the changes on your own. For any questions don't hesitate to contact me.

[andrvaut]

I found trouble:
1) not view enemy's transport. possible bug in mod art revolution.
https://youtu.be/OYZ1mC-q3DE?t=3h36m9s there are sounds of transport, but most not seen
2) Sometimes when the research menu is open, units are no longer managed. It seems to me that earlier in the campaign the game paused when opening the research menu. duplicate of #4651

[alfred007]

I found trouble:
1) not view enemy's transport. possible bug in mod art revolution.
https://youtu.be/OYZ1mC-q3DE?t=3h36m9s there are sounds of transport, but most not seen
2) Sometimes when the research menu is open, units are no longer managed. It seems to me that earlier in the campaign the game paused when opening the research menu.

In my tests with the latest master version, the enemy transport was also invisible in beta 01. I think this is a bug in the current master version and not a problem of the art revolution mod.

[Forgon]

[...] 2) Sometimes when the research menu is open, units are no longer managed. It seems to me that earlier in the campaign the game paused when opening the research menu.

"Sometimes" is vague. Could you investigate this further, and improve the description of ticket #4651?

[Berserk Cyborg]

Finally found out the enemy transporter is invisible. I will note that with the deity cheat active they are visible, which is why I failed to see this earlier. Hmm... then it is a visibility issue. Perhaps a side effect of 530d67a8b74025187edf80139edbd40bbfc4293a?

alfred007, you can do anything you want. You and Bethrezen are likely better at stat balancing than I am. Also, your weapons file for lassat missed a small tweak I made a few months ago. Change them back to this for future versions of your mod:

Code: Select all

"periodicalDamage": 380,
"periodicalDamageRadius": 425,
"periodicalDamageTime": 50,

[andrvaut]

[...] 2) Sometimes when the research menu is open, units are no longer managed. It seems to me that earlier in the campaign the game paused when opening the research menu.

"Sometimes" is vague. Could you investigate this further, and improve the description of ticket #4651?

I updated the ticket
video: ​https://youtu.be/DjjSI-21spo

[andrvaut]

3) error on log SUB_2_7:
error |09:18:44: [buildFeature:273] Warehouse(316) already placed at (11+2, 26+1) when trying to place Warehouse(180693) at (11+2, 26+1) - removing it
error |09:18:44: [buildFeature:273] Warehouse(320) already placed at (11+2, 20+1) when trying to place Warehouse(180701) at (11+2, 20+1) - removing it
4) If oil wells are not occupied, when going to the next mission, the beacons over them are extinguished.

[Berserk Cyborg]

Everone can forget about my libcampaign mod I uploaded earlier. It still asserts, and this time something is being morphed into null with savegames. :/

3) error on log SUB_2_7:
error |09:18:44: [buildFeature:273] Warehouse(316) already placed at (11+2, 26+1) when trying to place Warehouse(180693) at (11+2, 26+1) - removing it
error |09:18:44: [buildFeature:273] Warehouse(320) already placed at (11+2, 20+1) when trying to place Warehouse(180701) at (11+2, 20+1) - removing it

Weird, I don't remember seeing that one before. In any case, there were two warehouses almost merged with another warehouse next to them. So I removed the overlapping warehouses and it don't assert anymore. 820d5c64706cd5489c94fe4340f6d73bcf74efdf

And for invisible transporter (visibility.cpp's visibleObject()):

Code: Select all

if (game.type == CAMPAIGN && psTarget->type == OBJ_DROID && isTransporter(castDroid(psTarget)))
{
     return 0;
}

Is where it happens. I will add psViewer->player != selectedPlayer to it. Unless there is something better than that anyone can think of.

[pastdue]

Everone can forget about my libcampaign mod I uploaded earlier. It still asserts, and this time something is being morphed into null with savegames. :/

Could you please test a build of the work-in-progress reduce_qt_dependency_1 branch, and let me know whether it fixes the savegame issues for you? (If not, steps to reproduce the bug with that branch so I can fix it?)

[Bethrezen]

[...] 2) Sometimes when the research menu is open, units are no longer managed. It seems to me that earlier in the campaign the game paused when opening the research menu.

"Sometimes" is vague. Could you investigate this further, and improve the description of ticket #4651?

I updated the ticket
video: ​https://youtu.be/DjjSI-21spo

that bug is listed in the outstanding general bugs threat here
viewtopic.php?f=6&t=13753&sid=a06322e75 ... 583bf20e3c

2.) When you have the research window open there are occasions when it won’t let you select units if you try the computer automatically deselects the selected units.

To trigger this open the research window then close the side bar but not the bottom bar so the screen looks like this.



Then try to select a group of units the first time it should select the group of units as expected but if you then try to select a different group or reselect the same group of units it won’t let you and you will be unable to select any units until you close the research window fully. This doesn't seem to happen on every level and sometimes it does this with the unit build screen open in the same configuration as in the screen shot of the research window.

To make things a bit clearer here is a vid showing the issue https://youtu.be/DjjSI-21spo

its one of many outstanding general issue that we have come across while testing but that vid does do a good job of showing the problem because that's what i see when this issues happens to me, although it seems I'm going to have to reupload the screen shot again because they have been deleted DO'H !!!! stupid image hosts what i will do is add the link to the provided vid as that better demonstrates the issue than a screen shot and written explanation.

[edit]
Ok started replacing the missing screen shots but a couple of them haven't been replaced yet as they are a bit tricker to get so will require a bit more work to recreate the circumstances and it's to late at night to start doing that now so that will have to wait till tomorrow.

in addition i added the link to the video for issue 2.

[Berserk Cyborg]

Could you please test a build of the work-in-progress reduce_qt_dependency_1 branch, and let me know whether it fixes the savegame issues for you? (If not, steps to reproduce the bug with that branch so I can fix it?)

Using Qt 5.9.5 and compiled with GCC 7.3. Mostly everything appears to be working correctly on saveload (no libcampaign crashes/errors)!

I did receive two warnings while loading a save on Beta 1. It did not impact anything from my testing of the mission afterwards.

Code: Select all

warning |09:18:21: [toInt:149] Failed to convert string 'me' to int because of error: stoi (**Further warnings of this type are suppressed.)
warning |09:18:21: [toInt:149] Failed to convert string 'scriptName' to int because of error: stoi (**Further warnings of this type are suppressed.)

Edit:
I tested a skirmish and got the above warnings on saveload. Should an AI script be attached via the debug menu then these messages are shown on saveload:

Code: Select all

info    |09:54:25: [findEngineForPlayer:790] Script context for player 0 and script name semperfi not found
info    |09:54:25: [findEngineForPlayer:790] Assert in Warzone: qtscript.cpp:790 (false), last script event: '<none>'