Home » MODDING HQ 1.13 » v1.13 Coding Talk » Experimental Project 7
Re: Experimental Project 7[message #355957 is a reply to message #355956] Fri, 23 November 2018 03:51 Go to previous messageGo to next message
sevenfm

 
Messages:1935
Registered:December 2012
Location: Soviet Russia
buuface wrote on Fri, 23 November 2018 06:38
Personally do you play with NCTH enabled?

I usually test AI features with OCTH, but NCTH works well, too. It's just a bit different in some aspects.

Quote:
Did I also correctly recall somewhere in this thread you mentioned a 'focus' aim feature which can effect the chance of an interrupt?

Not in the stable release, though I plan to implement something similar.



7609+fix | 7609+AI (r1180) | Unofficial modpack | Win8+ fix | Experimental project | Youtube

"It's already "dog-eat-dog", friend. Not sure what worse a bunch of zombies could do."


Re: Experimental Project 7[message #355961 is a reply to message #355957] Fri, 23 November 2018 15:35 Go to previous messageGo to next message
sevenfm

 
Messages:1935
Registered:December 2012
Location: Soviet Russia
Ja2+AI r873

- suppression balance tweaks
- new option: DIRECTIONAL_PENALTY (TRUE) - OCTH penalty for shooting at opponent moving in perpendicular direction to shooter
- OCTH: lowered additional movement penalty for scopes
- limit maximum possible vision distance to double day vision distance
- LOS: check all smoke effects for sight penalty
- no vision range bonus from NVG when using night scope
- no vision range bonus from head items when using scope/binocs
- cannot use skills during interrupt
- AI: limit smoke grenade use (prefer other attack if best CTH >= 25, don't use smoke if found smoke near attack spot, reduce smoke attack value at close range)
- better formula for grenade throwing range: for better balance, demolitions/throwing trait increase effective strength instead of direct range bonus
- [X] always allow exchanging places in cheat mode
- MakeCivHostile: always apply custom side from CivGroupNames.xml (if it's defined for civ group)
- fix memory leak that caused random crashes in tactical
- code optimization for better AI speed
- [F] correctly shows CTH for throwing knives
- better CTH calculation for throwing knives, penalty when attacking unseen opponent, no aim bonus when attacking unseen opponent
- throwing: better miss spot calculation

Better behavior for blinded soldiers:
- use swatting when blinded (instead of running or walking)
- set max aiming levels for blind soldier to 1
- Panic AI: blind soldier cannot become chosen one, blind soldier cannot do panic AI except activating detonator
- blind soldier can only attack seen/heard personally, only with suppression fire
- cannot use mortar if blinded
- cannot use grenade launcher if blinded
- cannot use rocket launcher if blinded
- blind soldier cannot attack if not personally seen/heard opponent this turn
- blind soldier always has min cth when attacking

Externalized keys and locks to xml (by anv):
- new option XML_LOCKS (FALSE by default) enables support for xml locks and keys, if option is disabled, original locks.bin used for locks, original hardcoded values are used for keys
- Locks (TableData/Items/Locks.xml): Originally 64 locks were kept in BinaryData/LOCKS.BIN. Now up to 255 locks can be specified in TableData/Locks.xml. If it isn't found, LOCKS.BIN will be loaded normally.
- Keys (TableData/Items/Keys.xml): Originally 10 keys were hardcoded in KeyTable[]. Now up to 255 keys can be specified in TableData/Keys.xml. If it isn't found, hardcoded array will be used normally.
- see http://thepit.ja-galaxy-forum.com/index.php?t=msg&th=21824&goto=331829&#msg_331829
- updated map editor to support locks/keys
- new savegames cannot be loaded with older ja2.exe

[Updated on: Fri, 23 November 2018 18:01]




7609+fix | 7609+AI (r1180) | Unofficial modpack | Win8+ fix | Experimental project | Youtube

"It's already "dog-eat-dog", friend. Not sure what worse a bunch of zombies could do."


Re: Experimental Project 7[message #355962 is a reply to message #355961] Fri, 23 November 2018 18:23 Go to previous messageGo to next message
Taleman

 
Messages:64
Registered:October 2010
Location: Finland
Playing on wine, Debian GNU/Linux 9.6, Ja AI4866.exe. No mods.

Quote:
$ ja2.sh
PANIC: unprotected error in call to Lua API (not enough memory)
0009:err:ntdll:RtlpWaitForCriticalSection section 0x151055f8 "?" wait timed out in thread 0009, blocked by 0030, retrying (60 sec)
/home/taleman/bin/ja2.sh: rivi 7: 2401 Päätetty wine ja2_7609en+AI_r866.exe



I had half of memory free, from 8 GB. What memory is not enough here? Happened mid in battle, after shooting autofire. Save is from end of previous turn, but this turn was longish when panic happened. I can supply save file if needed.



Tapio Lehtonen
Pori, Finland
Re: Experimental Project 7[message #355963 is a reply to message #355962] Fri, 23 November 2018 18:44 Go to previous messageGo to next message
sevenfm

 
Messages:1935
Registered:December 2012
Location: Soviet Russia
Quote:
I had half of memory free, from 8 GB. What memory is not enough here? Happened mid in battle, after shooting autofire. Save is from end of previous turn, but this turn was longish when panic happened. I can supply save file if needed.

Please post here or send me a save if bug can be reproduced.
r866 should not have memory leak problem which caused crashes previously.
32 bit application cannot allocate more than 2gb of memory as far as I know.

[Updated on: Fri, 23 November 2018 19:06]




7609+fix | 7609+AI (r1180) | Unofficial modpack | Win8+ fix | Experimental project | Youtube

"It's already "dog-eat-dog", friend. Not sure what worse a bunch of zombies could do."


Re: Experimental Project 7[message #355969 is a reply to message #355963] Sat, 24 November 2018 04:29 Go to previous messageGo to next message
buuface

 
Messages:165
Registered:October 2011
Currently playing ARmod with +AI(874) on experienced 'soft' iron man. I'm just a few hours through the first day but thought i would share my feedback.
My squad of 8 has taken omerta and is now moving towards san mona to meet Tony and do some boxing. I've come across a patrol of 9 enemies an am waiting in ambush at a map with some farm houses along the road.

Despite using superior tactics/terrain advantage and ambushing the enemy, i'm finding myself consistently overwhelmed even against evenly numbered 'red' shirt enemies.

One problem I have is that my mercs seem to be very unreliable shooters. I have 2 imps one with machinegunner and one with auto-weapons trait. They are both armed with 10mm MP5 and have 85 marksmanship. Yet the advantage of wielding heavier guns than the enemies (who all have pistols so far) seems minimal. Outside of pistol range my smg's are too inaccurate, and inside pistol range the enemies pistols seem superior (less ap to fire, more accurate, and eerily large number of lucky headshots). My shots that do hit (inside of the 18 range of the mp5) do very little damage. It seems like an issue that because the enemy will now take shot from quite a long way outside the guns 'useful' range, 1 out of 5 of those quick shots with a pistol are as likely if not more likely to hit than a single aimed shot from a longer-ranged weapon.

I also have an m1 carbine with HP rounds and a x3 scope which is consistently doing 0 or 1 dmg per hit at optimal range. I suppose most of the enemy might be at least partially armored but doing 0 dmg seems to make HP near useless as far as i can tell.

Any tactical advice would be welcome. I love that the AI seems smarter than before but I can't seem to deal with the way guns/skills are balanced atm.

Another thing i noticed is that enemies are taking full 12guarge BS and slug rounds to the body, which are also doing very little damage and barely seeming to suppress them either. They turn around and are able to get off 4-5 pistol shots the next round as though totally unfazed.

[Updated on: Sat, 24 November 2018 04:56]

Re: Experimental Project 7[message #355971 is a reply to message #355969] Sat, 24 November 2018 09:38 Go to previous messageGo to next message
sevenfm

 
Messages:1935
Registered:December 2012
Location: Soviet Russia
@buuface do you have a save from the beginning of the battle or just before it? I could tell more if I could look closer at the situation.
Are you playing stock 1.13+official AR mod or my modpack?

With 8 mercs you shouldn't have any problem in the beginning :-)

Red shirts are army, and army soldiers in Ja2+AI always have some vest, and if the mod provides no uniform with 1 armour, they will wear at least flak jacket, but it shouldn't be too good at stopping bullets.
It's often better to shoot at legs with buckshot, as enemy will fall and lose some AP, and he usually has no legs armor.
As for BP losing from buckshot hit, can't say without looking at the save, but in general it should work the same as in stock 1.13, and is controlled by ini settings in APBPConstants.ini
HP rounds are useless if enemy wears armour, better shoot at heads/legs with them or use them again admins or cats.
SMGs being inaccurate when shooting beyond weapon range is a change to NCTH in Ja2+AI - you cannot use SMGs at any distance like you do in stock 1.13, you need to use a weapon with longer range (AR or MG) to hit/suppress enemy at distance.



7609+fix | 7609+AI (r1180) | Unofficial modpack | Win8+ fix | Experimental project | Youtube

"It's already "dog-eat-dog", friend. Not sure what worse a bunch of zombies could do."


Re: Experimental Project 7[message #355972 is a reply to message #355971] Sat, 24 November 2018 10:37 Go to previous messageGo to next message
buuface

 
Messages:165
Registered:October 2011
https://ufile.io/jrqho

Give it a try and see if you can do better than I am doing!

I am using your modpack and the AR mod you included with it. No other significant changes.

[Updated on: Sat, 24 November 2018 10:46]

Re: Experimental Project 7[message #355973 is a reply to message #355972] Sat, 24 November 2018 10:42 Go to previous messageGo to next message
buuface

 
Messages:165
Registered:October 2011
Does the ncth settings/skills settings differ according to the mod ini i'm running with?

Maybe my problems are specific to the AR mod?
Re: Experimental Project 7[message #355974 is a reply to message #355973] Sat, 24 November 2018 10:49 Go to previous messageGo to next message
sevenfm

 
Messages:1935
Registered:December 2012
Location: Soviet Russia
buuface wrote on Sat, 24 November 2018 13:42
Does the ncth settings/skills settings differ according to the mod ini i'm running with?

Maybe my problems are specific to the AR mod?

AR mod is map mod, it doesn't change weapons or weapon balance.
There are custom ncth and skills settings in Data-User\CTHConstants.ini which are applied to all mods in modpack, but you shouldn't have problems playing with them.
I will look at your save and maybe tell more.

Are you sure you are using ja2_7609en+AI_r874.exe? The game says it's older save and asks me to update it.
It seems you are playing with OCTH, for some reason I was thinking you are using NCTH.
Scope no.32 is in a bad condition, it seems it doesn't provide full bonus.

[Updated on: Sat, 24 November 2018 11:24]




7609+fix | 7609+AI (r1180) | Unofficial modpack | Win8+ fix | Experimental project | Youtube

"It's already "dog-eat-dog", friend. Not sure what worse a bunch of zombies could do."


Re: Experimental Project 7[message #355975 is a reply to message #355974] Sat, 24 November 2018 11:16 Go to previous messageGo to next message
buuface

 
Messages:165
Registered:October 2011
I think i'm playing with the version right before the one you most recently uploaded.
I got it from the link in your sig about 2 days ago.

I started with NCTH but was having more or less the problems i described to you, so i restarted and switched to octh and turned off nis.
Re: Experimental Project 7[message #355980 is a reply to message #355975] Sat, 24 November 2018 20:04 Go to previous messageGo to next message
sevenfm

 
Messages:1935
Registered:December 2012
Location: Soviet Russia
@buuface
It is possible to win this battle, but not without casualties, see video:


The question is - do you really need to win this battle? There's nothing important in this sector, it doesn't have mine income or anything, as far as I understand. So it may be better to just retreat. In Arulco Revisited mod it's better to play guerilla style war until you find some better weapons and train mercs a bit.
Also, this sector is not well suited for defense, unlike some sectors like SAMs which have fortifications. In this sector you are actually at disadvantage when defending because enemy has too much cover around your defense positions, and there are too many windows in houses so it's hard to defend there.
Also it may be good idea to disable reinforcements in Ja2_Options.ini, as it will make game easier and allow more place for tactical maneuvers in sector.

[Updated on: Sat, 24 November 2018 20:38]




7609+fix | 7609+AI (r1180) | Unofficial modpack | Win8+ fix | Experimental project | Youtube

"It's already "dog-eat-dog", friend. Not sure what worse a bunch of zombies could do."


Re: Experimental Project 7[message #355981 is a reply to message #355980] Sat, 24 November 2018 23:53 Go to previous messageGo to next message
LatZee

 
Messages:126
Registered:December 2015
Yeah, AR comes from pre reinforcement times and enemy group density makes it very hard if you use it with reinforcements allowed. So if you want to leave it on (and I like it that way, masochism FTW) you need to pick your fights better. Especially super early on when your equipment is trash. Fights like this one where you're going to get attacked from one side and reinforced from two more are basically suicide.

You should probably try to reach San Mona at least before taking on major fights like this one, there is enough equipment there to kickstart you in right direction. It is probably better to reach San Mona from Omerta by going the northern route alongside the edge of the map and then coming down south, even if you bump into something along the way, there will probably not be more than one group so you don't need to worry about reinforcements. Avoid roads, you are not an invading army fighting from position of strength (or at least equality), you are poor guerrilla trying to survive and outmaneuver enemies at this point.
Re: Experimental Project 7[message #355982 is a reply to message #355981] Sun, 25 November 2018 11:16 Go to previous messageGo to next message
buuface

 
Messages:165
Registered:October 2011
Ok thanks for the advice
Re: Experimental Project 7[message #355987 is a reply to message #355982] Sun, 25 November 2018 18:08 Go to previous messageGo to next message
crackwise

 
Messages:77
Registered:April 2013
@sevenfm: I have tested flashbang indoors on my mercs. They did not have any eye-wear (i.e. no sunglasses or night vision). However they seem to receive stun damage only and none of them got blind.

I am playing AR+SDO with your AI mod. Could it be that the blinding/deafening aspect of flashbangs are somehow disabled? If such, how can I enable it? Thanks a lot in advance!
Re: Experimental Project 7[message #355989 is a reply to message #355987] Sun, 25 November 2018 20:06 Go to previous messageGo to next message
sevenfm

 
Messages:1935
Registered:December 2012
Location: Soviet Russia
crackwise wrote on Sun, 25 November 2018 21:08
@sevenfm: I have tested flashbang indoors on my mercs. They did not have any eye-wear (i.e. no sunglasses or night vision). However they seem to receive stun damage only and none of them got blind.

I am playing AR+SDO with your AI mod. Could it be that the blinding/deafening aspect of flashbangs are somehow disabled? If such, how can I enable it? Thanks a lot in advance!

Fixed in r877.
Flashbang effect checked InBuilding() which only searched the spot for buildings with flat roof (those you can climb) and not just a room with any roof above as it should.
Another problem was that later blindess effect was only applied to person if it was also deafened, not when just blinded.

[Updated on: Sun, 25 November 2018 20:09]




7609+fix | 7609+AI (r1180) | Unofficial modpack | Win8+ fix | Experimental project | Youtube

"It's already "dog-eat-dog", friend. Not sure what worse a bunch of zombies could do."


Re: Experimental Project 7[message #355990 is a reply to message #355989] Sun, 25 November 2018 20:58 Go to previous messageGo to next message
crackwise

 
Messages:77
Registered:April 2013
Thanks a lot sevenfm! I have also noticed a couple of times (while playing r876) that the walking UI showed available movement (i.e. the footstep signs were yellow for running for one additional step) whereas my merc actually did not have enough APs to run anymore. This was out in the open, without any obstacles and such.
Re: Experimental Project 7[message #355991 is a reply to message #355990] Sun, 25 November 2018 21:06 Go to previous messageGo to next message
sevenfm

 
Messages:1935
Registered:December 2012
Location: Soviet Russia
crackwise wrote on Sun, 25 November 2018 23:58
Thanks a lot sevenfm! I have also noticed a couple of times (while playing r876) that the walking UI showed available movement (i.e. the footstep signs were yellow for running for one additional step) whereas my merc actually did not have enough APs to run anymore. This was out in the open, without any obstacles and such.

If you provide a save, I can probably check it when I have time.



7609+fix | 7609+AI (r1180) | Unofficial modpack | Win8+ fix | Experimental project | Youtube

"It's already "dog-eat-dog", friend. Not sure what worse a bunch of zombies could do."


Re: Experimental Project 7[message #355992 is a reply to message #355991] Mon, 26 November 2018 03:29 Go to previous messageGo to next message
sevenfm

 
Messages:1935
Registered:December 2012
Location: Soviet Russia
Another good example of defending against a horde of pistoleros in the early game (video provided by JAggernaut):



Notes:
- a good defending position in the middle of the sector between the trees
- actively moving according to changing tactical situation
- retreating from window position into the inner part of the house when in danger
- using buckshot instead of slugs for hunters with shotguns (helps to hit opponents behind obstacles and suppress them)
- opening doors diagonally

It would be also interesting to see if it's possible to defend sector by taking the lone building with flat roof to the south-east with the whole team.



7609+fix | 7609+AI (r1180) | Unofficial modpack | Win8+ fix | Experimental project | Youtube

"It's already "dog-eat-dog", friend. Not sure what worse a bunch of zombies could do."


Re: Experimental Project 7[message #355998 is a reply to message #355992] Tue, 27 November 2018 03:00 Go to previous messageGo to next message
buuface

 
Messages:165
Registered:October 2011
I still can't help but feeling pistols are just slightly out of whack. I have fox who has 67Mks and ambidexrous and she is by far the most lethal member of my team.

I may be wrong in my intuitions. Just starting to get hold of some higher tier weapons now, so i will continue to playtest and give more feedback.

PS, I had to give up on the previous Iron Man game, it was just too difficult. restarted AR-SDO with save anytime.
I also started WF607 with no other mods enabled and soft iron man, which also seems to be slightly more survivable.

[Updated on: Tue, 27 November 2018 03:06]

Re: Experimental Project 7[message #355999 is a reply to message #355998] Tue, 27 November 2018 03:10 Go to previous messageGo to next message
sevenfm

 
Messages:1935
Registered:December 2012
Location: Soviet Russia
Pistols are good for close range, especially with gunslinger and ambidextrous traits.
For example, if you can shoot with rifle with CTH = 40% and with a pistol with CTH = 20%, and if pistol is twice as fast as rifle and you have two of them and ambidextrous trait that Fox has, you will be roughly twice more effective at pistol range than if you used rifle.
Maybe I need to increase penalties when shooting beyond weapon range, but in general pistols allow much more shots, and you can move with them more like run closer to enemy, shoot, run back to cover.
You can also check merc statistics in the laptop and see who has better chance of hitting.



7609+fix | 7609+AI (r1180) | Unofficial modpack | Win8+ fix | Experimental project | Youtube

"It's already "dog-eat-dog", friend. Not sure what worse a bunch of zombies could do."


Re: Experimental Project 7[message #356000 is a reply to message #355999] Tue, 27 November 2018 05:25 Go to previous messageGo to next message
Hawkeye

 
Messages:1946
Registered:October 2005
Location: Australia
Personally I'd vouch for keeping the pistol modifiers as is. I still found at early stages of the game I could barely hit anything with a pistol, even at close ranges, even two or three tiles away, but have yet to try with a gunslinger. Enemies on the other hand seemed better at shooting, but perhaps Vengeance has slightly different values? Admittedly pistols have produced better results as the game has progressed and my mercs have had better training and improved their skills.


Re: Experimental Project 7[message #356001 is a reply to message #356000] Tue, 27 November 2018 05:46 Go to previous messageGo to next message
sevenfm

 
Messages:1935
Registered:December 2012
Location: Soviet Russia
The game can assign gunslinger trait if soldier has pistol, so they can be more accurate because of added bonus, you can check enemy traits with debug tooltip level.



7609+fix | 7609+AI (r1180) | Unofficial modpack | Win8+ fix | Experimental project | Youtube

"It's already "dog-eat-dog", friend. Not sure what worse a bunch of zombies could do."


Re: Experimental Project 7[message #356003 is a reply to message #356001] Tue, 27 November 2018 13:58 Go to previous messageGo to next message
LatZee

 
Messages:126
Registered:December 2015
Well, realistic solution to that would be to make shooting 2 weapons available only in "from hip" mode, because even if you are truly ambidextrous and therefore equally good with both hands, that doesn't mean that you have eyes on stalks like a snail and can therefore aim down 2 different sights cheeky although I don't think that pistols are all that good and therefore need any balancing
Re: Experimental Project 7[message #356009 is a reply to message #356003] Wed, 28 November 2018 04:04 Go to previous messageGo to next message
buuface

 
Messages:165
Registered:October 2011
I think one reason pistols feel too strong to me is because they are also more effective at suppression than shotguns or SMGs. Because of the AP cost you can fire 4-5 shots where the heavier weapons can only fire one, or with SMGS, one wildly inaccurate burst. I get that shotguns, especially the pump-action ones should be much slower firing than pistols but I would expect the buckshot (and maybe slugs) to do more suppression damage than a 9mm bullet. At the moment it appears that they don't.

Having said this. I've yet to really come up against an enemy wielding anything other than pistols (playing on slow progression), so this is mostly speculative and could be more of to do with the marksmanship of enemies having been improved generally.


[Updated on: Wed, 28 November 2018 04:04]

Re: Experimental Project 7[message #356010 is a reply to message #356009] Wed, 28 November 2018 04:21 Go to previous messageGo to next message
sevenfm

 
Messages:1935
Registered:December 2012
Location: Soviet Russia
Shotguns in stock 1.13 are not very good balanced, I can recommend setting
DAMAGE_SHOTGUN_MODIFIER	= 1.2
SP4T_SHOTGUN_MODIFIER = 1.2
BURST_AP_SHOTGUN_MODIFIER = 1.2
AF_SP5AP_SHOTGUN_MODIFIER = 1.2

in Item_Settings.ini, it will make shotguns more useful as powerful weapon for close range.

In SDO mod, shotguns are much more effective in the early game as they have much higher damage, so it's not unusual to kill enemy with one shot in early battles.

I don't think that pistols are better than SMGs, in my tests of your save game I was able to shoot 2 aimed bursts/turn with a good CTH of about 50% at medium range, if enemy was standing and not moving, so they were quite effective.



7609+fix | 7609+AI (r1180) | Unofficial modpack | Win8+ fix | Experimental project | Youtube

"It's already "dog-eat-dog", friend. Not sure what worse a bunch of zombies could do."


Re: Experimental Project 7[message #356013 is a reply to message #356010] Wed, 28 November 2018 09:45 Go to previous messageGo to next message
buuface

 
Messages:165
Registered:October 2011
Cool, will test that out.
Re: Experimental Project 7[message #356023 is a reply to message #332214] Wed, 28 November 2018 21:40 Go to previous messageGo to next message
CareBear

 
Messages:117
Registered:April 2016
Hi Sevenfm. Have you thought about implementing different accuracy faloff curves for different weapon classes ie. pistols have different curve than rifles. Because right now the settings that are good for pistols wont suit rifles and vice versa. Also would it be realistically possible to code a caliber/ammo based damage system ie. weapons dont have inherent damage value attributed to them, but it is determined by ammo type, ie 7.62x51 will have higher damage than 5.56x45. This would resolve problem with subsonic[cold] ammo having same damage as normal loaded ammo, or 12 gauge slugs having incredible damage, while you balance the buckshot.

[Updated on: Wed, 28 November 2018 21:56]

Re: Experimental Project 7[message #356025 is a reply to message #356023] Wed, 28 November 2018 23:18 Go to previous messageGo to next message
sevenfm

 
Messages:1935
Registered:December 2012
Location: Soviet Russia
CareBear wrote on Thu, 29 November 2018 00:40
Hi Sevenfm. Have you thought about implementing different accuracy faloff curves for different weapon classes ie. pistols have different curve than rifles. Because right now the settings that are good for pistols wont suit rifles and vice versa.

Are you talking about OCTH or NCTH?
In both systems chance to hit quickly drops when shooting beyond weapon range which is the biggest difference between weapons. Pistols have a short range so you can hit with them at distance no more than 15-20 tiles.

Quote:
Also would it be realistically possible to code a caliber/ammo based damage system ie. weapons dont have inherent damage value attributed to them, but it is determined by ammo type, ie 7.62x51 will have higher damage than 5.56x45. This would resolve problem with subsonic[cold] ammo having same damage as normal loaded ammo, or 12 gauge slugs having incredible damage, while you balance the buckshot.

I don't plan any changes that will break compatibility with stock 1.13 7609 and mods. We already have main trunk for that kind of changes.
If you want to change damage for a certain type of ammo, why not just use tags like <armourImpactReductionMultiplier>, <armourImpactReductionDivisor>, <afterArmourDamageMultiplier>, <afterArmourDamageDivisor>, <beforeArmourDamageMultiplier> and <beforeArmourDamageDivisor>?



7609+fix | 7609+AI (r1180) | Unofficial modpack | Win8+ fix | Experimental project | Youtube

"It's already "dog-eat-dog", friend. Not sure what worse a bunch of zombies could do."


Re: Experimental Project 7[message #356027 is a reply to message #356025] Thu, 29 November 2018 00:05 Go to previous messageGo to next message
CareBear

 
Messages:117
Registered:April 2016
Quote:
Are you talking about OCTH or NCTH?
In both systems chance to hit quickly drops when shooting beyond weapon range which is the biggest difference between weapons. Pistols have a short range so you can hit with them at distance no more than 15-20 tiles.


NCTH. Fallof for pistols should be larger than for rifles. Currently no way to do that.

Quote:
I don't plan any changes that will break compatibility with stock 1.13 7609 and mods. We already have main trunk for that kind of changes.
If you want to change damage for a certain type of ammo, why not just use tags like <armourImpactReductionMultiplier>, <armourImpactReductionDivisor>, <afterArmourDamageMultiplier>, <afterArmourDamageDivisor>, <beforeArmourDamageMultiplier> and <beforeArmourDamageDivisor>?


Ive already tinkered with that before making post, but the results are only somewhat satisfactory.
Re: Experimental Project 7[message #356028 is a reply to message #356027] Thu, 29 November 2018 00:19 Go to previous messageGo to next message
sevenfm

 
Messages:1935
Registered:December 2012
Location: Soviet Russia
@CareBear
I think it's the design of NCTH system - aperture increases linearly over distance (which means quadratic CTH drop when using uniform shot distribution), with gun's accuracy adding small penalty which is also probably linear, I didn't look deep into the code.
I plan to review some parts of NCTH code some day, specifically max aperture dependency from visual sight/gun range (which shouldn't be linear) and scope effect (which shouldn't just divide aperture size by a magnification value).



7609+fix | 7609+AI (r1180) | Unofficial modpack | Win8+ fix | Experimental project | Youtube

"It's already "dog-eat-dog", friend. Not sure what worse a bunch of zombies could do."


Re: Experimental Project 7[message #356083 is a reply to message #355991] Sun, 02 December 2018 15:16 Go to previous messageGo to next message
crackwise

 
Messages:77
Registered:April 2013
crackwise wrote on Sun, 25 November 2018 20:58
Thanks a lot sevenfm! I have also noticed a couple of times (while playing r876) that the walking UI showed available movement (i.e. the footstep signs were yellow for running for one additional step) whereas my merc actually did not have enough APs to run anymore. This was out in the open, without any obstacles and such.


Hi sevenfm, here is a screenshot regarding the bug that pathfinding UI does show more "footsteps" than it is actually available:
https://imgur.com/a/Qo8Na3O

And here you can find the savegame file: (AR+SDO with +AI r883)
https://ufile.io/dmokz

[Updated on: Sun, 02 December 2018 15:21]

Re: Experimental Project 7[message #356104 is a reply to message #356083] Thu, 06 December 2018 03:40 Go to previous messageGo to next message
sevenfm

 
Messages:1935
Registered:December 2012
Location: Soviet Russia
Ja2+AI r900

- HandleSuppressionFire: fixed critical bug
- marbles work only for walking/running soldiers (swatting soldier will not fall)
- fixed flashbang effect in buildings, blindness effect (previously flashbang code could only recognize buildings with flat roof)
- fixed finding wrong flagged soldier (seems to be the bug that caused enemy instantly finding mercs when there is enemy tank in sector and causing realtime slowdowns)
- fixes: r7819, r7931, r8643
- better random number generator (it seems that stock 1.13 random is not providing uniform distribution, allowing strange sequences of extremely lucky/unlucky shots)
- fix jumping with backpack cost
- don't allow fence jumping from prone stance using [j] key
- breaking AI deadlock: show AI action name, allow breaking AI less often when fast forwarding enabled
- always allow dropping item at the same tile (allow going into negative APs to prevent situation when merc is stuck with item in cursor and without APs to drop it)

AI:
- allow grenade/mortar attacks more often if sector curfew = 2
- sector leader determines extra flanking or restricts flanking when attacking sector (for the first 10 turns after raising alert only)
- stationary/snipers don't advance if have no cover, try to avoid roof edge
- Black AI: better crouching/turning code when nothing to do
- allow shooting through walls when returning fire or if enemy is in a room and high percent of teammates killed
- AI speed optimization
- minor fixes for AI soldiers changing stance/turning to prevent AI deadlocks

OCTH:
- AllowedAimingLevels: reduce max aim levels by 1 when firing with two pistols
- CalcChanceToHitGun: additional penalty 2%/tile when shooting beyond weapon range, bonus up to 10% if weapon range is greater than distance to target
- always allow 1% when shooting at visible target, for invisible targets only when aiming at torso

NCTH:
- disable increasing deviation when shooting beyond effective range (now deviation formula is the same as in stock 1.13)
- use original stock 1.13 shot distribution code (uniform distribution inside aperture)

[Updated on: Thu, 06 December 2018 03:45]




7609+fix | 7609+AI (r1180) | Unofficial modpack | Win8+ fix | Experimental project | Youtube

"It's already "dog-eat-dog", friend. Not sure what worse a bunch of zombies could do."


Re: Experimental Project 7[message #356105 is a reply to message #356104] Thu, 06 December 2018 08:48 Go to previous messageGo to next message
buuface

 
Messages:165
Registered:October 2011
Thanks for the update!
Re: Experimental Project 7[message #356107 is a reply to message #356104] Thu, 06 December 2018 16:37 Go to previous messageGo to next message
zwwooooo

 
Messages:83
Registered:February 2009
Hi,

I get the error when compiling AI r900. Missing 2 files?

error C1083 "XML_Keys.cpp": No such file or directory
error C1083 "XML_Locks.cpp": No such file or directory
error LNK1181: "Tactical.lib"
Re: Experimental Project 7[message #356108 is a reply to message #356107] Thu, 06 December 2018 16:40 Go to previous messageGo to next message
sevenfm

 
Messages:1935
Registered:December 2012
Location: Soviet Russia
zwwooooo wrote on Thu, 06 December 2018 19:37
Hi,

I get the error when compiling AI r900. Missing 2 files?

error C1083 "XML_Keys.cpp": No such file or directory
error C1083 "XML_Locks.cpp": No such file or directory
error LNK1181: "Tactical.lib"

They are there, you just need to include them in Tactical project if you don't use the same vs2013.sln project that I use for compiling.

UPD: uploaded missing files in r901

[Updated on: Thu, 06 December 2018 17:10]




7609+fix | 7609+AI (r1180) | Unofficial modpack | Win8+ fix | Experimental project | Youtube

"It's already "dog-eat-dog", friend. Not sure what worse a bunch of zombies could do."


Re: Experimental Project 7[message #356138 is a reply to message #356108] Mon, 10 December 2018 03:44 Go to previous messageGo to next message
buuface

 
Messages:165
Registered:October 2011
Hi sevenfm,
I have some more feedback for you after playing your modpack (wf607 mod) a bit longer.

Regarding grenades;

I noticed that stun grenades are ineffective after trying to use them multiple times. They enemy is knocked down after the detonation but are immediately able to stand up in their turn and act as though they have a full AP (move/5-6 shots with a pistol). I realise this may be default behavior for this version 1.13, but i thought i should let you know in case you want to try to fix it.

The same applies to mini-grenades, they seem to do almost no breath dmg to enemies.

The m79 launcher; Though the one i picked up wasn't at 100% quality, I was appalled by how inaccurate short range it was. Used by both Grunty (heavy weapons) and Wolf, both of them were far more reliable and could hit longer ranges with a thrown grenade than with a 40mm from the m79. Have a try and see if the same goes for you.
Re: Experimental Project 7[message #356139 is a reply to message #356138] Mon, 10 December 2018 03:58 Go to previous messageGo to next message
buuface

 
Messages:165
Registered:October 2011
Returning to the interruption system - I understand you usually test with the new interrupt system disabled - and that you have modified the chance-to-interrupt bonus for pistols to be slightly higher within the weapons range. Is that correct?

What I cant get used to is enemies with pistols often being uninterruptible/being able to beat my mercs 'to the draw' even when they come into view with an un-aimed pistol against my merc with an aimed gun (shotgun/smg/other cqc weapon). One hack for this is for my merc to pull out his side-arm and aim it but this feels kind of silly since it's usually more effective (unless down to -20ap) to simply put the pistol away and fire with my main weapon once I win the interrupt.

A possible solution would be to allow the AI to sidestep into view in an aimed stance like the merc is able to do (I don't know if that is possible within the limits of the game-engine, but if so, it certainly must be possible for your coding skills, judging by what else you have been able to achieve already).

Would it also be more realistic and also gameplay-friendly to make the pistol interrupt bonus only apply to um-aimed weapons? I can't see why an aimed pistol would be faster to fire than an aimed p90/mp5 which are specifically designed for close-quarters/room-to-room combat.

Re: Experimental Project 7[message #356140 is a reply to message #356138] Mon, 10 December 2018 04:42 Go to previous messageGo to next message
sevenfm

 
Messages:1935
Registered:December 2012
Location: Soviet Russia
buuface wrote on Mon, 10 December 2018 06:44
Hi sevenfm,
I have some more feedback for you after playing your modpack (wf607 mod) a bit longer.

Hi buuface, thank you for feedback!

Quote:
Regarding grenades;

I noticed that stun grenades are ineffective after trying to use them multiple times. They enemy is knocked down after the detonation but are immediately able to stand up in their turn and act as though they have a full AP (move/5-6 shots with a pistol). I realise this may be default behavior for this version 1.13, but i thought i should let you know in case you want to try to fix it.

I cannot reproduce it, in my tests with SDO+Wildfire, 2 stun grenades (item 131) are usually enough for the enemy to not stand up on the next turn.

Quote:
The same applies to mini-grenades, they seem to do almost no breath dmg to enemies.

Mini grenades have low damage and breath damage, they should not work as stun grenades.

Quote:
The m79 launcher; Though the one i picked up wasn't at 100% quality, I was appalled by how inaccurate short range it was. Used by both Grunty (heavy weapons) and Wolf, both of them were far more reliable and could hit longer ranges with a thrown grenade than with a 40mm from the m79. Have a try and see if the same goes for you.

Cannot reproduce. In my tests with SDO+Wildfire, I had the following results:
Igor (no traits):
mini grenade: 17 tiles, 35% cth
mk2 grenade: 14 tiles, 42% cth
M79: 16 tiles, 42% cth or at max range 35 tiles, 28% cth
Grunty (heavy weapons):
mini grenade: 16 tiles, 39% cth
mk2 grenade: 13 tiles, 47% cth
M79: 16 tiles, 58% cth or at max range 35 tiles, 44% cth
Barry (demolitions):
mini grenade: 19 tiles, 62% cth
mk2 grenade: 15 tiles, 72% cth
M79: 16 tiles, 45% cth or at max range 35 tiles, 31% cth

As you can see, M79 has better cth at grenade range, unless your merc has demolitions skill, and m79 has much higher range than any grenade.
Also, merc trait is not the only thing that matters, you also need regular skills for good cth:
basic CTH for throwing (grenades, throwing knives): (dexterity + marksmanship) / 2
basic CTH for launched (mortars, grenade launchers): (dexterity + marksmanship + wisdom + 10 * level) / 4
Effective skills are used (so for example, suppression shock can lower effective marksmanship and dexterity).
Later, other modifiers are applied like morale/fatigue etc.

I can look at particular situation if you provide a save, but currently I don't see any problem in thrown weapons balance.
I will slightly lower range penalty for calculating CTH with grenade launchers, but that's all I can do at the moment.

Quote:
Returning to the interruption system - I understand you usually test with the new interrupt system disabled - and that you have modified the chance-to-interrupt bonus for pistols to be slightly higher within the weapons range. Is that correct?

Raised weapon in raised direction adds +1 interrupt level, if target is within weapon range.
If soldier has pistol in hand, and target is within pistol range, add +1 interrupt level.

So max possible difference between your merc and enemy soldier with pistol will be 1 level if you aim in his direction.
If your merc has interrupt level 5 and enemy has interrupt level 6, that means that 54% of the time he will win interrupt, and 46% of the time you will win. I don't see any problem here.

Quote:
What I cant get used to is enemies with pistols often being uninterruptible/being able to beat my mercs 'to the draw' even when they come into view with an un-aimed pistol against my merc with an aimed gun (shotgun/smg/other cqc weapon). One hack for this is for my merc to pull out his side-arm and aim it but this feels kind of silly since it's usually more effective (unless down to -20ap) to simply put the pistol away and fire with my main weapon once I win the interrupt.

There are no uninterruptible enemies in the game, you always have a chance for interrupt, unless:
- enemy can see you but you cannot see enemy
- your merc was under attack this turn (he was suppressed and changed stance, just a stray bullet that caused no reaction doesn't count)
- your merc was hit this turn (with bullet, fragment, grenade etc)

Quote:
A possible solution would be to allow the AI to sidestep into view in an aimed stance like the merc is able to do (I don't know if that is possible within the limits of the game-engine, but if so, it certainly must be possible for your coding skills, judging by what else you have been able to achieve already).

AI cannot use alternative movement mode (moving backwards, sidestepping, rolling), and I'm not sure if it's possible technically to teach them.

Quote:
Would it also be more realistic and also gameplay-friendly to make the pistol interrupt bonus only apply to um-aimed weapons? I can't see why an aimed pistol would be faster to fire than an aimed p90/mp5 which are specifically designed for close-quarters/room-to-room combat.

Yes, I will change that so the game will first check for raised weapon and check for pistol bonus only if weapon is not raised.

[Updated on: Mon, 10 December 2018 04:46]




7609+fix | 7609+AI (r1180) | Unofficial modpack | Win8+ fix | Experimental project | Youtube

"It's already "dog-eat-dog", friend. Not sure what worse a bunch of zombies could do."


Re: Experimental Project 7[message #356161 is a reply to message #332214] Thu, 13 December 2018 08:28 Go to previous messageGo to next message
buuface

 
Messages:165
Registered:October 2011
Thanks for the reply.

A couple of questions:

I noticed to the Vengence:Reloaded link in your sig; does this mean the current version of V:R has your AI modifications integrated already?

Do you have any plans to update your mod in line with some of the newer development versions of 1.13 in order to take advantage of some of the new features. It seems like a lot has been added to the trunk since 7609 from flugente's workshop and other places.

In fact it would make a lot more sense for them to incorporate your AI mod into the 'official' version of 1.13. Has this been talked about?
Re: Experimental Project 7[message #356162 is a reply to message #356161] Thu, 13 December 2018 08:42 Go to previous messageGo to previous message
sevenfm

 
Messages:1935
Registered:December 2012
Location: Soviet Russia
buuface wrote on Thu, 13 December 2018 11:28
I noticed to the Vengence:Reloaded link in your sig; does this mean the current version of V:R has your AI modifications integrated already?

No but I applied a few fixes to the Vengeance code. The AI in VR is the same as in stock 1.13, as it is based on stable 7609 code with some improvements.

Quote:
Do you have any plans to update your mod in line with some of the newer development versions of 1.13 in order to take advantage of some of the new features. It seems like a lot has been added to the trunk since 7609 from flugente's workshop and other places.

No, the idea is using stable code base to improve AI and implement some experimental features that I find interesting.

Quote:
In fact it would make a lot more sense for them to incorporate your AI mod into the 'official' version of 1.13. Has this been talked about?

Not possible for many reasons, the code is too different. It would be easier to make new AI from scratch than to merge Ja2+AI into the trunk.



7609+fix | 7609+AI (r1180) | Unofficial modpack | Win8+ fix | Experimental project | Youtube

"It's already "dog-eat-dog", friend. Not sure what worse a bunch of zombies could do."


Previous Topic: [WIP] Extended Music Mod Alpha
Next Topic: New Attachment System Beta
Goto Forum:
  


Current Time: Thu Sep 19 11:44:03 EEST 2019

Total time taken to generate the page: 0.02746 seconds