Questionable priorities

Found a bug? Great! Post about it here!

Re: Questionable priorities

Postby Ferret » Tue Mar 07, 2017 5:53 am

Sorry it took me a bit to remember to check into this :P

The Infection thing is, as I suspected, a bit messy. The solution, oddly enough, may be to change Infection itself to just be "no healing" :P That will give the AI a pretty clear cut choice on how to handle it, no? :D Not sure I want to do that yet, but AI is always tricky for situations where it has to weigh "short term triage" (heal through Infection because OMG you're gonna die!) and "long term benefit" (healing it first, even if it means you have lower HP for a little longer because it didn't heal you instead)

As for the SP cost, I looked at the numbers and I admit I winced a little.

Here's how that apparently works.

1) If an ability and target combo's evaluation is less than 30000 (for reference, 30000 is a really, good evaluation, usually requiring multiple targets, healing a near-dead ally, or damaging a near-dead enemy to reach), and the ability user has at least half their health...

2) ..deduct 50% from the evaluation if the ability will drop you below maxAbilityCost * 2 Stamina...

3) ..deduct 75% (total, doesn't stack with #2) from the evaluation if the ability will drop you below maxAbilityCost Stamina.

I can easily see a 50% penalty being enough to override the difference in damage your Lilim would be getting with the situation you outlanded. These numbers are probably a bit too high, and I'm going to experiment with adjusting them. Thanks for the report. :D
User avatar
Ferret
 
Posts: 1785
Joined: Tue Sep 16, 2014 3:18 pm

Re: Questionable priorities

Postby oneeyedjoe » Tue Mar 07, 2017 4:24 pm

Glad to see that you have time to deal with silly bugs while working on a huge update with a bunch of new content. Btw found another one for you :P Why do demons often misuse execution light/dark skills (don't remember their names, they do 130 power attack if damage would be lethal and miss otherwise). So yeah, most of the time demons miss with that skill.
oneeyedjoe
 
Posts: 205
Joined: Tue Feb 14, 2017 4:30 pm

Re: Questionable priorities

Postby Ferret » Tue Mar 07, 2017 5:07 pm

The threshold for the AI attempting to use one is 50%+ projected success rate... but there is a lingering and quite annoyingly persistent bug that makes the AI think using them on Guilt/Pariah'd targets is an awesome idea. :P This one is unfortunately tabled until after this build though: given the previous attempts to fix it have whiffed, a proper fix is probably going to take longer than I can justify on a build already 8 weeks old for a non-crashy bug.
User avatar
Ferret
 
Posts: 1785
Joined: Tue Sep 16, 2014 3:18 pm

Re: Questionable priorities

Postby wizzzargh » Tue Mar 07, 2017 10:52 pm

I think infection disallowing healing completely is a pretty nifty idea. I doubt anybody ever said 'Zombie/Chindi/Acheri/Druj Nasu are too strong and keep killing me!'
wizzzargh
 
Posts: 353
Joined: Tue Oct 27, 2015 5:31 pm

Re: Questionable priorities

Postby Ferret » Tue Mar 07, 2017 11:14 pm

I think it has been said of Chindi and Druj Nasu at least, but not really because of Infection I admit. :P
User avatar
Ferret
 
Posts: 1785
Joined: Tue Sep 16, 2014 3:18 pm

Re: Questionable priorities

Postby Remora » Thu Mar 09, 2017 1:09 am

I'm not so sure about that. I feel that infection is fine as it is, and if paired with poison it can be very strong in killing enemies.
Remora
 
Posts: 85
Joined: Thu Sep 22, 2016 6:27 am

Re: Questionable priorities

Postby Ferret » Thu Mar 09, 2017 1:17 am

Infection changes wouldn't be in the upcoming build anyway even if I do decide to do that, just to clarify: this build's already got "too many" balance changes for me to comfortable with as it is, but that was inevitable for the first batch of Tier 3 content so I'm just going to have to live with it. :P I don't need to go making it worse though. :D
User avatar
Ferret
 
Posts: 1785
Joined: Tue Sep 16, 2014 3:18 pm

Re: Questionable priorities

Postby Bormoth » Sun Apr 09, 2017 2:45 am

Ferret, Have you tried a 3gate AI aproach

gate 1: are hard conditionals, (Ally should die I can kill one of sources of death, than kill(check if ally would die anyway, if yes than try heal) and such. Some just critical interactions. (If nothing applies go to gate 2.)
gate 2: are criticals, they are going in strict order(or could be weighted but go in their own pool) of their critical approach(remove infection and desease(Concern here that stun might prevent you from saving yourself but deasease usually just locks you out of preservation), remove deadly dots, remove disables and charms, remove dots or other low occurance debuffs (Poison ++ or poison +, shocked) (All that wouldn't kill you, but need to be dealt ASAP) (If if nothing goto gate 3)
gate 3: is where most flexibility is. Tagged weighted system(each skill has tags, so if tags with target(s) correspond you add weight depending on targets. Which you seem to use.(which you seem to use to great success I like how it works.)

This system probably would be able to make miracles a bit more usefull because they could be predicted unless you decide them to be commandable.
---
And second I noticed strange behavior while using lucky charm and lucky aid. (My zaltys tends to use lucky charm instead of preserving SP. 25% chance to gain random buff for 25SP doesn't seem like good justification to use it instead of any other ability or just preserving SP for later use if there are, no luck based abilities on cooldown in your team(but there are luck cooldown based abilities just not on cooldown). (maybe it is just me, and I just imagining things, but this could help AI not to overthink, if it happens so) I mean Lucky aid every other turn(almost always either SP recovery or heal or status recovery) is more precious than 25% chance to get buff.

Also I dunno if it is already done I would be glad if lucky aid prioritised targets which have (heal other SP costed ability, on chance it might recover SP to your healer. Could lead to strange interaction with Draw wounds, but it can give extra weight based on how much healer with draw wounds is injured(After all now you can get 66% chance to make healer more usefull again either through draw wounds or through SP healing. ) When there is no need to save anyone from death or maybe even when HP is not critical strict order.
If previous condition is false you might weight how much it might help (if target is low SP or low health, it is better target than just low SP. Maybe low health targets should be weighted a bit more. But in general I enjoy this skill, saved my ass so many times, and preserves several slots being earliest heal you can get.
Bormoth
 
Posts: 32
Joined: Sat Apr 08, 2017 4:51 pm

Previous

Return to Bug Reports

Who is online

Users browsing this forum: No registered users and 3 guests

cron