Angband.oook.cz
Angband.oook.cz
AboutVariantsLadderForumCompetitionComicScreenshotsFunniesLinks

Go Back   Angband Forums > Angband > Vanilla

Reply
 
Thread Tools Display Modes
Old July 11, 2018, 16:31   #11
Sideways
Swordsman
 
Join Date: Nov 2008
Posts: 318
Sideways is on a distinguished road
Part of the problem with randarts is that the correspondence between an artifact's "power" and how strong that artifact actually is isn't very good; so a randart with nominally moderate "power" can actually be silly-powerful, and the game doesn't realize how game-breaking the artifacts it generates are. The main reason Pos-likes have better randart balance is that the item scoring system works better.
__________________
The Complainer worries about the lack of activity here these days.
Sideways is offline   Reply With Quote
Old July 11, 2018, 23:34   #12
Nick
Vanilla maintainer
 
Nick's Avatar
 
Join Date: Apr 2007
Location: Canberra, Australia
Age: 53
Posts: 7,222
Donated: $60
Nick is on a distinguished road
Quote:
Originally Posted by fizzix View Post
1) Finish the set of features to go into 4.2, right now presumably this is dominated by the new classes and spell reworks, but I'm not sure if Nick plans to do a monster rework for 4.2 also. If this is the case, then rebalancing should occur after the monster rework.
Agreed. I was considering releasing before the monster rework, but I think actually all this stuff needs to be done at once because the bits need balancing against each other.

Quote:
Originally Posted by fizzix View Post
2) Start a discussion about how to improve randart generation. One of the recent moves was to allow stuff like different numbers of rings/amulets/artifact lights etc. However, this has a few problems, notably that you can get really strong gear for slots that are typically not so strong. I don't understand the randart generation code, and I don't know how to fix it.
I suspect (and hope) that it's mainly that I've been too generous with the power distribution that I'm drawing randart powers from; it's possible that Sideways is correct too, and that the power algorithm needs some work. There is also the problem that Estie in particular has brought up about artifacts getting too many little boosts. I think I have a reasonable handle on this.

Quote:
Originally Posted by fizzix View Post
3) Increase monster density later in the game.
Yes; I will keep this in mind when redoing monsters.

Quote:
Originally Posted by fizzix View Post
4) Reduce vault spawn rate (I believe this has already been done, but it might need to be reduced even further).
Maybe done as part of a rethink of vaults/template rooms.

Quote:
Originally Posted by fizzix View Post
5) Reduce quality of drops through the ego/artifact promotion rates

6) Penalize out of depth items more. Based on some initial stat results, it looks like OoD items can spawn much more commonly now. We should look and adjust that function to make it more of a rare occurrence rather than once every 3-4 levels.
Seem plausible given the facts.

Quote:
Originally Posted by fizzix View Post
7) Consider weakening some standarts
Some adjustment of artifacts is in order (again, Estie has expressed this well).

Quote:
Originally Posted by fizzix View Post
My personal preference is to use 3.4 as a point of comparison since that is the last version that we did a dedicated rebalance for. I'd also be ok with taking the lead on 3, 5 and 6. I'm less comfortable with 2 and 7.
Convenient, because I have plans already for 2 and 7 (and maybe 4)

Quote:
Originally Posted by fizzix View Post
I do think the key is to get any major feature upgrades for 4.2 up and running before doing the rebalancing. I also think it's ok for 4.1.x to be an easier version of the game, just like 3.1 and 3.2 were super easy. It's the price we pay for new features.
Yes, yes, yes, this is all exactly correct. Thanks, excellent post.
__________________
One for the Dark Lord on his dark throne
In the Land of Mordor where the Shadows lie.
Nick is offline   Reply With Quote
Old July 11, 2018, 23:59   #13
Pete Mack
Prophet
 
Join Date: Apr 2007
Location: Seattle, WA
Posts: 5,014
Donated: $40
Pete Mack is on a distinguished road
Be careful when you squash out of depth objects. The suggestion of an exponential tail is *much* too strict. 1/x (logarithmic) tail is what you want--basically 50 and 100 levels OOD should be roughly as likely, as should 5 and 10 (or 1 and 2.) The way to get that is an O(1/x) probability distribution, and a logarithmic expectation value.

For example: if you use base 2, the expected depth for an OOD object is 1+7= 8, with the odds of it being in depth range 51-100 = 1/14.
Pete Mack is offline   Reply With Quote
Old July 12, 2018, 01:55   #14
luneya
Adept
 
Join Date: Aug 2015
Posts: 235
luneya is on a distinguished road
Quote:
Originally Posted by Pete Mack View Post
@Grotug--much of the difference in character power is randarts vs standarts. But even with standarts, a 4.1 character would be stronger.
Do we have any useful information on the average power of equipment drops in standarts, as regards the comparison of 3.4 vs 4.1 vs feature branch? It seems to me that if there's a balance issue in standarts, we need to get that fixed first, and then worry about how to balance randart generation.
luneya is offline   Reply With Quote
Old July 12, 2018, 02:33   #15
Pete Mack
Prophet
 
Join Date: Apr 2007
Location: Seattle, WA
Posts: 5,014
Donated: $40
Pete Mack is on a distinguished road
Fizzix did an analysis that showed a 17% difference in total artifact count over 100 levels. However, u am not sure which variant has more, and I am not sure about which has more powerful ones in vaults. The ones carried by monsters don't really count unless you are assuming (a) people are clearing levels and (b) monster count per level has been properly adjusted. By and large I only kill about 5% of all generated monsters (looking at Fizzix numbers and monster kill counts in recent games.)

To do the analysis properly, you need to analyze ladder posts by variant, and normalized by turn count somehow.
Pete Mack is offline   Reply With Quote
Old July 12, 2018, 02:40   #16
Derakon
Prophet
 
Derakon's Avatar
 
Join Date: Dec 2009
Posts: 8,549
Derakon is on a distinguished road
Quote:
Originally Posted by luneya View Post
Do we have any useful information on the average power of equipment drops in standarts, as regards the comparison of 3.4 vs 4.1 vs feature branch? It seems to me that if there's a balance issue in standarts, we need to get that fixed first, and then worry about how to balance randart generation.
I don't think we can have that analysis unless we trust our power rating system, and it's not entirely clear we do.
Derakon is online now   Reply With Quote
Old July 12, 2018, 02:53   #17
Pete Mack
Prophet
 
Join Date: Apr 2007
Location: Seattle, WA
Posts: 5,014
Donated: $40
Pete Mack is on a distinguished road
@Derakon--
Surely we have a sufficiently accurate power model to distinguish Dwarven Green DSM, say, from Caspannion? I used the former in the endgame twice in 3.4. Never, in 3.5 and later.
Pete Mack is offline   Reply With Quote
Old July 12, 2018, 04:02   #18
jevansau
Adept
 
Join Date: Jan 2009
Age: 59
Posts: 158
jevansau is on a distinguished road
When working on the randart generation code, it would be good to also look at the max level choices.

Particularly for object types with low average base powers, the max level should be much higher.

Conversely, for weapons and maybe body armor, the max level should tail off quicker.

I think min levels are less of an issue.
jevansau is offline   Reply With Quote
Old July 12, 2018, 04:49   #19
fizzix
Prophet
 
Join Date: Aug 2009
Location: Madison, Wisconsin, US
Posts: 3,001
fizzix is on a distinguished road
For OoD items my initial inclination was to do some form of a/(mx + b) where x is the level difference and the rest are constants. It will take a bit of effort to fit it into the item generation formulation, but it shouldn't be that bad. And with stats generation we can ensure we're getting roughly the behavior we want.
fizzix is offline   Reply With Quote
Old July 12, 2018, 05:08   #20
Pete Mack
Prophet
 
Join Date: Apr 2007
Location: Seattle, WA
Posts: 5,014
Donated: $40
Pete Mack is on a distinguished road
@Fizzix-- that'd work fine. It's softer near 0 than pure 1/x behavior, but asymptotically the same. That makes sense- the difference between 1 and 10 levels OOD really doesn't matter, so they should be fairly uniform.

Edit: I went and looked at your data again for speed rings. This is an excellent proxy for all OOD items. And the real problem (compared to v3.4) seems to be between dl 40 and dl 75, not very early in the game. (Of course it is hard to tell on a linear plot; semilog would be easier to compare.)
Pete Mack is offline   Reply With Quote
Reply


Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)
 
Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump

Similar Threads
Thread Thread Starter Forum Replies Last Post
Noob - gear question Czurin Vanilla 18 June 5, 2015 14:53
I'm having a hard time having a feel of my gear. Amnekian Vanilla 3 April 20, 2014 16:07
Gear Issues Monkay AAR 1 February 25, 2012 18:32
Gear Spells UglySquirrell Vanilla 7 September 27, 2011 19:22
Best gear for end-game priest? Zikke Vanilla 19 August 4, 2009 21:29


All times are GMT +1. The time now is 17:42.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2018, vBulletin Solutions Inc.