[research] iwd2 xp calculation cont.

[lynx]

As most will know, iwd2 doesn't grant fixed xp amounts for killing nasties. The amount is calculated using the creature's challenge rating (stored in-place of the xp value) and mean party level. The problem is that the table with the mapping starts with cr values of 0.25 and 0.5, while the creature files all contain just plain integers. Even taking that into account and just shifting your lookup, you still have to divide the resulting xp by 2 to get the intro goblins to die for 37xp.

So, some data mining is required in the original. To keep it simple, start with a non-hof party with no ECL characters and never multiclass. Just play until you get near some goblins and save there. Then just repeat a version of the following:
- load game
- cluaconsole in extra xp to level-up the same pc, leaving others alone; http://gemrb.org/iesdp/appendices/clua/iwd2.htm
- kill a goblin
- note pc level and xp gained
- repeat (levels 1, 4, 7 and 11 will cause good changes in the party average)

Then repeat by killing nasties with higher challenge rating. The closest one is Phaen of the Rags in the next area.

For perfection, also chars with ECL could be tested, since it looked like this was a bit buggy in the original. And multiclassed chars, since averages of averages are not usually the same as direct averages (iwd2 has a levelsum stat though and probably uses that). But this is beyond the initial test.

[PracticalKat]

Ran a quick test on Targos docks. Have a party of 4 non-ECL characters. The one is raised from Level 1 to 4 to 7 to 11 and the encounter rerun, staying a pure cleric. So far the XP for one goblin remains at exactly 37. Have cross-checked the reported XP gain on screen with the XP change for each char.

 

Am I doing something wrong? Testing IWD2 install from GOG, installed under PlayOnLinux. V2.01 (101615)

 

Have installed the widescreen patch as well as several of the IWD2Fixpack patches, but nothing that should affect XP rates.

[lynx]

Thanks!

 

I picked those values with a party of 6 in mind, but it's ok, you just increased the mean faster. Looking at the table, the values are fine — they match up.

 

Try leveling an additional pc to 11 and another to 9, so your mean will rise to 8. You should get 33 xp then.

With:

- 3x11 and 3: 25xp

- 3x11 and 7 (or anything higher): 12xp

[PracticalKat]

Got back to testing again. Levelled as suggested: pure classes, non-ECL races, levels 11, 11, 9 and 1 to give average level 8. XP per goblin is now 25. Can attach my spreadsheet with detail before and after per character if that helps.

 

Now to test with Phaen..

 

1. Same party members as above all at Level 2. Then XP for Phaen's house is 823.

 

2. Level distribution of 11, 11, 9 and 2 gives XP for the same encounter as 600.

[lynx]

great, the goblin actually matches — upon review I messed up the prediction by a row. If you didn't install the xp extender, a higher mean should make you get no xp for them at all.

 

Phean has CR of 6. Your mean of 2 and 823 makes no sense (tabled under 1350=2*675; dividing you get 0,611111). :S Is this just him or also the goblins and quest xp?

 

At mean of 8, it matches the value in the table, except it isn't halved anymore. Perhaps the first two (0,25 and 0,5) cols are special in that regard. Or low party levels.

 

Try some more means, this is getting really interesting! Thanks

[PracticalKat]

1. great, the goblin actually matches — upon review I messed up the prediction by a row. If you didn't install the xp extender, a

higher mean should make you get no xp for them at all.

 

2. Phean has CR of 6. Your mean of 2 and 823 makes no sense (tabled under 1350=2*675; dividing you get 0,611111). :S Is

this just him or also the goblins and quest xp?

 

 

At mean of 8, it matches the value in the table, except it isn't halved anymore. Perhaps the first two (0,25 and 0,5) cols are

special in that regard. Or low party levels.

 

Try some more means, this is getting really interesting! Thanks

 

1. Great

 

2. This is everything: Phaen, goblins plus quest XP. I have auto-pause on, but still it gets tricky separating things out. Is there a switch to enable detailed logging in the ini file?

[PracticalKat]

Ran the docks goblin kill again as you suggested, 3 chars at Level 11 and one at Level 3. Also 3 chars at Level 11 and one at Level 7. Both times get 12 XP for one goblin.

[lynx]

Only xp for Phaen himself is important, I can't do much with composites. There is a hidden option for extra logging, but it doesn't print xp. You'll just have to pause.

 

Do you have a moncrate.2da in override?

[PracticalKat]

Only xp for Phaen himself is important, I can't do much with composites. There is a hidden option for extra logging, but it doesn't print xp. You'll just have to pause.

 

Do you have a moncrate.2da in override?

 

 

XP for Phaen only, parties as posted yesterday at 3.47pm:

1. Same party members as above all at Level 2. Then XP for Phaen only is 300.

2. Level distribution of 11, 11, 9 and 2 gives XP for Phaen as 56.

 

 

Yes, I have moncrate.2da in override folder.

[lynx]

That's odd, but we have different tables, so please attach that 2da to a post. Rename it to .txt if the forum will give you grief.

[PracticalKat]

I installed the mod to avoid ever getting no XP. File attached as .txt

moncrate.txt

[lynx]

hmm! I guess the goblins were too good to be true. For you to get 300/56 xp at those levels, the table suggest Phaen was at CR 4 ("2", not 6) and the 11+11+9+2 rounded to 10 instead of 8 or 9 (8,25). My model is therefore significantly flawed.

 

I can't think of anything that would explain the apparent CR drop*, but the second thing is explainable by use of median (10 exactly) instead of mean. Please try with 11,11,9,5 and 11,11,9,9. For both the median stays the same as before, but mean increases by 1 until it matches. A 11,11,9,15 reverses the pattern and would provide the decisive nail.

 

*he's not embedded in the area, no difficulty variable spawning, so the raw CR value is likely massaged before use

[lynx]

Testing with the AddExperiencePartyCR() action gives results as initially expected. CR maps to columns with an offset of 2 and xp is halved like everyone agrees.

[PracticalKat]

Results for Phaen for class combos:

 

Class Mix: Phaen XP:

 

11,11,9,5 => 112

11,11,9,9 => 112

11,11,9,15 => 56

Edit: Figures correct now. Originally misread from spreadsheet

[lynx]

ups, the last one was supposed to be 11,15,9,9 to be certain. Looks like it does use median though.

 

Do you have any CRE file that starts with 11 in override?