Mod: UnitGen - Random nations with procedural sprites!

[elmokki]

http://koti.kapsi.fi/elmokki/dom3/un...26_2_12-v2.zip

I might release a v3 today if I feel productive. This includes magic paths for primary mages and one kind of secondary mage magic path set that's duplicated to every single secondary mage at the moment.

Also an easy .exe file for windows users.

[elmokki]

http://koti.kapsi.fi/elmokki/dom3/un...26_2_12_v3.zip

18 minutes before it would've been 27_2_12.

It actually REMOVES secondary mage paths as they were bugged as hell. It has a complete rewrite of primary mage paths.

[elmokki]

http://koti.kapsi.fi/elmokki/dom3/un...en_27_2_12.zip

Derp, now it should finally be bug free. Also generates all possible primary mages of certain path set per nation.

[elmokki]

Primary mages can be now generated with following patterns for magic:
3-3-?
3-2-1-?
3-2-?
3-1-1-?
3-1-2?
3-2? (linked)
3-2?
3-?
2-2-2? (linked)
2-2-2?
2-1-2? (linked)
2-1-1-2? (linked)
2-1-1-2?
2-1-1-1?
2-1-1-1-?
2-3? (linked)
1-3? (linked)
1-1-1-2? (not in 27-2-12 version)

In general randoms are such that only one path can reach level 4 and none of the paths can reach level 5. Exceptions to this is the linked 2-2-2? which allows both of the two paths to be level 4. Current version generates one of every setup for test, but in the future some will be rarer than the others.

I just went through secondaries and tertiaries of MA and I believe this is a fairly comprehensive list of levels possible for secondaries and tertiaries. Do tell if you can think of one more to these or primaries! One setup is listed only once even if it is available for multiple nations.

Sec:
1-1 - MA Tien'chi - caponly primary (7 picks)
2-1-1 - MA Pythium - caponly primary (7 picks)
1-2? - MA Man - caponly primary (6 picks)
2-1 - MA Marignon - caponly primary (6 picks)
2-1-? - MA Ashdod - caponly primary (3 picks), special
2-1-1-? - MA MAchaka - caponly primary (6 picks)
2-1-1 - MA Agartha - caponly primary (4 picks)
3 - MA Abysia - caponly primary (6 picks), special
2? - MA Ashdod - caponly primary (3 picks), special
1-1-1-? - MA Eriu - caponly primary (7 picks)

Tert:
1 - MA Pythium
1-? - MA Man
? - MA Tien'chi
1-1-1 - MA MAchaka
2 - MA Abysia

The 2? ones will probably get linked and non-linked versions and the ones where randoms would give level 3 will probably have versions where they can and where they can't.

[elmokki]

Nation naming, especially Lizard one, was pretty crappy far too often so I rewrote it.

Now nation name files consist of three files. Long syllables (misleading name), short syllables and suffixes. Nation name format is: Long syllable + 33% chance of short syllable + suffix. Idea is to simply have tons of entries for especially long syllables and suffixes to make nations varied enough.

You can also define very precisely what a name part can be after or before. Syntax is for example:
test "cantbebefore consonant a" "cantbebefore consonant e" "derp"
which would make the name part unable to be generated before anything that starts with:
- any two letters where first letter is an consonant and the second a
- any two letters where first letter is an consonant and the second b
- derp

The keywords that contain multiple single letters are consonant, vowel, softconsonant and hardconsonant. Here's what they contain:
String[] vowels = { "a", "e", "i", "o", "u", "y", "å", "ä", "ö", "ü" };
String[] consonants = { "b", "c", "d", "f", "g", "h", "j", "k", "l", "m", "n", "p", "q", "r", "s", "t", "w", "v", "x", "z" };
String[] softconsonants = { "f", "h", "j", "l", "m", "n", "r", "s", "w", "v", "x", "z" };
String[] hardconsonants = { "b", "c", "d", "g", "k", "p", "q", "t" };

So basically if your limitation is "hardconsonant vowel" you get every single combination where the first letter is a hard consonant and the second letter is a vowel. If your limitation is "hardconsonant derp" you get: bderp, cderp, dderp, gderp, kderp, pderp, qderp, tderp. So yeah, you should be able to limit generation far more than you will ever need to.

My test files are written white humans in mind, but here goes a set of 50 names and with heavy recycling of current name parts, especially suffixes, a great deal of which are by kianduatha:

Ondnuiu, Duxsoum, Emia, Arcoheim, Dogegard, Doror, Cunvia, Ishnom, Swedod, Gilmitha
Lunbalstan, Warmon, Wuhullum, Nefland, Lamsorea, Helmon, Ermdod, Norgaor, Faeurnia, Lejiheim
Etten, Ishvik, Krophale, Funangea, Gaeilna, Pyjaland, Burvavik, Caance, Ugliburg, Gomovo
Netaimer, Moran, Eren, Agth, Vanmon, Warjais, Oben, Kritia, Narmark, Hamheim
Ashugal, Olaance, Erbogundy, Karnoiraine, Faeengmark, Osaany, Dorni, Ungmia, Mortium, Elegard

I won't implement these to the nation generation itself before I have sets for all the three current races, but NameGen.jar will generate names based on longsyllables.txt, shortsyllables.txt and suffixes.txt at UnitGen root directory so you can write lists yourself and test them easily and submit good lists for UnitGen releases!

[elmokki]

Magic site generation works and should actually not lump every command in one site and continue to next. It'll try to combine gem commands to mages with suitable paths or units with tags of being associated to a magic path (some sacreds definitely will be such)

Site naming will get a better rehaul than it did earlier later too. I want to get rid of stuff like "Melancholic Office", which basically means I'll add tag "epic" to certain name parts and force at least one epic tag per name.

Mage generation has gotten tweaks, but it isn't ready. When it is, it'll have a ****ton of balance problems (in this context: mage setups that are horribly powerful) when I consider it ready enough too and those will have to be sorted out.

Then what's left to do is in order of effort taken:

- Fix cavalry generation
- Tweak chariot generation slightly
- Fix troop naming
- Good mage naming
- Add elite generation
- Proper commander generation that generates commanders based on what a nation needs. Basically this means that nation with flying recruits gets at least some sort of a flying commander, nation with stealthy troops gets at least one stealthy commander etc. This does not mean that a nation with a flying sacred stealthy amphibian gets one commander with all of those features though (though in case of sacreds it's probably likely)
- Priest generation
- Generation of funky special abilities for units (ranging from more Ulmish stats to stealthiness to cold resistance and stuff like that) based on if they are eligible for them and the rng-gods favour the nation (all members of one race in a nation would probably get the stats, but stealth or berserking is probably just for specific lightly armored units)
- Add sacred generation
- Add custom weapon generation for sacreds
- Allowing multi race nations (well, honestly, probably less terrible than I expect it to be since I have a fairly good idea of how it can be done fairly easily, but it needs some rewriting of item generation)
- Nation/unit descriptions (this is potentially huge loads of work, but also in general generating names / descriptions is most fun)

I probably miss something too.

[elmokki]

Regarding mage generation. I had one too complex system, current one is less complex in one way but still too complex in another.

Question is, what sort of mage patterns ignoring mages that aren't really part of the official hierarcy (Fortune Teller for LA Ulm or Nahualli for Mictlan as examples, but LA Marignon sailors don't suit) should be possible?

The lingo here would be that:
- School is a collection of mages of about the same available paths.
- Primary mage is the top tier mage of a country. Generally having 4 to 7 picks with the rare exception of 3? linked pattern. Maximum level reachable with 100% randoms is 4, but most of the possible pick patterns only one path can reach 4.
- Secondary mages are well, weaker versions, having 2 to 5 picks or so. Maximum level with randoms is 3, but quite a few patterns won't reach that as opposed to primaries where majority can get that one level 4. Pretty much all nations will probably have at least one secondary of some kind, but it might be gone from nations with enough options already (in base game Mictlan doesn't have a school-relevant secondary mage in my eyes)
- Tertiary mages are rarer. They range from 1 to 3 picks with maximum of 2 in a path. I would imagine they would happen mostly in cases with total mage count being rather low or no secondaries existing.
- Pattern: distribution of paths, for example 3-2-? refers to what MA Marignon's grand master has, 1 path at 3, 1 path at 2, 1 100% random.

In general the goal is to generate about maximum of 5 mages from schools for a nation, with a possible extra non-school mage (mentioned above) or a minor secondary school (LA Marignon sailor mages could be such, or for that matter LA Ulm S/B-mages if they had a major mage there too).

For school mages I see following options mostly.

3 primaries, all cap only, either one secondary or one tertiary or both.
3 primaries, at most one cap only, either one secondary or one tertiary or both.
Both cases have fairly specialized mage patterns with 2, possibly sometimes 3 paths available, probably the exact same pattern for all, except in the non-cap only case if there's one cap only variant, that might be slightly more powerful. Magic paths will vary for primaries like they do in EA Agartha or MA Mictlan - one path varies, rest stay the same. These are pretty mucn EA Agartha and EA Mictlan in terms of ingame versions.

2 primaries, 1-2 secondaries, possibly a tertiary. 0-2 cap only primaries. Either the primaries work pretty much exactly like in the 3 primary cases with same or very similiar pattern and a bit different paths or they can get a very different pattern, either with very different paths or not. Regardless of the case a nation will not get more than 6 paths and even then in those cases patterns with many randoms won't happen so you're going to be left with low path amounts. With one secondary it's a mix of the two path setups if they are different, with two they are both mini versions of the mages.

Of this setup you can find probably less in the game. I guess MA Jotunheim would suit, Gygja has 5 picks and so does Skratti, but patterns and paths are quite different. MA Tien'chi and MA Pangaea have two primary mages too obviously. Pangea and Tien'chi have rather similiar pattens for both of them.

Lastly one primary setups. They can have the "traditional" 1 primary, 1 secondary, 1 tertiary setup with paths just getting weaker and weaker as tier goes down, or they could scrap the tertiary. Primary being cap only or not is fairly arbitary too. With that few mages you are bound to get an unrelated bonus mage with different paths too. As a bonus you could get 2 or 3 secondaries instead of just one, all being variations of a cap only primary. Once again, either by varying the paths or varying the patterns.

Holy picks or not are a different matter handled elsewhere.

Did I miss something?

[elmokki]

Mage generation with a single primary is looking fairly good. Most of the code can go to three or two mages with fairly little edits - probably anyway - so those should get ready soonish.

I'm going to rewrite the dom3db-reader (direct copy from Randomocalypse, just set to read armor/weapon data instead of unit data) a bit so I can inject UnitGen-generated equipment in it without causing issues. It MIGHT work even now actually, but it's a fairly small change in how stuff is stored anyway so I'm going to do that. I can't be arsed to start work on sacred generation before that is done since seriously, getting all sorts of funky weapons (and possibly some armor too) is half of what excites me in generating sacreds. That said sacred generation will contain a boring mode that copies stats and sprite from existing sacreds sometimes instead of actually generating a sacred. Hydras at the very least deserve to get a shot at getting generated here too.

As pointless as descriptions are for actual gameplay, I'm fairly sure I'm interested enough to code them in anyway, and since this just sounds plain awesome the design principle will be adjusted according to that (results are, however, not guaranteed to be in line)

17:00 < kianduatha> one last thing before i go for real this time: My totally unrealistic dream description thing is to have a chance for linked histories between nations(if it makes sense). So one nation would be like a former slave state of some other empire
17:02 < kianduatha> "Falchions were the only weapon heavy enough to cut through Thuringia's Heavy Knights. The days of the revolt are long past, but Forinth's Palace Guards still wear the falchion proudly in remembrance of that time" is mydream procedural description

...but descriptions will have to wait for now.