Getting Fallen Empires draft to work

[lujo]

What would be better implementation of pick order? Perhaps a two-dimensional system where value of any card could be increased or decreased by the specific cards you've already picked?

I'm still pondering it, and there's several ways to go about it, but I think one of the better ones means that the array of attributes which define a card for draft (and also limited deckbuilding) in a particular set needs more fields.

So that would enable, for example, cards to have tags like "splashable", so a card like Derelor or Icatian Phalanx could be picked but not cause other cards to go up and down in pick priority based on their color. Then you can have a tag called "dud", which would prevent a particular card from ever going up and down, so a card like Icatian Moneychanger would never get priority unless everything in the booster is also "dud". Then you can have a "quantity X" tag which would make that particular card go up or down depending on how many copies the AI already has - Goblin War Drums and Goblin Kites are like that.

Then you could have the triggers which make particular cards more appealing depending on whether you have another card, or which makes certain commons a priority if you early-pick a certain rare or uncommon. Deep Spawn is always an early pick, but if you have one or two, getting 1-2 High Tide becomes a priority.

Then you could have a tag which define a cards color identity (already probably present) so that, say, Dwarven Ruins would go up or down depending on the players color, even though they're a land.

And so on and so forth. There's another issue:

A major problem with a linear set-wide pick list is that not all colors have an equal distribution of bombs / key cards / highly efficent creatures at all rarities. This is, I think, why most set reviews don't use an unified pick list for the entire set, but separate it into Rares / Uncommons / Commons. In a vacuum, if a color has one or two highly efficient commons, it's easier to get locked into it early, and a universal pick list is going to be skewed in favor of those colors resulting in many people fighting over them. This is probably bad for the AI.

If im reading things correctly, the bestiaire list causes the AI to check the booster for certain black commons before it even checks for key rare and uncommon stuff in other colors, which leads to the situation where it will often go for a black common unless it opens one of the White rares. Which probably leads to a situation where every booster is checked for black commons before anything else, which then causes 3 things:

1) The colors are more often drafted one by one as more people prioritize black not because it's the strongest color but because it has the highest positions on an universal list
2) The AI is more likely to get stuck in black (causing legit bombs from other colors to drift around the table as low picks even if the AI can use them), as picking up black cards further increases the likelyhood of picking up more black cards.
3) All the fallout that comes from that, such as blue always being open to the player because the highest pick on the list is an unplayable rare, and it's key cards are all below a lot of commons and the AI's getting 4 color decks because they "fought over" the same colors in every booster, one color at the time, sequentially.

I'm still thinking about how to properly arrange cards in tiers where the AI would consider important rarer cards before the important commons. In any case making a more sensible universal list first is the way to go for now, even if it just means separating all the duds and moving stuff up and down, seeing how that works, then moving onto set-specific limited deckbuilding tweaks, and then moving forward.

EDIT: And ofc, getting as many cards in a set properly implemented for the AI as possible, and getting the AI to be able to put most of them in a deck. This means there'll probably be a bunch of bug reports and AI requests and questions about FE cards for a while, but that's actually a good thing. It just ends up with forge having more properly implemented cards and a more draftable set. It's like a "second pass" on general card implementation

I think. XD

[lujo]

So, the next batch of cards on the placeholder "whole set list" should probably look like this (for now):

The "sac" lands:

Info | Open

If these could be drafted "on color" especially, prioritizing them here, if not higher, is a good idea. They're more reliably useful than most of the stuff below. I'm not sure about the order. Having the AI use them to accelerate into key bombs would be enough.

#60|Havenwood Battleground|U|FEM (=> Feral Thallid)
#61|Ruins of Trokair|U|FEM (=> Hand of Justice / Icatian Town)
#62|Ebon Stronghold|U|FEM (=> lots of stuff)
#63|Svyelunite Temple|U|FEM (=> Homarid Warrior / Deep Spawn. Nice to pay for High Tide)
#64|Dwarven Ruins|U|FEM (=> Orgg, Dwarven Catapult etc.)

One-off stuff:

Info | Open

These are things which you do want one of the appropriate decks, but no more than that, and sometimes not even that many. Ideally they someone who already has one of these should probably rate these cards as bottom of the heap and only play one copy.

#65|Tidal Flats|C|FEM (One of if you allready have a decent blue deck going, but no more than one)
#66|Elven Fortress|C|FEM (Same as above, it's pretty good, but no more than one)
#67|Goblin Grenade|C|FEM (If you have no goblins, you don't need a single one. The chance of having enough goblins to support even one is slim, but one is worth it)
#68|Night Soil|C|FEM (This is a constructed sideboard card, and not entirely awful, but more than one is overkill.

General on color filler:

#69|Basal Thrull|C|FEM (These guys are more key than they look, have plenty of sinergy with other stuff, but are still a 1/2 for BB and shouldn't be picked too high. They have little usefulness outside of black, so I put them here to try to avoid other colors picking them too highly)
#70|Farrelite Priest|U|FEM (This guy is a mediocre wall with no combat applications, but ok support for the pump guys and will hold some agrro down)
#71|Icatian Scout|C|FEM (Not that bad, just not a priority and the AI shouldn't overstuff their deck with them)
#72|Icatian Infantry|C|FEM (Same as above, even some sinergy with stronger white cards, but are still at the bottom of the playable heap due to the AI not handling banding too well)
#73|Vodalian Soldiers|C|FEM (Cheap, barely playable blue fodder)

Filler Artifacts:

These are here as an alternative to the niche/duds

#74|Conch Horn|U|FEM
#75|Implements of Sacrifice|U|FEM

Really Narrow Stuff:

#76|Thrull Wizard|U|FEM (The only point of these guys is the creature type, and possibly color hosing, so they're only really good in CERTAIN black decks, where basal Thrulls are still better. But they're not completely unplayable, and should be played in mono-black)
#77|Vodalian Mage|C|FEM (These guys go from dud to playable depending on whether you have Homarid Spawning Bed and possibly 2 Vodalian War Machine)
#78|Initiates of the Ebon Hand|C|FEM (These guys are better left to the player, although having them in black/x wouldn't hurt the AI. Still, a 1/1 for B isn't what I'd encourage the AI to play, but if the AI could play them with pump knights it'd be good to have one or two)
#79|Tidal Influence|U|FEM (This is... It's not bad actually, and it kills you really dead with Homarid Spawning Bed or even Deep Spawns, but the AI would need to know how to handle it, and the wording is awful. Timing it is also REALLY tricky as you can easily kill yourself with it if you're only blue for Homarid, Homarid Warrior and River Merfolk)
#80|Orcish Captain|U|FEM (Can still win any game and might even be fun to play against, but for now let's leave him safely here.)
#81|Elvish Scout|C|FEM (Also, not a bad card, but probably not higher untill the AI can actually use it well, vanilla 1/1 for 1 isn't good.)
#82|Spore Flower|U|FEM (Not bad at all, I'd have to be sure the AI can use it before I'd put it higher, though)

#83|Delif's Cube|U|FEM (Interesting card, but would need AI attention before it should really be picked too high.)

[lujo]

Ok, the question of why Fallen Empires of all things has been raised, and in my reply in the Card AI thread I forgot what is probably the most important reason (or rather I haven't explained it well enoguh).

The oldest sets can't rely too well on the bestiaire lists for many reasons. This leads to all sorts of things, but the biggest problem with badly drafted Fallen Empires, for example, is that AI vs AI matches devolve into endless ground stalls. This makes playing FE draft (and in Forge there are plenty of reasons to do so), a painful experience - the game slows down as the AIs struggle to keep track of all the possibilities and activations and you have to sit there cursing yourself for even attempting it or cursing the game for making you have to win a FE draft to get a token to play the draft you really want.

And what happens then is that there's some random gamecrashing bug with one of the cards which makes the whole experience 10 times worse as the matches take forever, and if an AI draws a card like that - bam, thoroughly screwed, restart the whole endless grind again. Which then makes the whole thing unbearable.

This is not necessarily so because FE block games have to devolve into ground stalls, but more so because the AI isn't drafting, putting into decks or using cards which are there to prevent this sort of thing (and are often bombs and archetype cornerstones). So fixing a more recent set would be more about getting the AI to draft fancy stuff and make the draft experience deeper, this is about making the draft experience BEARABLE.

And I chose FE, in this regard, precisely because the difference between a well drafted and well played FE draft and a terrible one is unbelievable, so I can clearly see improvements if there are any. Everyone's got their favourite sets they'd like to be infinitely draftale with a ton of skill and variance, but these actually need the most help as they're the furthest from being justifiably called "implemented". And I wanna get them there first.

Ok, so I'll put up the next batch of updates, I've reconsidered some choices so far because I forgot to take into account some junk rares (Elvish Farmer, I'm looking at you), and I'm getting to the stuff that's supposed to be drafted over them.

[lujo]

I'm not too sure on the ordering of these particular lists and we're in the teritory where the universal list concept slowly falls apart, but it's still guaranteed to work better than the current one, so let's experiment with this for now


Utility Uncommons & Some Rares:

#43|Spirit Shield|U|FEM (It's rare, it's an equipment and is usable by anybody and adds a "playable" to any deck.)

#44|Icatian Priest|U|FEM (It's rather good, but it's not too splashable and picking up too many leads to too many 1/1's in a deck)

#45|Breeding Pit|U|FEM (Is killer with thrulls, but leads to ground stalls, locks up mana and most stuff you can use it with should be picked higher as it CAN work without Breeding Pit, while the Pit is quite lacklustre without other stuff)

#46|Soul Exchange|U|FEM (This is actually the only Pit-friendly thing which you don't want the AI playing without the pit too often. There is a turn 3 Deep Spawn combo with Dwarven Armorer in the set, and I'd love to make the AI able to draft it but that's for a much more advanced stage. With proper AI adjustment, this does work rather ok with green, though.)

#47|Dwarven Lieutenant|U|FEM (You do want to pick other dwarves above him, but if you do, you want to check packs for him before other stuff)

#48|Thallid Devourer|U|FEM (If you're green, this gus is what you check the pack for before other things, could be higher.)

#49|Balm of Restoration|U|FEM (This is better than it looks, but should only be used as to save creatures from dying and is rare so you check for it here)

#50|Homarid Shaman|U|FEM (This is actually mostly worth picking up due to sinergy with HSB and random blowouts against green guys.)

#51|Rainbow Vale|U|FEM (I'm not sure where to place this, it could go much lower, but it oddly seems to randonly work out agaisnt all reason, due to allowing AI to cast stuff in 4 color decks)


Important utility commons:

Category Info | Open

Ok so these are commons you want 1-2 off in your deck, but being commons and not worth stuffing your deck with too much, and you generally don't want AIs picking these over other stuff, splashing them or moving into colors because of them.

#52|Goblin War Drums|C|FEM (This is key to breaking ground stalls with proper use, splashable and all that, but also a common common and 2 per deck max.)

#53|Hymn to Tourach|C|FEM (Lol wut? But yeah, it's a BB card and a 4 art common, you want a few in a black deck, but you don't want AI's to go black because they opened a Hymn to Tourach, or passing decent creatures and ending up with 4-5 Hymns and a load of 1/1 vanilla dorks.)

#54|Goblin Chirurgeon|C|FEM (1 or 2 in any red deck is priceless, but you don't really want more even if it would allow for Goblin Grenade, as it's more of a "Seal of Regeneration" in disguise than a creature. Should probably not be counted as a creature while deckbuilding)

#55|Combat Medic|C|FEM (He's a bomb, but also a common and you never really need more than one on the board to completely ruin the other guy's day)

#56|Merseine|C|FEM (This NEEDS to be implemented. But it's a 4 art common, so you'll always get the several you need)

#57|Spore Cloud|C|FEM (This is a huge blowout combat trick, but you don't want more than 2-3 in a deck. Off color people shouldn't be picking them up and moving into green because of it)

#58|Armor Thrull|C|FEM (He's great, but also a 4 art common "seal in disguise" you don't want the AI overloading on, or forcing black for.)

#59|High Tide|C|FEM (This is not really useful outside of Deep Spawn blue and would need a block friendly AI tweak. No rush to pick them up over other useful stuff)

[drdev]

I must say, I'm very impressed by the amount of effort you're putting into this project. Keep up the good work.

[lujo]

^ Ty and I hope I can do a fine enough job to persuade you to give implementing Merseine a try. Coming up with a nice "Forge Coupon" to put into boosters in place of unimplemented cards (Such as Raiding Party in FE) would also be appreciated to no end. I know I keep restating this, but if people are reading these posts and appreciate the effort it might help seeing those suggestions accompanying the usual read to give it some credibility and an incentive for a coder to jump in and help this whole thing out at least with that.

All right, getting really controversial here.

Here's the meat-and-potatoes of FE draft, the commons which should form the backbone of any FE limited deck. The ordering is particularly problematic, as is the entire concept of having a "big ole set list", as these guys can really pile up to influence what colors someone ends up in. I'm thinking my reasoning is flawed for now, and that maybe a lot more uncommons and rares should be higher than these guys, but then the AI would be getting too many "niche" or "clever" cards and too few beaters and the player could snap these up and always win.

I'll try the following first, ordering them in one "one color each" groups of 5 with different emphasis, and I'll test how what works out for the AI. I'm sure it's not gonna work out too well, but I have to start somewhere and this is valuable research for finding a robust system which can then be applied to other sets in the future. This thing really is the granddaddy of modern sets, moreso than anything up to invasion, tbh.

Most of these guys being 3-4 art commons does mean a good idea might be to move them, or some of them, down below a bit as with 5 (maybe should be 6, really) boosters you can always sorta catch up on them. If that attribute array for draft cards is made a real thing some of these guys should get optimal numbers past which they drop down once you've got a nice core of playables.

The best common dudes:

#22|Order of the Ebon Hand|C|FEM (Main reason being that they're immune to Icatian Javelineers and can gain first strike against most other stuff, as well as block white bands)

#23|Icatian Javelineers|C|FEM (They ARE Icatian Javelineers, and can kill a bunch of stuff on top of being a white creature for Hand of Justice and a soldier for Icatian Liutenant. Mogg Fanatic, if you will, reprinted in Time Spiral, to boot. If you ever wondered why there's so many 0/2 guys in this set, well, there you have it. "Doesn't die to Icatian Javelineers" used to be a thing.)

#24|Brassclaw Orcs|C|FEM (They're actually a deadly beater with Goblin War Drums, Goblin Kites, great when buffed and prevent Orgg from attacking. And laugh at Icatian Javelineers. The AI does need to be agressive and very willing to trade them in combat as they trade well)

#25|Homarid Warrior|C|FEM (The "big guy" of commons, kills anything, can be powered out with any sac land, takes 3 colored mana to be killed by pump knights, and turns into 5 tokens with HSB. Pump this guy via black and he's a beast, band him with a white bander he's a beast, etc.)

#26|Thorn Thallid|C|FEM (Has GG in his cost, and could use some AI adjustments, and looks slow as hell, but combos with Fungal Bloom better than anything, kills anything and playing as many as possible if green is a good idea. He also kills Pump Knights. And Icatian Javelineers.)

The slightly less best common dudes:

#27|River Merfolk|U|FEM (Ok, this guy isn't common, he's RARE, and he has some evasion, but he's here because... he dies to Icatian Javelineers and Thorn Thallid and combos less well with Homarid Spawning bed. Otherwise, a fine, fine aggro pick, just not something you see and immediately decide you're blue. If you ARE blue, you do want him as many as you can get.)

#28|Order of Leitbur|C|FEM (The reason this guy isn't in the first list is because he dies to Icatian Javelineers, and, well, isn't one.)

#29|Necrite|C|FEM (Getting this guy through is a mite difficult, and the AI doesn't seem to like putting him in decks, but he's a 2/2 for 3, a thrull, and if he does go through he's removal. Gotta get the AI putting him in decks somehow, though.)

#30|Dwarven Soldier|C|FEM (Now, he does die to stuff, but he's really aggressive and if you snap up the Dwarven Liuetenant and possibly the Armorer to make him not die to pings, he can dish out a crapton of damage. All that with Goblin Kites is often 5 damage per turn if you have enough of these guys to Fling, or if you keep wining coinflips.)

#31|Thallid|C|FEM (He does die easily and doesn't beat face, but he's a common source of saprolings, and there's plenty of uses for those. Also keep in mind that green likes to sit behind Elven Fortress and Spore Cloud a lot, while it pings you and stacks forests for Thelonite Druid, once you take that into account the value of this guy becomes more apparent. Soul Exchange, Thallid Devourer, Thelonite Druid, Ebon Praetor - they all love the tokens, and if I could get the AI to use Thelonite Monk to lock you out of colors in FE draft that would be very cool.)

The rest of the guys:

#32|Farrel's Zealot|C|FEM (This guy will straight up lock you out if he starts going through. The only reason Order of Leitbur goes above this guy is that it blocks Black bombs, and he doesn't, but a case could be made that spamming banding + this guy is the strongest deck in FE, as you can't block the band, and you can't NOT block the band. Funny thing is, this guy was actually playable in constructed, believe it or not.)

#33|Elvish Hunter|C|FEM (He keeps stuff locked down, and with some AI work he's extra good when supported by Spore Clouds. Green removal, but I'm not sure overdoing them is a good idea, thy're 1/1 and tie up your mana. Still, he can keep 2 dudes tapped down for a long time)

#34|Orcish Veteran|C|FEM (Whether this guy is better than Brassclaw orcs is debatable. He can attack into first strikers, but he can't attack into as many bombs as Brassclaw Orcs can or combo with anything like Dwarven Warriors. Still, no reason not to pick him up most of the time.

#35|Mindstab Thrull|C|FEM (Recently I've begun to suspect this guy is better than Necrite. If you get him in early you can really mess someone up with a 3 for 1 on turn 4.)

#36|Homarid|C|FEM (Silly wording and awful on the turn it comes into play, but not actually bad, all in all. A 2/2 when you untap with him, a 3/3 on the next turn, a 2/2 after that and if he's not HSB'd by then you're doing something wrong)

Here's the list of rares worth checking packs for after the choicier commons have been snapped up (or possibly before, I'm sure I'll reconsider stuff after few whirls).

T2 Rares:

#37|Elven Lyre|U|FEM (My current line of reasoning is that this can be played in any deck, so it probably should be checked for first.)

#38|Draconian Cylix|U|FEM (Same deal, and it can be pretty brutal, could possibly go higher or lower. I'll have to try out different arrangements with emphasizing different stuff, it'll take a lot of testing.)

#39|Vodalian Knights|U|FEM (This is actually a wall in effect and flying is only relevant against blue (no other fliers in the set, and it can't attack anyone else). It is, however, a fine little defender to snap up if you've got a few more versatile blue cards already, but by no means a true incentive to go blue. With HSB even blue crap becomes a white rare. Icatian Town, to be precise.)

#40|Icatian Lieutenant|U|FEM (Well, I'm not actually sure this goes here - white has a bunch of soldiers, so this looks like an obvious build around. The thing, however is, that you don't have to "draft around" this thing if you're white, so you might as well snap it up later than some other stuff. Not sure.)

#41|Dwarven Armorer|U|FEM (Eh, I really need to see if there's a point of this being this high at all. Still, you can pump it with Dwarven Liuetenant, and it's the only discard outlet for Soul Exchange, and I wonder if it's best to always let the player get it for now, until some AI work can be done on it. The AI should be able to put a +0/+1 counter on 1 toughness creatures because of all the annoying pinging.)

#42|Tourach's Gate|U|FEM (This is... This is a build around card. It needs some AI work, as it can basically be a simple one time +2/-1 attack boost for some decks, or a more sustained buildaround for several black/X archetypes. I'm not sure if this should be higher/lower of what, but I'll test and see)

[lujo]

Top Rares:

Info | Open

The ordering of these isn't exactly important, as they can't be present in the same booster, so it just means tha AI will pick them up before other stuff. I didn't bother with ordering them until I figure out what exactly the order of snap-pick rares supposed to achieve.

#1|Aeolipile|U|FEM (Because it's removal which goes into any deck and gets rid of a variety of stuff)

#2|Zelyon Sword|U|FEM (Being able to get anything to be dangerous is big)

#3|Fungal Bloom|U|FEM (This lets you power out Thallids)

#4|Ring of Renewal|U|FEM (Card draw, slow, but steady and can go into any deck. More than one is not necessarily good)

#5|Hand of Justice|U|FEM (The biggest rare in white, and should steer tha drafter towards white)

#6|Icatian Town|U|FEM (This is highly splashable, and combo's well with Goblin War Drums and several other things)

#7|Orgg|U|FEM (This is better than it looks especially if the AI can power it out with sac lands.)

#8|Derelor|U|FEM (Was played as Ernham Djinn 5-8 in constructed decks for a long time, this is rather great but should not steer the drafter towards black)

#9|Ebon Praetor|U|FEM (This is the black bomb, needs a bit of precise AI but is otherwise sort of unkillable in the format)

First Pick Uncommons:

Info | Open

The ordering of these, however, is exceedingly important, as there's a good chance the AI will pick one over the other. As gectrix has noted in the card AI thread, getting the AI to not always choose deterministically from this list would probably be beneficial. The temporary ordering, below:

#10|Homarid Spawning Bed|U|FEM (Up to two copies, this should be snapped up).

#11|Thelonite Druid|U|FEM (So should this, and if the AI can use it well, should go a long way to preventing endless ground stalls)

#12|Deep Spawn|U|FEM (Second good reason to go blue, a deck can support up to 3, oddly enough.)

#13|Farrel's Mantle|U|FEM (This + River Merfolk = you destroy red, + Goblin Flotilla = you destroy blue, + anything with banding = you probably destroy anything as you probably can't block the band and you can't NOT block the band. I think I would've evaluated banding differently had I been making the list from the top, but more on that when I round up the archetypes.)

#14|Feral Thallid|U|FEM (The big green beater, very strong, but does have trouble due to low toughness and is a bit slow rolling without Fungal Bloom. Havign guys with banding helps it on offense).

#15|Dwarven Catapult|U|FEM (This is removal. The AI should pick it, and use it agressively as it doesn't resolve a ground stall, but it does wonders to prevent one if used on one or 2 creatures)

#16|Icatian Phalanx|U|FEM (This blocks anything, is nigh-unkillable, and escorts guys with Farell's Mantle, Farrel's Zealot, Necrite or Mindstab Thrull on suicide missions, on top of being splashable into pretty much anything.)

#17|Thrull Retainer|U|FEM (It's a bit strange to see this so high, but it's sort of priceless as it allows you to be more agressive and keeps your dangerous stuff alive.)



Ok, so that's a first attempt, and there's a lot of stuff I'm not happy with, which should change after a bit of playtesting, when and then I'll nail down the archetypes, apply findings to improve the list, diagnose deckbuilding issues, jot down the in-match play issues and outline the first draft of properties for card evaluation arrays.

[lujo]

Ok, the AI is definitely less keen on passing playables but it's deckbuilding NEEDS work, some cards desperately need to be enabled, some need their AI adjusted/made, some stuff has to get the priority arrays. 5 out of 7 decks build were terrible even though the AI had fine cards to make much better decks, some deckbuilding moves are hilarious (really you'll laugh when you see them, they're actually funny), and one guy drafted a perfect mono-black thrull deck, all of which ended up in his sideboard in favor of an unplayable 4 color deck o.O

The good news is that it's deffinitely drafting stuff more appropriately, I'm seeing Dwarves, even (one guy had 3 Dwarven Warriors and a Liuetenant in his deck, which is very refreshing). Plenty of stuff to reorder, but first thing is to get the AI to use everything it can without inconvenitencing any coders TOO much, and build decks the way you're supposed to.

I'll post the decklists and a bit of analysis shortly, but first, please, anyone - how can I disable whatever is preventing the AI from putting cards it can't use very well into decks, so that I can actually see what that looks like and where it would be easy to fix? At least having insight into this is crucial.

[Sloth]

I'll post the decklists and a bit of analysis shortly, but first, please, anyone - how can I disable whatever is preventing the AI from putting cards it can't use very well into decks, so that I can actually see what that looks like and where it would be easy to fix? At least having insight into this is crucial.

You can edit the card scripts directly. The following line in the script will prevent the AI from putting the card into a deck:

Code: Select all

SVar:RemAIDeck:True

EDIT: And by the way. I've enabled the AI to use Deep Spawn today.

[lujo]

I'll post the decklists and a bit of analysis shortly, but first, please, anyone - how can I disable whatever is preventing the AI from putting cards it can't use very well into decks, so that I can actually see what that looks like and where it would be easy to fix? At least having insight into this is crucial.

You can edit the card scripts directly. The following line in the script will prevent the AI from putting the card into a deck:

Code: Select all

SVar:RemAIDeck:True

EDIT: And by the way. I've enabled the AI to use Deep Spawn today.

^ Very cool, ty! I was going to point out that he was one of the easy candidates! Gonna be interesting (Hope I finally get to have my ass kicked by a humongous lobster, that's worth trying the set out at least once XD)

Also, I spent better part of the afternoon getting to know eclipse and the way cards are scripted. I think I still couldn't build a version, but I could do the simplest things, and noticed that bit of code

Can that be edited without having to rebuild or whatever it is I have to do? Is this code stored in txt files where you can edit it safely?

(If someone told me I was going to be learning how to use java so I can port Merseine of all things into a simulation of MtG at the time I first saw that card I would've called him crazy, and with good cause XD)

EDIT: Oh, and could you also allow Necrite? The AI seems to be way too eager to sacrifice Mindstab Thrull every single time it goes though, and it plays Mindstab Thrull. If those two could check if the opponents hand had at least 2 cards in it (Mindstab Thrull) and that their target wasn't a token or a 1/1 or a 1/2 (Necrite) they'd both be pretty legit. Heck, they'd be legit if the AI never even used their abilities.

Also, I'm not seeing Zelyon Sword used, and it's basically equipment, and the AI can use those, right? I mean, it's 2nd pick, I sure hope it can.

For some strange reason I have a feeling somoene also allowed Derelor lately, as I've had a bout recently where everyone played him and that never happened before

[Sloth]

Can that be edited without having to rebuild or whatever it is I have to do? Is this code stored in txt files where you can edit it safely?

Each card should have a txt file in res/cardsfolder/. If you have a release version you have to unpack the big zip file there and delete the zip file, or edit the files in the zip file (otherwise your changes will not be recognized).

[lujo]

Awesome! I'll make notes on how the cards behave, this is gonna be so sweet! Always wanted to be a MtG dev (of some sort ^^)

Ok, ok, sorry for keeping you away from more pressing things, but I've come up with this for Merseine so far:

Code: Select all

Name:Merseine
ManaCost:2 U U
Types:Enchantment Aura
K:Enchant creature
A:SP$ Attach | Cost$ 2 U U | ValidTgts$ Creature | AILogic$ KeepTapped

T:Mode$ ChangesZone | Origin$ Any | Destination$ Battlefield | ValidCard$ Card.Self | Execute$ AddNetCounters | TriggerDescription$ CARDNAME enters the battlefield with three net counters on it.
SVar:AddNetCounters:AB$PutCounter | Cost$ 0 | Defined$ Self | CounterType$ NET | CounterNum$ 3

S:Mode$ Continuous | Affected$ Creature.EnchantedBy | AddHiddenKeyword$ CARDNAME doesn't untap during your untap step. | CheckSVar$ X | Description$ Enchanted creature doesn't untap during its controller's untap step if Merseine has a net counter on it.
SVar:X:Count$CardCounters.P1P1

A:AB$ RemoveCounter | ValidPlayer$ EnchantedController | Cost$ ??? | CounterType$ NET | CounterNum$ 1 | SpellDescription$ Remove a net counter from CARDNAME.

SVar:Picture:http://www.wizards.com/global/images/magic/general/dehydration.jpg
Oracle:Enchant creature
Merseine enters the battlefield with three net counters on it.
Enchanted creature doesn't untap during its controller's untap step if Merseine has a net counter on it.
Pay enchanted creature's mana cost: Remove a net counter from Merseine. Any player may activate this ability, but only if he or she controls the enchanted creature.

That would need NET counters added somewhere maybe, and I kinda pulled bits off Cocoon and Festering Wound and Dark Depths and probably some other stuff. If by some miracle this works it'd still need to be able to calculate the cost, could you possibly help me out with that, or at least maybe point me in the right direction?

Oh, and it'd need art and a proper place and stuff...

And god knows what else I messed up. I'm notoriously bad with coding. Only coding I ever did was Fallout 2 stuff

[lujo]

Sorry, cleaned it up a bit, posted leftovers from Festering Wound and complications, apologies.

So what I'm unsure of is whether it would count the counters right, whether NET counters need to be defined somewhere else and how to calculate the exact cost. And whether I used the right way to even put the counters on, as for example dark depths uses only:

Code: Select all

K:etbCounter:ICE:10

So it would be:

Code: Select all

K:etbCounter:NET:3

Except I'm not sure how that would interact with the enchant trigger for it being an aura.

I suppose this looks hilarious to someone who actually knows how to do these things right.

[Marek14]

Well, I guess SVar:X:Count$CardCounters.P1P1 should be SVar:X:Count$CardCounters.NET or something. But of course the main problem is Cost$ ???. Maybe use EnchantedManaCost like Elemental Resonance which also refers to full mana cost of the enchanted permanent? If that doesn't work, Back from the Brink might help.

[lujo]

Well, I guess SVar:X:Count$CardCounters.P1P1 should be SVar:X:Count$CardCounters.NET or something. But of course the main problem is Cost$ ???. Maybe use EnchantedManaCost like Elemental Resonance which also refers to full mana cost of the enchanted permanent? If that doesn't work, Back from the Brink might help.

Ok, so there are 3 problems, not one.

This part is obviously fine, and works | Open

Code: Select all

Name:Merseine
ManaCost:2 U U
Types:Enchantment Aura
K:Enchant creature
A:SP$ Attach | Cost$ 2 U U | ValidTgts$ Creature | AILogic$ KeepTapped


This should be a replacement effect or something, and probably not use the stack which it currently does: | Open

Code: Select all

T:Mode$ ChangesZone | Origin$ Any | Destination$ Battlefield | ValidCard$ Card.Self | Execute$ AddNetCounters | TriggerDescription$ CARDNAME enters the battlefield with three net counters on it.
SVar:AddNetCounters:AB$PutCounter | Cost$ 0 | Defined$ Self | CounterType$ NET | CounterNum$ 3

Then this part works, thanks to changing the counter type to NET. | Open

Code: Select all

S:Mode$ Continuous | Affected$ Creature.EnchantedBy | AddHiddenKeyword$ CARDNAME doesn't untap during your untap step. | CheckSVar$ X | Description$ Enchanted creature doesn't untap during its controller's untap step if Merseine has a net counter on it.
SVar:X:Count$CardCounters.NET

-
-

But then here, whatever I try I can't get it to get the creatures manacost:

Code: Select all

A:AB$ RemoveCounter | ValidPlayer$ EnchantedController | Cost$ Mana<B B B> | CounterType$ NET | CounterNum$ 1 | SpellDescription$ Remove a net counter from CARDNAME.

I've tried various things, but maybe I'm messing the syntax up or something because it SHOULD work somehow if Elemental Resonance and Back from the Brink work, that means you can get the correct mana cost from a creature, and this should not be the problem, except I can't seem to figure it out.

And it's really weird, because I've had any immaginable ammount (including "1 snow mana" ???) appear as the cost except the casting cost of the enchanted creature o.O

The big problem is actually this bit:

Code: Select all

ValidPlayer$ EnchantedController

I've tried it out with changing the cost to a fixed number, and casting it on the other guys creature, but the other guy couldn't use it, and I still could.



-------------------


In other news, having enabled all the cards which should be interesting to the AI, and disabled a bunch which were getting into decks for no good reason, the improvement in deckbuilding is clearly visible - I just had a draft pod with 5 out of seven AI decks in 2 colors! And with pretty sensible cards, too.

There's still a crapton of work that needs to be done, but I have a better idea about what needs to be added to the deckbuilding and card code, and I've spoted holes which can use fixing. Plus, actually getting most of this stuff to work right can and should be used on other sets eventually, as getting some of these guys truly playable by the AI would make a ton of other guys work, too.

Making notes and trying to get Merseine in.

Also - figuring out how to change the draft format for FE to 6 boosters to try that out would help a bunch. I has it's pro's and cons, but it would help the AI decks not unreasonably splash due to a lack of playables.

Also, in the vein of "Disable of AI" tags, the tags I'm pretty certain about ATM which need to be in Forge are "Splashable"/"Nonsplashable", "LimitedColorIdentity" for nonbasic lands (as the AI tends to put any in there), and "Draft"/"RndDeck" + "Limit", which would make the AI only consider drafting or putting a limited ammount of a certain card in.

That's a bare minimum, the basic stuff. Maybe some is allready in and noone applied it to Fallen Empires, but that's absolutely necessary.

--------

Oh, and lol, now I've got a nonfunctional Merseine in the draft, gonna try a pod, see how it goes, and see if it's getting all the 4 arts (so far I've only seen one).