Nascent variant

[danaris]

Greetings.
I hope I'm not violating any protocols with this post; if so, please enlighten me and I shall endeavour to rectify my mistake!

I've been working on a variant on and off for a while, and it's still very rudimentary, but I feel like it's starting to have enough "shape" to it that I'd like to share it with the community and start getting feedback, if only on the basic ideas I've got for it.

Thus far, here are some of my thoughts for what I'd like to do, some of it still pretty sketchy, and very little final. Please note that they are based on my observations of the current state of Vanilla, which I recognize is in turn based on my own play style. Other people may quite like the aspects of Vanilla that annoy me, and that's fine; it's just a different style of play. There may also be some things I've overlooked that would mitigate the things that annoy me, and if so, I'm happy to hear about them!

First of all, I feel like the current state of Vanilla has pretty much reversed the common D&D complaint of "linear warriors, quadratic wizards," with spells able to do quite a bit of damage, but a good item setup able to do truly staggering amounts with plenty of +to-dam and enough blows per round. So I'm looking for some ways to balance that out, mainly by increasing the variety of buffs that apply to spellcasting classes.

Second of all, and proceeding pretty directly from the above, I feel that many of the classes just aren't differentiated enough. By and large, my impression of The Optimal Path To Victory in Angband (setting aside personal challenges and deliberate increasing of difficulty through sub-optimal, but more interesting, choices) is that you just find ways to get your melee damage and AC as high as possible, with some method of escape from a battle that looks unwinnable. I'd really like to create a variety of different, but (more or less) equally viable paths to victory...or at least to a strong, high-level character. (I haven't thought far enough ahead to say whether or how much I'll be changing Morgoth's behaviour to account for all this!)

So with those basic premises laid out, here are the general thoughts I have on classes:

Mage:
- Existing mage archetype, at higher levels, tends to either shoot off a bunch of spells in a burst, then run away to regenerate mana and come back to kill the bigger monsters later, or be a hybrid type that fires some blasting spells, buffs up, and goes to hit them in the face until they die. While I have no problem in principle with either of these, I'd like to make the blaster-mage require less running away, and enable a full-on "buff and bash" mage without (much of) the blasting. Oh, and make debuffing monsters actually meaningful.

I've already added the following modifiers, all planned to be available on items (and the first two planned to be increased innately in mages):
- Spell speed: Spells can be cast faster than 1 per turn, to compete with multiple blows/round; however, allowing spells to also be slower than 1 per turn creates some interesting possibilities for balancing very powerful damage spells.
- Spell power: Increase damage, duration, and ability to overcome resistance by various percentages (currently, each +1 spell power on an item increases the spell power by 25%)
- Half and quarter mana usage

Priest:
- I feel that Priest and Paladin play very similarly right now—as in-your-face warriors who happen to be able to heal and cast some other spells. I actually quite like that role, but I think it more properly belongs to a Paladin, not a Priest. I'm still sort of stumbling around in the dark trying to find a new role for the Priest. My first vague idea, inspired by one of the Priest archetypes in Blizzard's Hearthstone, is a mind-control based Priest: turn evil creatures to the side of good and use them to fight their former fellows. There are a variety of difficulties with this idea, but I haven't come up with a better one yet (so I'm open to suggestions!).

Paladin:
- In general, as I said, I like where the Paladin is right now. I do like the idea of adding an innate Protection from Evil effect to Paladins starting at, say, level 30.

Ranger:
- I'm not sure exactly where I want the Ranger going yet, either. It's possible that the "buff-and-bash" mage role I mentioned above should go to the Ranger; it's also possible I should attempt to improve missile combat to parity with melee combat, similar to what I'm trying to do with the blaster-mage.

Warrior:
- My original idea for creating an Angband variant was actually to essentially implement the Legend game system, or at least its class/track system, in a roguelike. While I mostly want to defer that to a separate variant someday, I still like what the Legend Barbarian does, which is basically: go into a berserk rage, intimidate and later straight-out injure everyone around you by sheer force of presence, and gain the ability to shrug off damage that would turn a lesser being into a smear on the floor. I think it needs a certain amount of adjustment for Angband, but that's the general idea for what I'd like to do with the Warrior.

Rogue:
- This is where I actually started my efforts, again inspired by Legend, and, indeed, it was playing a Vanilla Rogue that led me to this in the first place, as it seemed to play like just a "Mage Lite". So the first thing I did was to implement Sneak Attack, which greatly increases the Rogue's chances of doing a critical hit—and damage if he manages it—when a monster is asleep. Thus, it greatly rewards stealth, and already changes the rogue game significantly at earlier levels. I've also implemented "bonus move" (inspired by Legend), which allows for moving further than is normally possible without monsters being able to do anything about it, but not attacking or otherwise acting more. (Basically, you can take 2 steps before the monster can do anything, but if you hit him, he gets a turn.)

Some thoughts about non-class-specific stuff to do:
- Already implemented "enemy memory", remembering a specific monster from the last 10 that you have fought, but not yet killed; this is also the first step towards allowing at least some class(es) to detect (track?) any monster in that list. Inspired by Smeagol getting away too damn many times.
- Already implemented a display of specific monsters' actual speed under its health display. I like that some monsters are a little bit faster or slower than their type's stated speed, but it's really aggravating to be trying to chase or run away from a monster that's supposed to be exactly the same speed as you, only to find that it's actually just a little bit faster and you didn't know it. Also helps with monsters that can haste themselves.
- More complex aggro: add ability for monsters to be hostile to each other, and/or friendly/neutral to you. This will make a Summoner role feasible (either as a separate class or another Mage role), as well as the aforementioned mind-control Priest, and possible animal companions for ranger. Probably requires fiendishly complicated changes to monster AI, but I think it might be worth it.
- Several additional buffs/debuffs, including two already implemented: Entangled (can act, but not move) for monsters and players, and Blind for monsters (which presently prevents spellcasting and makes them flow by smell).
- Standing effects, like an area of obscuring mist or acid pools on the ground, and temporary walls, like a wall of ice created by a spell.

So...that's all I can think of off the top of my head that I've got planned right now. The code is available on GitHub at https://github.com/danaris/tgcangband (TGCAngband being the provisional name of the variant)...and, as I said, feedback is welcomed!

[Derakon]

Hey, good luck with your variant! Greater differentiation is always welcome. I haven't taken a look at the actual specifics of your changes, but I do have some comments about the "current state of things" in Vanilla. Since you seem to want to keep/make things reasonably balanced, it'd probably be a good idea to be aware of how they are currently balanced:

1) Damage spells. There are two tradeoffs going on here.
1a) First, if you're not in melee range, then you take vastly decreased damage, since all monster turns that would be spent attacking you are instead spent moving closer to you.
1b) Second, attack spells always hit, and spell failure rates are almost always lower than melee miss chances. Thus, the average damage per round of melee weapons is closer to that of spells than it looks. For example, if you have 400 average damage/round in melee, but a hit rate of 80% (which is actually pretty high), then your actual average damage is 320 -- which is quite comparable to Mana Storm.

2) Mages can make excellent use of magical devices as a source of damage. All classes get a damage boost of (magic device skill - level of device)%. Mages have huge device skill, with the net result that attack wands typically do more damage than their spells do, for no at-the-time mana cost. Wands of Annihilation do more damage than Mana Storm does. This is horribly documented though.

3) Rangers get free extra shots with bows. A ranger with a good bow is absurdly powerful, at least for as long as their ammo lasts. And, like with spells, they take less damage by avoiding melee range. Moreover, since extra shots mean that each action they take is costing them half or a third of a turn, they get extra chances to react to monster actions (much like if they had double or triple their actual speed).

4) Priests get access to 0%-failure-rate healing spells, which allows them to be considerably more reckless than paladins. They also get Banish Evil in the late game, which is the only area-effect teleport-other spell in the game, and is consequently a major game changer. I will grant however that they do spend most of the game bashing monsters over the head and casting Orb of Draining, which doesn't make them a heck of a lot different from paladins.

Anyway, I look forward to seeing what you come up with. Good luck!

[danaris]

1) Damage spells. There are two tradeoffs going on here.
1a) First, if you're not in melee range, then you take vastly decreased damage, since all monster turns that would be spent attacking you are instead spent moving closer to you.

While this is true, and I will need to try and balance for this when making my changes (possibly by making some monsters much nastier from a distance), I have always felt, at least, that there are two types of monsters to fight: those who are primarily nasty from close up, and those who are especially dangerous from far away. The latter type tend to include primary spellcasters and high-damage breathers, and when I play a primarily melee character who can still do a fair amount of damage from a distance, I do find that some monsters it pays to close with, since once you're face-to-face with them, they're much less likely to breathe, summon, or otherwise cast nasty spells.

1b) Second, attack spells always hit, and spell failure rates are almost always lower than melee miss chances. Thus, the average damage per round of melee weapons is closer to that of spells than it looks. For example, if you have 400 average damage/round in melee, but a hit rate of 80% (which is actually pretty high), then your actual average damage is 320 -- which is quite comparable to Mana Storm.

This, too, is true, and I may very well end up giving most damage spells a save-for-half or save-to-evade for monsters, or something similar. However, while I can't swear to it, my recollection is that my damage per round with my last character (a priest) was more like 800-some odd under optimal circumstances.

2) Mages can make excellent use of magical devices as a source of damage. All classes get a damage boost of (magic device skill - level of device)%. Mages have huge device skill, with the net result that attack wands typically do more damage than their spells do, for no at-the-time mana cost. Wands of Annihilation do more damage than Mana Storm does. This is horribly documented though.

Yeah; I only recently discovered this, and have not actually played a character who tries to take advantage of it, for exactly the reason you note in your last sentence

I think this is worth making clear in some way within the game, probably in a similar way to seeing your weapon damage per round ('I'nspect the magic device and you'll see what your expected damage with it is.)

Incidentally, my implementation of Spell Power actually uses this exact mechanism, which I was somewhat surprised to find in place in just the way I wanted. All I needed to do was change a function argument from FALSE to spell_power-100.

3) Rangers get free extra shots with bows. A ranger with a good bow is absurdly powerful, at least for as long as their ammo lasts. And, like with spells, they take less damage by avoiding melee range. Moreover, since extra shots mean that each action they take is costing them half or a third of a turn, they get extra chances to react to monster actions (much like if they had double or triple their actual speed).

Yeah, this is definitely going to be a factor in my decision process for what, exactly, to do with the Ranger.
In my experience, however, as someone who really likes to keep his distance whenever possible, keeping enough ammo around can be a royal pain. It's possible that the new quiver mechanics in restruct will mitigate the difficulty of carrying enough ammo, but even if that is the case, it will remain a finite resource, which melee blows are not.

4) Priests get access to 0%-failure-rate healing spells, which allows them to be considerably more reckless than paladins.

Oh, yes, I enjoyed that considerably with my last bash-em-in-the-face character. The main thing I needed to bring with me aside from my weapons and spellbooks was !Restore Mana. It's kind of amazing when one !Restore Mana can do the work of about a dozen !*Healing*.

They also get Banish Evil in the late game, which is the only area-effect teleport-other spell in the game, and is consequently a major game changer. I will grant however that they do spend most of the game bashing monsters over the head and casting Orb of Draining, which doesn't make them a heck of a lot different from paladins.

Right. If anything, as you point out, because of their easy access to healing spells, they can be more of a bash-em-in-the-face class than paladins, which just seems very backward.

Anyway, I look forward to seeing what you come up with. Good luck!

Thanks!

[getter77]

Always good to see new variants trying to mix things up a bit---keep at it!

[fizzix]

While this is true, and I will need to try and balance for this when making my changes (possibly by making some monsters much nastier from a distance), I have always felt, at least, that there are two types of monsters to fight: those who are primarily nasty from close up, and those who are especially dangerous from far away. The latter type tend to include primary spellcasters and high-damage breathers, and when I play a primarily melee character who can still do a fair amount of damage from a distance, I do find that some monsters it pays to close with, since once you're face-to-face with them, they're much less likely to breathe, summon, or otherwise cast nasty spells.

One idea I've played around with, is mitigating some of the advantages of super-heavy breathers by making their breath spell take two turns and be telegraphed. So turn 1 is spent with something like "the dragon rears back" and then turn two it would breath. You could even say what the dragon would breath, so you'd get, "the ancient multi-hued dragon rears back and belches." This gives the player one turn to prepare. So, the next thing you could do is allow defensive moves like "block" which prevent some static percentage of incoming damage (based on shield?) or you could even include 1 turn elemental invulnerability consumables. The upshot of this all is that all the sudden fights with multiple heavy breathing monsters can become tactical as opposed to requiring you get rid of at least one of them immediately.

Regarding priests. You could always go the "evil priest" route, where you essentially resurrect an undead army and use that as your minions. Alternatively, you can have a "holy priest" route which has devastating attacks against undead and demons, but may have some difficulty against natural creatures (maybe combine it with a druid type character that can use natural creatures to fight?) Angband has never used allies partly because it's a pain to code the AI in a reasonable way.

Yeah, this is definitely going to be a factor in my decision process for what, exactly, to do with the Ranger.
In my experience, however, as someone who really likes to keep his distance whenever possible, keeping enough ammo around can be a royal pain. It's possible that the new quiver mechanics in restruct will mitigate the difficulty of carrying enough ammo, but even if that is the case, it will remain a finite resource, which melee blows are not.

Tome4's mechanic may interest you, although it seems a bit weird. The player equips a quiver, and that quiver has an infinite number of arrows, but only a given amount at one time. So you'd have a quiver that had 10/10 arrows in it. After you shoot one it's 9/10 and so on. When you get to 0/10 you can't shoot any more. At any point you can "reload" which puts some set number of arrows back in your quiver. It doesn't make much sense realism wise, but it does make some sense gameplay wise. Unfortunately, it removes the current mechanic in Angband where you can have super-powerful ammo which is very limited in use.

Oh, yes, I enjoyed that considerably with my last bash-em-in-the-face character. The main thing I needed to bring with me aside from my weapons and spellbooks was !Restore Mana. It's kind of amazing when one !Restore Mana can do the work of about a dozen !*Healing*.

You could consider splitting spell points from arcane casters from those of holy casters. So that RM only works for the arcane caster pool. Holy casters can regain some set number of points by praying, perhaps with some negative effect (chance of perm stat drain?)

[Derakon]

I have always felt, at least, that there are two types of monsters to fight: those who are primarily nasty from close up, and those who are especially dangerous from far away. The latter type tend to include primary spellcasters and high-damage breathers, and when I play a primarily melee character who can still do a fair amount of damage from a distance, I do find that some monsters it pays to close with, since once you're face-to-face with them, they're much less likely to breathe, summon, or otherwise cast nasty spells.

I'm guessing you haven't had a chance to look at the monster "AI" routine, which is almost literally:

1) If (1 in cast chance): cast a random spell.
2) Otherwise, if adjacent to player, attack in melee.
3) Otherwise, move closer to player.

In other words, the chance of a monster choosing to cast a spell does not depend on whether or not they're in melee range. The only monsters that are less dangerous in melee range than at range are those who you need to restrict LOS on, like summoners.

Maybe this is something you should change in your variant.

[danaris]

I'm guessing you haven't had a chance to look at the monster "AI" routine, which is almost literally:

No, I haven't yet.

1) If (1 in cast chance): cast a random spell.
2) Otherwise, if adjacent to player, attack in melee.
3) Otherwise, move closer to player.

In other words, the chance of a monster choosing to cast a spell does not depend on whether or not they're in melee range. The only monsters that are less dangerous in melee range than at range are those who you need to restrict LOS on, like summoners.

Huh. That's interesting. I could swear I remember some (though by no means all casting) monsters breathing on me or casting spells essentially every step as I approached them, then doing so only rarely when I was in melee with them...but at the time, I was just playing along, not taking careful records. I'll have to pay closer attention to the actual timings of such things in future.

Maybe this is something you should change in your variant.

I think I should, actually

I've always liked AI, and though I don't have anything specific in mind yet, I think there are probably a lot of little (and maybe some big) ways in which the Angband AI could be improved—whether with the aim of making it smarter, more human-like, or just better balanced for the particular needs of my variant.

[danaris]

One idea I've played around with, is mitigating some of the advantages of super-heavy breathers by making their breath spell take two turns and be telegraphed. So turn 1 is spent with something like "the dragon rears back" and then turn two it would breath. You could even say what the dragon would breath, so you'd get, "the ancient multi-hued dragon rears back and belches." This gives the player one turn to prepare. So, the next thing you could do is allow defensive moves like "block" which prevent some static percentage of incoming damage (based on shield?) or you could even include 1 turn elemental invulnerability consumables. The upshot of this all is that all the sudden fights with multiple heavy breathing monsters can become tactical as opposed to requiring you get rid of at least one of them immediately.

Yeah, that's actually kind of a cool idea. It even raises the possibility of "attack scripts" of various kinds, rather than simple, single (or multiple) attacks: rather than having to give a monster 6 attacks per round, or several spells, to be able to deal 6 different kinds of damage, you could give them a series of different attacks to be done on separate turns. Multi-turn breath weapons would then just be a particular case of those, with turn 1 of the "breathe fire" attack script being "rear back, take no offensive action," and turn 2 being "breathe."

(Or that could be more complicated than necessary, and I could just implement it specially for breath attacks...but either way, an interesting idea!)

Regarding priests. You could always go the "evil priest" route, where you essentially resurrect an undead army and use that as your minions.

Hm. That's an interesting idea for a class, but I would call it a "necromancer" rather than a "priest." And if I decide to implement that role and can't come up with anything good for priest in the end, then I'd just replace priest with necromancer—I'd rather have things named in a sensible, predictable manner than have all the "standard" classes everyone expects, if I can't do them well.

Alternatively, you can have a "holy priest" route which has devastating attacks against undead and demons, but may have some difficulty against natural creatures (maybe combine it with a druid type character that can use natural creatures to fight?)

It's a thought, but it does feel a lot like a minor variation on the Paladin role. I'd really prefer to have something more different. I think the fundamental problem is that the "Priest" role as it was originally created in D&D is a support role—the party's healer, hanging back and keeping everyone alive while the warriors hacked and the mages blasted. That's part of why I like the idea of a mind-control priest: it would let the priest heal his controlled creatures, and be in that support role again.

Angband has never used allies partly because it's a pain to code the AI in a reasonable way.

Yeah, I'm expecting it to be a bit of a bear. My initial thought, without having looked much at the monster AI code yet, is that the first step should be to try replacing references to the player in the monster AI with references to a "target," which can be the player or another monster.

But there's a lot more to it than just that, especially once you get into the idea of actually mind-controlling a monster: do you then get to actually be in direct control of multiple units at once? Or do you just get to give them more general commands, like "kill that creature, then anything else around", or "stand there and kill anything that comes in range"?

Tome4's mechanic may interest you, although it seems a bit weird. The player equips a quiver, and that quiver has an infinite number of arrows, but only a given amount at one time. So you'd have a quiver that had 10/10 arrows in it. After you shoot one it's 9/10 and so on. When you get to 0/10 you can't shoot any more. At any point you can "reload" which puts some set number of arrows back in your quiver. It doesn't make much sense realism wise, but it does make some sense gameplay wise. Unfortunately, it removes the current mechanic in Angband where you can have super-powerful ammo which is very limited in use.

That is interesting, but the lack of realism does bother me. And I do like the option to have powerful ammo. For years, I've toyed with the idea of an artifact arrow of some sort, either as something you can shoot at the enemies and it will automatically reappear in your quiver, or something you can just collect infinitely without having to worry about breakage. Neither of those seems to offer a truly optimal missile-weapon experience (the former being probably too powerful, the latter being probably too weak with a single shot), but I'm sure there's some kind of approach that will work.

You could consider splitting spell points from arcane casters from those of holy casters. So that RM only works for the arcane caster pool. Holy casters can regain some set number of points by praying, perhaps with some negative effect (chance of perm stat drain?)

That's actually a good idea. I'm not sure what the "pray for restored divine SP" mechanic should look like, but removing the ability of priests to cast Heal 10 times, drink a !RM, then cast it another 10 times, and keep doing this through their 20-odd !RMs, is probably good for balance.

[Nick]

This looks very interesting.

I commend your choice of starting from the restruct branch, and encourage you to keep updating as it changes - the structure of the monster AI code should become easier to tweak soon.

[danaris]

This looks very interesting.

I commend your choice of starting from the restruct branch, and encourage you to keep updating as it changes

Oh, definitely. I've already been very pleased with the way things have moved, having looked at the code a little bit back sometime in the pre-3.4 era.

the structure of the monster AI code should become easier to tweak soon.

Then I shall await these changes with bated breath—and not try to work on that part just yet!

[fizzix]

But there's a lot more to it than just that, especially once you get into the idea of actually mind-controlling a monster: do you then get to actually be in direct control of multiple units at once? Or do you just get to give them more general commands, like "kill that creature, then anything else around", or "stand there and kill anything that comes in range"?

As a first step you can code it as "friend attacks nearest monster to it."

A second step would be to give the player a command which is essentially "attack that monster". Then you'd target the enemy and all allies would move towards and attack that enemy, only attacking something else if it's in the way.

A third step would be to allow each ally to be targeted towards a different enemy.

Then you can continue adding other commands to give to allies, such as "wait" or "pick up weapon" or "use wand" or other things. You can set up exclusion zone to prevent allies from walking towards them.

The hardest part is ally pathing, especially if you want to do anything clever. You don't want to be regenerating a heat map for each ally each turn. So, you can start with only being able to target enemies that the ally can see (pathing is then easy) or targeting a place on the ground (you only need to calculate the heat map for that target once).

[danaris]

As a first step you can code it as "friend attacks nearest monster to it."

A second step would be to give the player a command which is essentially "attack that monster". Then you'd target the enemy and all allies would move towards and attack that enemy, only attacking something else if it's in the way.

A third step would be to allow each ally to be targeted towards a different enemy.

That all sounds like the stuff I've been (vaguely) thinking of. One inspiration is Escape Velocity's escort commands interface, which lets you give several general commands to all escorts, and also select specific escorts in your targeting display to give specific commands to.

As an aside, another extension of all this I've been considering is expanding the possible different "sides", allowing monsters to aggravate each other to some extent—for instance, replace the "trample weaker monsters" with an ability to attack another monster that gets in its way, which could then get angry about that, and potentially have its "friends," if it has any, help fight back against the bigger monster that just tried to walk all over it.

Then you can continue adding other commands to give to allies, such as "wait" or "pick up weapon" or "use wand" or other things. You can set up exclusion zone to prevent allies from walking towards them.

Those sound cool; I wasn't (yet) thinking of adding the capability for monsters, whether allied or enemy, but it's definitely a logical extension of the foregoing changes.

The hardest part is ally pathing, especially if you want to do anything clever. You don't want to be regenerating a heat map for each ally each turn. So, you can start with only being able to target enemies that the ally can see (pathing is then easy) or targeting a place on the ground (you only need to calculate the heat map for that target once).

Here, I just have to take your word for it thus far.
I honestly don't have a problem, at least for the time being, with only being able to target enemies that the ally and the player can see.

Though that raises a new question: Does the player get to see what his allies can see? And that raises another question, of what monster vision is really like. Can all, no, or some monsters see invisible? See in the dark?

That suggests it's easier to leave player vision belonging solely to the player, at least for a version 1.0 of TGCAngband...though perhaps I might let the player "sense" the positions of allies at all times.

[wobbly]

Poschengband has "pets" and summons and seems to handle them reasonably well, so may be worth taking a look at.

[EpicMan]

In terms of different mechanics for Priests, you might look at Scthband (downloadable from the Variants tab).

In Scthband (which may call them Shamans instead of priests, don't remember) priest casting used connections with Life or Wild spirits instead of books. The spirits functioned more or less as spellbooks, except that you had to equip them (and had limited slots) and that casting spirit magic didn't use SP.

Instead, casting a spell ("asking a favor") would be bothering a spirit, and add a cooldown before that spirit could be asked for a favor again. You could cast from that spirit before the cooldown ran out, but you ran the risk of annoying the spirit, which could add a bunch to your cooldown or even result in monsters being summoned around you as punishment. If extremely angry the spirit could severe your connection, essentially destroying your spellbook.

[half]

The hardest part is ally pathing, especially if you want to do anything clever. You don't want to be regenerating a heat map for each ally each turn.

For what it's worth, since version 1.2, Sil does this for every monster every turn and there are no noticeable slowdowns. There are something like 20 - 100 monsters on a level.