Our community module

[lukep]

Bodies are composed of parts; these parts are variable, but follow a standard scheme for type.
Parts have sizes and damage resistances/susceptibilities. The latter may or may not be inherited from armor.
Attacks can hit the whole body or individual parts of the body.
Some attacks have precision ratings; these determine re-rolls for the purposes of accuracy. (i.e; weapon precision 2 allows you to take the better of two rolls. Noninteger rolls could work; 2.6 could mean 2 rolls 40% of the time, and 3 rolls 60% of the time.)
Damage to overall health is taken as the sum of the individual hits.
Body parts themselves may or may not also take damage; the definition of "damage" may be different for these.

My expanded idea of accuracy:

Each body part has an individual size, which may or may not be changed by swapping out parts. In addition, attacks can miss by hitting the miss "size" instead, which is dependent on range. Example: arm (x2)= 3 size each, head = 1 size, legs = 8 size, chest = 10 size for a total of 25 for the player (other enemies could be bigger or smaller, eg, a rat could be 5 size, a car could be 100, a mech walker could be 500, unless all enemies are human sized)

"Miss" size is (tiles away^2 * 5).

Body parts are ranked from most desirable to hit to least, then miss. The hit check rolls a number between 1 and (body size + miss size), as many times as the precision rating is. The highest number is kept, and damage is done using that part, or it is a miss.

Examples:

An enemy shoots a gun (precision 6) at the player (size 25, parts as above) from 5 tiles away (size 125 miss zone). 6 numbers are randomly generated between 1 and 150, 1 = head, 2, 3, 4 = left arm, 5, 6, 7 = right arm, 8-15 = legs, 16-25 = chest, 26-150 = miss. Chance to hit would then be 3.9% head, 11.0% left arm, 9.7% right arm, 21.4% legs, 18.7% chest, 32.4% miss (total of 97.1% ... rounding errors?). This seems to be a reasonable chance to hit or crit (head shot) to me.

The player swings a club (precision 3) at a gelatinous cube (size 200, one part) from melee range (size 5 miss zone). 3 numbers are randomly generated between 1 and 205, 1-200 = hit, 201-205 = miss. Chance to hit would then be 99.9985% hit, 0.0015% miss.

The player swings a club (precision 3) at a mouse (size 5, 1 head, 4 body) from melee range (size 5 miss zone). 3 numbers are randomly generated between 1 and 10, 1 = head, 2-5 = body, 6-10 = miss. Chance to hit would then be 27.1% head, 63.8% body, 9.1% miss.

Although this looks quite complex, I don't think that it would be confusing to the average player with a good UI.

[bricks]

I like it, lukep; very similar to what I had in mind, and your precision and "miss" size concepts really round things out. The fact that a flamethrower is more effective than a sniper rifle against a swarm of flies will be a rewarding consequence of this system. I think those "spread"-style weapons could just have a Precision of 0, and have a different handling for damage. The "size" calculation could perhaps be penalized for melee-ranged combat with a long-range weapon; that might make for a good talent (either overcoming the penalty or causing enemies to be penalized for being up close). For special hits (legs == chance to cripple, arms == chance to disarm, etc.): These could be given fake "damage" ratings (i.e. disarming = 20 damage) so the precision rating takes into account these extra effects (damage_value = weapon_damage + effect_value). The formulas for these needn't be flat number, they could prioritize hitting explosive robots on their fuel tanks when they are far away, but aiming elsewhere when up close.

[lukep]

I like it, lukep; very similar to what I had in mind, and your precision and "miss" size concepts really round things out. The fact that a flamethrower is more effective than a sniper rifle against a swarm of flies will be a rewarding consequence of this system. I think those "spread"-style weapons could just have a Precision of 0, and have a different handling for damage. The "size" calculation could perhaps be penalized for melee-ranged combat with a long-range weapon; that might make for a good talent (either overcoming the penalty or causing enemies to be penalized for being up close). For special hits (legs == chance to cripple, arms == chance to disarm, etc.): These could be given fake "damage" ratings (i.e. disarming = 20 damage) so the precision rating takes into account these extra effects (damage_value = weapon_damage + effect_value). The formulas for these needn't be flat number, they could prioritize hitting explosive robots on their fuel tanks when they are far away, but aiming elsewhere when up close.

Just to clarify, you are proposing that body parts are sorted based on value of a hit, where 1 damage = 1 "value", 1 turn disarmed = 20 "value", 1 turn pinned = 15 "value" etc...

I'm not sure how well this would work, as it is very difficult to determine what amount of damage has equivalent utility to a status effect in every situation.

As for penalizing short range combat, one thing that I think would work would be to have a minimum calculated range for each weapon. For example, a rifle could have a minimum calculated range of 8, resulting in a miss size of 320 for anything in range 1-8. A pistol could have a minimum calculated range of 4, giving a miss size of 80 up to that range, and a pike could have a range of 2, giving a minimum miss size of 20.

EDIT: weapons would also have a maximum range, though I imagine it being almost completely ineffective at max range against some enemies. This would also simplify the display, giving a pistol (4-15) range, a rifle (8-30?) range, a pike (2-3) range etc...

[bricks]

Yeah, that's exactly what I'm saying. It would definitely be tough to define, it's more of a band-aid for dealing with the interaction between precision and special hits. If the latter are even included, of course.

Minimum range sounds good, though anything over 5 would be pretty mean if we're talking ToME-styled maps. Gonna do some math to figure out how that balances with precision...

Edit:

Assuming a pistol (min. range = 4, precision = 6) and rifle (min. range = 8) aiming at a roughly humanoid target (size = 25), the break-even point for accuracy at short range is precision = 21 for the rifle. Obviously you'd want rifles to be less effective than this at this range (that is, pistols should be more accurate between range = 1 and range = pistol min. range). Pistols range in accuracy from 80% at min. range to 36% at rifle's min. range. To surpass a pistol's accuracy at its minimum range by 50%, a rifle needs approximately 10 precision (53% accuracy at or below minimum range). This seems like an acceptable trade-off. Larger enemies require less precise weapons, naturally. Rifles and pistols would be equally effective against the hypothetical gelatinous cube at short range, and the benefit of using a rifle at long range is minimal (about 5% more accurate). The mouse is hell to hit with any ranged weapon; at minimum range, 30% with the pistol and 14% with the rifle. At long range that drops to 9% for the pistol. Break-even points: 6-7 range for humanoids, 6-7 for mouse, 5-6 for gelatinous cube. Pretty consistent. For a hypothetical size = 1, the break point is 6-7. Break points don't reach 4 until sizes of over 1800, at which point you shouldn't ever miss anyway (about once in every 10^9 hits, I believe).

Seems like a solid system! Naturally this doesn't account for differences in damage and the value of reduced variance (which is a complicated function).

[Devorius]

Realistically long range weapons are generally still accurate at close range; though that is relative anyway, as we're talking about a game... However, what do you think about a speed penalty rather than directly reducing accuracy? Something like: a rifle takes a turn to fire, or better yet long range weapons require you to perform an "aim" action prior to actually firing.

A hybrid style would also be good: long range weapons require "aim" action before firing OR drop accuracy if the player chooses not to.

Just some more food for thought...

[bricks]

I agree that a speed penalty would make the most sense, I just don't think action speed modifiers are a good design choice for roguelikes. DoomRL has alternate fire methods for weapons; for pistols you can choose to "aim" which takes longer. Flipping that idea on its head, long-range rifles could perhaps get an accuracy bonus/decreased minimum range if you choose to take a turn to aim (which could just be an implied action when standing still). I'd also like to see some sort of run-and-fire action, too, which could have an accuracy penalty. All good tactical options.

[Goblinz]

With speed penalization for longer weapon how about giving the player the chance to set up shot. If I have a sniper rifle and want to spend 6 turns setting up that shot I should have really high accuracy for that shot. So each weapon would have a maximum number of turns to aim (there presumably be diminishing returns), after all I can spend an hour looking down the sights of an pistol and I am not going to get any more accurate.

[lukep]

I would rather have accuracy be a (displayed) variable from turn to turn over increasing steadily. This variable boost would only apply when standing still, and would add anywhere between 0% and 100% of a "aim bonus" specific to the weapon.

For example, a pistol could have 6 (+2) precision, meaning that you can gain a bonus of up to 2 more precision on top of the base 6 by choosing to shoot on a turn that has a 100% aim bonus, bringing the precision rating from 6 to 8.

Meanwhile, a sniper rifle could have 8 (+10) precision, meaning that you can more than double your precision by waiting for a good turn.

This would force more decisions onto the player, for example, if a 70% bonus is good enough, or if they should risk the enemy noticing them/closing more.

[bricks]

I think a single round bonus would be good. Beyond that it's going to be hard to convince the player that sitting around for multiple turns is better than just taking a bunch of pot shots. Doubling the precision score works out nicely in the math. I don't think this should count for anything other than ranged weapons; you can't exactly "aim" a melee weapon, though a range-2 spear is a corner case. lukep's suggestion sounds good too, though I hesitate to attach a bunch of numbers to weapons. Depending on the route taken for talents, these sorts of bonuses could be granted on that basis, and groups could be given bonuses categorically - +2 for all pistol-type weapons, +10 for all rifle type weapons.

For shotgun-style weapons, I think hit rate should be 100%, but the damage should scale using the same formula lukep proposed for accuracy - damage = (max damage) * size/(size + k*distance^2). k could be 5, or it could vary depending on the type of shotgun, k being a "spread" parameter.

[lukep]

For shotgun-style weapons, I think hit rate should be 100%, but the damage should scale using the same formula lukep proposed for accuracy - damage = (max damage) * size/(size + k*distance^2). k could be 5, or it could vary depending on the type of shotgun, k being a "spread" parameter.

For a shotgun, I was thinking it would be something like 40x 1-precision attacks, depending on balance. This seems more intuitive to me, though it could lead to huge amounts of damage in edge cases.

lukep's suggestion sounds good too, though I hesitate to attach a bunch of numbers to weapons.

It does seem to be getting more complex, but I'm not sure if it's gone too far. If all of my presented ideas are used, weapons would look something like:

Rifle
Damage: 45 (kinetic/bullet)
Range: 8-30
Precision: 8 (+12)

SMG
Damage: 10 (x3) (kinetic/bullet)
Range: 5-20
Precision: 5 (+4)

Pistol
Damage: 15 (kinetic/bullet)
Range: 4-15
Precision: 6 (+2)

Shotgun
Damage: 0.75 (x40) (kinetic/bullet)
Range: 2-9
Precision: 1

Pike
Damage: 25 (kinetic/edged)
Range: 2-2
Precision: 3

Sword
Damage: 30 (kinetic/edged)
Range: 1-1
Precision: 3

I don't think that three statistics with two parameters each is too much for a weapon, and I'm not sure how much less you could have without sacrificing meaningful variety.

I think a single round bonus would be good. Beyond that it's going to be hard to convince the player that sitting around for multiple turns is better than just taking a bunch of pot shots. Doubling the precision score works out nicely in the math. I don't think this should count for anything other than ranged weapons; you can't exactly "aim" a melee weapon, though a range-2 spear is a corner case.

By going significantly past double, it would be possible to hugely incentivise waiting a turn between shots. For example, if the rifle was something like 4 (+24) precision, the character could get an average of 16 precision every second turn. The choice that the player would then need to face is is the bonus would show up as +10% or some other low value, whether to wait one more turn to reroll it, or take the shot and need to wait again before the next one.

[yufra]

Some really great discussion guys. I want to point out some implicit assumptions you guys are making, as I have done on the wiki already. You guys are talking about instantaneous damage projection, and there has been no discussion of a projectile-based system. One system that I used in the past for ranged weapons is to have a projectile for each attack (bullet, acid spit, etc). When you initially set the projectile on its way you take the angle of the line between the source and the target and add a random number from a Gaussian distribution, the width of which is controlled by an "accuracy" parameter. This handles poorer accuracy at longer range since the deviation will grow with distance. I do like the emphasis on size, and we could add a check so that a projectile entering a tile with an actor does not automatically hit that actor (like it does in ToME).

For anything but fast moving projectiles I think the defender should have some ability to dodge/block/etc attacks. Any thoughts on that?

[Rectifier]

We talked about this before crashing on irc yufra and I agreed with the premise of projectile deviation and dodge/block, along with cover being a factor.

In the dodge/block concept, this could be accomplished a number of ways, such as a "dex" type attribute adding to a dodge chance or the armor type affecting the dodge/block chance. In any case, cover should be a big part of a sci-fi encounter with obvious obstacles such as corners or pillars; but what about interact-able cover, such as a low wall? Ideally the player could choose when to peek their head out to shoot, duck, or fire blindly. Or to balance something like this a cover system could only be accessed by certain classes...

Another issue that came up during that conversation was size. Most encounters will likely be with humanoids or anthropomorphic robots, but what about small critters and larger beasts? Obviously small targets should not be as easy to aim and hit as say...barn door sized monsters and would need dodge/block chances reflecting this. Furthermore, it doesn't appear that multi-tile actors are supported by tome (and such a system is a pain in the ass to implement, constrains map layouts, and can create weird rules), so if people want large creatures, they'll probably be within a single tile and size represented by a statistic rather than graphics. That isn't anything new, but might as well clarify.

Regardless we should come to a consensus on the basics of combat, movement, and field of vision so that they can be tested.

[bricks]

Doing an actual spread for shotgun attacks would be fun, and more consistent with overall mechanics; great idea. I'd like the stand-still-to-aim mechanic to be a little more deterministic since we already have a random chance to hit. I also hesitate to suggest to the player that they should always wait before shooting, except in the case of very specific weapons like sniper rifles (likewise, there should be weapons that get little to no precision bonus from waiting around).

yufra, I think the gaussian variation in attack angle does look really nice and adds a good amount of realism. Hopefully the math can be worked out such that the player doesn't get overestimates on weapon accuracy.

Dodging and blocking should definitely be options. There should be many types of dodging. In addition to gaining accuracy, standing still could cause you to go "prone," reducing your effective size by some factor. DoomRL has a "running" tactic that both reduces your accuracy (with non-shotgun/melee weapons) and the chance of an enemy of hitting you for a few turns. DoomRL also has side-stepping, a really interesting mechanic. If you move roughly perpendicular to an enemy's attack, there is a chance that they will target the last tile you were standing on instead of the new one. I can see certain cybernetic and genetic pieces providing special dodge conditions, perhaps even as a "reactive" dodge which causes you to leap in a random direction when targeted (which has it's obvious benefits and drawbacks). Blocking can also follow many schemes. Glancing vs. square blows and "deflection shields" which cause projectiles to fly off in random directions come to mind. I like low walls, too, like Rectifier pointed out.

Perhaps hitting tab/some easy-to-reach button could flip between "prone," "running," and "normal" stance. Prone would be canceled on moving, running would be canceled if the player didn't move on the last turn. Running would provide an accuracy penalty but make you harder to hit; prone would allow you to hide behind short walls, increase accuracy, and reduce your effective size, but also require you to take a turn to stand before moving. I don't think three tactical options is excessive, especially when one is basic and the other two are mutually exclusive.

A little clearer on the gameplay side: you hit "tab" once and it flips you from "Normal" to "Going prone." So long as you don't move on the next turn, you will have a reduced effective size and gain a precision bonus. You will also be blocked from sight by short walls. To indicate your stance, it now reads "Prone." If you try to move, you will stay in place but return to "Normal." Now you can hit tab twice, going from "Normal," through "Going prone," to "Starting run." If you do not move, you'll simply return to "Normal." If you do move, you will now be "Running." Attacks against you receive a precision penalty, but you also lose precision on your attacks. You will return to "Normal" if you don't more in a turn where you are "Running." While "Running" you can vault over low walls (perhaps other agility effects). While "Running" or "Prone" you can't directly change your tactical stance. Energy could be drained while running, perhaps dependent on body weight and leg type. Some bodies might be incapable of going prone.

Edit: Long discussion on damage types on the wiki; my opinion is to just pick a few and see if they work. Kinetic, Heat, Shock, Toxic get my vote.

Edit2: Not sure how committed people are to "Gene Revolution." I think "Re-Evolution" would be a better title, since it incorporates the whole "post-human" theme, including both genetics and cybernetics. Kinda moot since there's no plot yet, but I thought I'd get it down. The name isn't entirely original, but the other contexts include a webcomic, a new-age blog, and a punk clothing store, so I don't think there'd be any concern for cease-and desist. Time to play with shotgun mechanics...

Edit3: I know it's the wrong language, but I made a python script that tests out weapons on different targets, using the info that lukep typed up. So far, SMGs look very powerful. There are a few features missing that I'm going to play with, then I may expand it into different body types. Then I might actually do some useful coding. I think LUA is too close to python for my comfort.

[lukep]

I'm not sure how well it would work, but precision could target at a cone, with increasing values being closer to the centre. For example, a result of 40 on a 25 sized target would mean that the bullet flies at a point 3 tiles (0.3 tiles? 1 tile?) to the side of your targeted point because it is off by 15, and each tile represents a precision-size of 5 (or 50, or 15).

One side effect of this would be that moving at all would cause the most precise attacks to miss, so it would need to be fixed to account for that. One thing that would help would be to have various velocity values, but that may be getting to be too complex, with it being the eighth statistic for weapons (added to damage magnitude, damage type, number of attacks, minimum effective range, maximum range, precision, and aim bonus)

As for the SMG being very powerful, I haven't really balanced the presented weapons at all, they merely illustrate variety and a general idea for weapons. The damage could easily be reduced, or the rifle and shotgun damage increased (and have the handgun stay weak to compensate for it needing one hand).

I like the transhumanist angle for a theme and background. How about rEvolution as a (somewhat gimmicky) name?

[lukep]

In terms of leveling up:

All parts/mutations level up individually, giving one lots of different imrovements almost continually. But each part has a max level --- maybe not?! Almost I guess like iron vs steel vs titanium etc. Also when removing a part some of its power lingers. So you can either build so that you keep you parts on you for long before switching (less frequent levelling up), thus optimising the permanent gains, or switch to more powerful parts, whose permanent gain wouldnt be as close by and perhaps not as strong.

My idea for class/levelling/parts (note: I am working off the assumption that weapons are intrinsic to parts, not separate items, though it could work either way without much modification):

Each body part (eg. arm of smashing) has 10(?) levels, gaining in statistics each level. They start with various intrinsic talents as well. In addition to the part improving with levels, it gives perks to the part slot (eg. right arm) that it is in. Example:

---

Arm of Smashing (lvl 1)
Damage: 30 (kinetic/blunt)
Range: 1-1
Precision: 3

Intrinsic talents (talent level = item XP level / 2):

Knockback: 5 energy, 6 cooldown, melee attack
Hits for (150-200)% damage, knocking back (2, 2, 3, 3, 4) tiles.

Power Sweep: 10 energy, 4 cooldown, melee attack
Hits for (200-250)% damage in a frontal arc.

Ground Pound: 15 energy, 10 cooldown, melee attack, 1 radius
Hits all in area for (50-75)% damage as kinetic/concussion damage.

Levelling bonuses:
1: part + 1 precision
2: body slot perk -> (+5 kinetic/blunt damage on melee attack)
3: part + 5 damage
4: part + 5 damage
5: body slot perk -> (Knockback talent level 1)
6: part + 1 precision
7: body slot perk -> (+1 precision, melee attacks only)
8: part: + 10 damage
9: part: +3 precision
10: body slot perk -> (Power Sweep talent level 3)

---

That would mean that a fully levelled Arm of Smashing would look like this (including bonuses from the improved right arm slot):

Arm of Smashing (lvl 10)
Damage: 55 (kinetic/blunt) (50 base + 5 from right arm bonuses)
Range: 1-1
Precision: 9 (8 base + 1 from right arm bonuses)

It would have access to all of the talents at talent level 5.

---
If, after levelling up your arm of smashing fully, you decided to switch to an arm of slicing, some bonuses would be retained, as shown here:

Arm of Slicing (lvl 1)
Damage: 30 (kinetic/blade) + 5 (kinetic/blunt)
Range: 1-1
Precision: 4 (3 base + 1 from right arm bonuses)

Intrinsic talents (talent level = item XP level / 2):

Pin: 5 energy, 6 cooldown, melee attack
Hits for (130-180)% damage, pinning for (2, 2, 3, 3, 4) turns.

Stab: 8 energy, 10 cooldown, melee attack
Hits for (200-250)% damage, bypassing armour.

Parry: passive
(5, 10, 15, 20, 25)% chance to deflect an incoming melee attack, negating it.

In addition to those three talents, you could also use Knockback (lvl 1) and Power Sweep (lvl 3) as well.

Levelling bonuses:
1: part + 5 damage
2: part + 1 precision
3: body slot perk -> (+5 kinetic/blade damage on melee attack)
4: part + 5 damage
5: body slot perk -> (Stab talent level 2)
6: part + 1 precision
7: body slot perk -> (+1 precision, melee attacks only)
8: part: +3 precision
9: part + 10 damage
10: body slot perk -> (Pin talent level 3)

---

That would mean that a fully levelled Arm of Slicing (with the bonuses from the Arm of Slicing as well) would look like this (including bonuses from the improved right arm slot):

Arm of Slicing (lvl 10)
Damage: 55 (kinetic/blade) (50 base + 5 from right arm bonuses) + 5 (kinetic/blunt)
Range: 1-1
Precision: 10 (8 base + 2 from right arm bonuses)

It would have access to all of the Arm of Slicing talents at talent level 5, as well as Knockback (lvl 1) and Power Sweep (lvl 3).

---

I don't know if there should be a limit (say, 5 or 10) to the amount of body slot perks that can be applied to a specific part, or if it should just get unequivocally better with levels.