Author Topic: Weapon Discussion Thread  (Read 50281 times)

Offline Arthanor

  • Commander
  • *****
  • Posts: 2488
  • XCom Armoury Quartermaster
    • View Profile
Re: Weapon Discussion Thread
« Reply #60 on: August 22, 2016, 10:58:15 pm »
That's why I don't leave home without Meridian's .exe which features gaussian multipliers. Fuck linear distribution 0-2x; that shit is dumb.

That said, I'm definitely of the opinion that

A: Gaussian damage multipliers should be default or at least optional.

and

B: Auto-only weapons should be able to Auto Reaction fire to compensate for some of their loss of appeal due to Gaussian multipliers making precision weapons significantly more consistent; also to compensate for their ridiculously high TU costs and weight + relatively short range.

Some kind of reaction fire would allow for proper suppression fire that these weapons are supposed to be doing. This is especially crippling for the AI: See a machine gun? That's ok, just rush the guy in melee, he can't shoot! It was especially noticeable with the siberian mission, where the big power armored guys were pretty much harmless between smoke and getting rushed on my turn by melee gals who could just run behind them and smash them with a bardiche. When I implemented auto-weapons in vanilla, I gave them a snapshot which didn't make sense to use generally (it was way worse than auto-fire in terms of Accuracy/TU) but at least allowed them to react. I prefer that.

Quote
I agree with the sentiment that there should be a 'finesse' damage bonus (or armour ignoring) for accuracies in excess of 100%; I believe I actually mentioned it to Dioxine on Gaudium007's stream.

That said, with or without ridiculous multiplier distributions, snipers do have a clear role in terms of smoke and snipe which heavy weapons simply cannot do at nearly the same ranges, but I generally agree with most of what you're saying.

What kind of range issues do you have with HMGs, Vulcans and Rocket launchers? All of those have unlimited accurate range like sniper weapons. In fact, they are often able to blow away both the cover and the target, making them far superior to sniper weapons. The XG-Chaingun and Fatty don't have unlimited range, but their TU cost and weight allows you to bring them with power armored or assault gals, to areas where they can definitely function.

Buggy engine. Problems that date back to 1994.
I believe there is hope for a fix soon, though.
There is always hope! But for now I trust in "learning the engine" more than fixing it...

Offline Star_Treasure

  • Captain
  • ***
  • Posts: 96
    • View Profile
Re: Weapon Discussion Thread
« Reply #61 on: August 23, 2016, 02:11:23 am »
This problem will hopefully disappear soon, as there is a patch to fix Line of Fire behavior available - if it works, it'll probably be merged. When it does, you will be simply getting 'No LoS' messages when trying to shoot through such obstacles.
Which is why you miss, the game lies to you that hitting is possible, while it isn't.

What about over penetration? Surely bullets ought to be able to pass through a fence with only minimal reduction in power.

Offline Surrealistik

  • Colonel
  • ****
  • Posts: 484
    • View Profile
Re: Weapon Discussion Thread
« Reply #62 on: August 23, 2016, 02:30:25 am »
I thought the AI cheated (as in they could RF with autofire) so I never approached the Supersoldiers from the front. It's possible they didn't fire on you because they have pretty awful reactions. Personally I utilized Ghost Daggers against these dudes to stab at their souls; usually one-two shot them for minimal TU cost.


At any rate, yes, I definitely feel that autofire only weapons could and should benefit greatly from auto-fire reaction. This would preserve the utility of the snap shot, while allowing those weapons to fulfill their suppression function without being OP as I mentioned previously to Dioxide.


Concerning range issues, HMGs/Vulcans spray and pray aren't as reliable as an aimed shot from a sniper weapon at the longest distances. Say you have a dude with 100% Aim; with an HMG you still have a chance of missing vis a vis a sniper weapon that will always hit (or should; we're apparently getting a fix for buggy projectiles). The chance of missing on all shots is very small, yes, but I'd much rather be sure to hit once with the sniper weapon that will reliably kill with gaussian multipliers, vs have a chance of hitting multiple times. Also spreading into your own troops is also a definite concern vis a vis a sniper weapon which the volume of fire combined with much less then perfect accuracy presents.

Rocket launchers are basically sniper rifles albeit with massive HE rounds.

Chaingun and fatty require actually moving into position which defeats the point of long range/cross map firepower.


« Last Edit: August 23, 2016, 02:48:02 am by Surrealistik »

Offline Star_Treasure

  • Captain
  • ***
  • Posts: 96
    • View Profile
Re: Weapon Discussion Thread
« Reply #63 on: August 23, 2016, 05:46:33 am »
Is it possible to code a snap shot that fires more than once?

Offline Arthanor

  • Commander
  • *****
  • Posts: 2488
  • XCom Armoury Quartermaster
    • View Profile
Re: Weapon Discussion Thread
« Reply #64 on: August 23, 2016, 06:16:47 am »
@Star_Treasure: No, unfortunately snap shots are only one shot. It's not something modders can alter.

@Surrealistik:
The humanist super soldiers cannot react, their weapons only have autoshots. It was not a fluke.

As for the reliability of HMG type weapons, given the likelihood of getting screwed with the 0-200% rule, which is what these weapons are using by default and what prompted the comment. It's false to claim that the sniper is more reliable. Until you reach 100% (which means firing acc = 100 on a regular sniper rifle), you actually have more odds to miss your one shot because of the quadratic dependency on accuracy of the sniper rifles and the forgiving nature of the HMG.

Ex.: at acc = 99
- Sniper rifle accuracy is 0.01*99^2 = 98%, so 2% to miss
- HMG chance to hit is 49.5%, chance to miss is 50.5%, to the 8 for the chance of missing all 8 is 0.42%

Once you factor the chance of the RNG screwing you over on the damage roll, it becomes even more significant.

Still at 100 acc:
- Sniper rifle with Normal-PS ammo: 30-36+100/4 = 55-61 damage
- HMG with Normal-PS ammo: 55-65 damage

The damage being similar, you have similar odds of rolling under the armor value. But if you crank up the accuracy of the firer beyond 100 to increase the damage of the sniper (thus reducing the odds of not doing damage), the odds of missing all HMG shots become even smaller, the odds of hitting with multiple shots increase, and the HMG pulls even further ahead. The vulkan pushes the advantage with the 20 shots, despite the reduced accuracy.

Sure, if you make the sniper rifles more dependable, they get more dependable (but they lose some of their ability to roll really high, which is where they shine with very very high accuracy soldiers to overcome very tough enemies).

With the current rule, you are more likely to get screwed by the RNG on your one damage roll with a sniper rifle than to miss all and/or get screwed on multiple shots of the HMG. You hit often enough that some of the hits will likely roll you good damage, plus each hit does a little bit of armor damage, helping the next to punch through.

The niche of the sniper rifle is with really good accuracy soldiers to push the damage really high, but that is rarely required since the Vulkan can whittle down common power armored foes reliably. Or those trick shots where you have soldiers in the zone, which given your preference for smoke and snipe, shouldn't be much of an issue (it isn't even for me, and I play pretty aggressively with ~50% melee soldiers). So that leaves reaction fire, but with the decreased accuracy and still high TU cost, it's not great either.

The hunting rifle and the like are great for training gals (one hit, regardless of damage, is a chance at a stat increase) and reaction fire (they tend to perform better with snapshots, although the damage is worse), the HMG does better with applying damage at a distance (and I enjoy blowing away the cover too, without blowing away the items like with HE damage).

Anyhow, that's my experience and the statistics to back it up. If you are enjoying the sniper rifles, great for you. I just wanted to pitch in for those weapons which, on paper, might not look too great (crap listed accuracy and really heavy) but can also be really useful (and lead to maniacal laughing while blowing stuff away).

Offline Surrealistik

  • Colonel
  • ****
  • Posts: 484
    • View Profile
Re: Weapon Discussion Thread
« Reply #65 on: August 23, 2016, 06:27:56 am »
@ Arthanor:

I agree that if you want to reach out and murder something on the default ridiculous 0-200% spread the HMG and Vulcan are better than sniper weapons, yes, assuming you can afford wayward shots (again, the scatter risk is real, particularly for the Vulcan, even if you ultimately hit and drop the target). The statistics are known to me, hence the statement that the likelihood of missing all shots is very small.

For gaussian distributions though (which is how damage should be calculated IMO and what I play with personally, and what I was assuming per my prior mention of it), Sniper class weapons all the way, particularly the upper echolon sniper weapons such as the Sniper Gauss, which will reliably one shot anything that doesn't have insane Piercing resist + armour.
« Last Edit: August 23, 2016, 06:46:59 am by Surrealistik »

Offline Star_Treasure

  • Captain
  • ***
  • Posts: 96
    • View Profile
Re: Weapon Discussion Thread
« Reply #66 on: August 23, 2016, 10:27:41 am »
The biggest advantage of HMGs over Sniper rifles is that they won't be frustrated by chain link fences. Heck, you don't even need line of sight. Just a general idea of where an enemy might be.

Offline hellrazor

  • Commander
  • *****
  • Posts: 2013
  • Deep Ruleset Digger & Bughunter
    • View Profile
    • Github Account
Re: Weapon Discussion Thread
« Reply #67 on: August 23, 2016, 10:44:47 am »
Sorry but isn't the Autocanon kinda the heavy machine gun?

I personally employ (outside of piratez in my own mod), Knives/Sword, Pistols, Shotguns, Rifles, Sniper Rifles, categorywise.
I already saw the implementations of the Heavy Machine Guns in other ModPacks, but i personally find those weapons how they are balanced to cumbersome and to uneffective.

Sorry to intrude...

Offline Solarius Scorch

  • Global Moderator
  • Commander
  • *****
  • Posts: 11408
  • WE MUST DISSENT
    • View Profile
    • Nocturmal Productions modding studio website
Re: Weapon Discussion Thread
« Reply #68 on: August 23, 2016, 12:08:40 pm »
What about over penetration? Surely bullets ought to be able to pass through a fence with only minimal reduction in power.

That would be interesting mechanics, but completely new. Possibly deserving a new thread at least. I'd be happy to discuss it.

the default ridiculous 0-200% spread

Well, there are two big advantages of this system:
1) getting hit with a bullet and not wounded much is pretty normal in RL. Same with most other types of damage. I like how bullets are not death rays in X-Com - sometimes they hurt, sometimes you're lucky, sometimes you get a headshot.
2) While Piratez are more extreme, in vanilla you can pretty much kill everything with anything. Dioxine told me how in the original 1994 version his base was attacked by Mutons and he only had access to early weapons, because of a bug that prevented him to scroll the items on the ground to get to plasmas. So he managed to win against Mutons with rifles and cannons. It would be pretty much impossible with Gaussian distribution, at least within a reasonable time.

I'm not sure which system's advantages prevail, but I'd personally stick with what I know.

Offline Arthanor

  • Commander
  • *****
  • Posts: 2488
  • XCom Armoury Quartermaster
    • View Profile
Re: Weapon Discussion Thread
« Reply #69 on: August 23, 2016, 09:48:08 pm »
@Surrealistik/Solarius: Alright, I must have misunderstood that the comment was strictly reserved to Gaussian damage. I'm not a specialist on how it is to get shot (and how often it doesn't hurt), but as a player I intend to follow the modder's choice of damage randomness, especially since Dioxine changes it per weapon to create the experience he intends. I fully respect tweaking master mods to better fit what is enjoyable for yourself though, and I have a "Piratez tweaks" mod of my own to do that.

I used to play with the TftD damage rule for everything because I like my hits to count (it's already hard enough to hit sometimes in vanilla...). That was before I saw a comment from Dioxine on how it made low armor values useless and high armor values too good, which I agree with, so I reverted back to the 0-200%, especially for Piratez. I feel like the Gaussian damage, especially once  you reach high damage values (ex.: sniper gauss) becomes too reliable and too good. Essentially you recover the reliability of TftD damage rule, plus the value is so high that pretty much everything becomes "low armor value".

Sorry but isn't the Autocanon kinda the heavy machine gun?

I personally employ (outside of piratez in my own mod), Knives/Sword, Pistols, Shotguns, Rifles, Sniper Rifles, categorywise.
I already saw the implementations of the Heavy Machine Guns in other ModPacks, but i personally find those weapons how they are balanced to cumbersome and to uneffective.

Sorry to intrude...

No need to apologize! All able bodies are welcome in the Piratez' corner. We be equal opportunity folks! (although I've yet to see a male soldier...!)

The autocannon, as implemented in vanilla, is quite a different beast from the HMG and the like in Piratez/FMP. What you saw of HMG in "other ModPacks" (assumed to mean the FMP) is probably closer. I'm a big fan of the autocannon in Piratez, since it's got good damage, decent enough accuracy (with blast ammo) and a nice variety of ammo (personal favorite: poison gas, followed closely by HE. I wish there were chem and stun gas too!). But in Piratez the autocannon is more of a short range weapon due to its limited accurate range, whereas the HMG type weapons are for long range mayhem and damage projection (provided you can afford to risk some scatter, but not as bad as shooting a full on rocket and blasting everthing away).

Offline Surrealistik

  • Colonel
  • ****
  • Posts: 484
    • View Profile
Re: Weapon Discussion Thread
« Reply #70 on: August 24, 2016, 02:26:43 am »
In general, the only way a bullet is going to do truly negligible damage if it connects is if it grazes you or deflects off something; the probability of this vs say a leg or body shot is substantially smaller. Gaussian distributions make way more sense than flat distributions because it accurately reflects the probability of getting fucked up by a bullet (or worse!) to the probability of not getting fucked up or say shot through the heart/brain. The only argument I can really see for linear distribution 0-2x is that of gameplay and the argument that high variability prevents armours from being too effective, but even there, Gaussian distributions are not quite like TFTD in that you can still get ridiculous extremes, but they're just less likely as opposed to outcomes being clamped to a range. From first hand experience, even with a very conservative playstyle and running full plate, I'm still eating occasional damage from las/heavy conventional weapons, including against frontal armour on tower shield gals, so those outliers are still a thing I have to consider and concern myself with.

Offline Dioxine

  • Commander
  • *****
  • Posts: 5412
  • punk not dead
    • View Profile
    • Nocturnal Productions
Re: Weapon Discussion Thread
« Reply #71 on: August 24, 2016, 02:51:25 am »
This is all true only if we agree that accuracy model and armor model are 100% accurate, which they aren't. In normal conditions, there is no such a thing as absolute accuracy, or armor free of any weak spots. How does it relate to damage? Bear with me. While the bullet damage in real cases follows gaussian distribution, accuracy of a shot is not linear (it only can approach 1.0, never achieve it), as a gust of wind, sudden target movement, or a twitch of shooter's muscle causes an otherwise sure shot miss. Likewise, a lucky shot can score damage on an otherwise impervious target; maybe not against powered armor, which is fully enclosed, but certainly against non-powered ones. Since neither case is accounted for in the engine, flattening of the Gaussian distro of bullet damage seems to be the only way to account for them. Most no-damage shots account for actual misses, while most of hi-damage shots account for hitting a particularly weak spot.
Systems which account for this (and therefore should have Gaussian distro of projectile damage) use various tables of critical hits and critical misses, which XCom doesn't have; and, if critical hit/miss mechanics are linked directly to skill/accuracy, some sort of accuracy cap is employed (like 95% cap in Fallout).

Offline Surrealistik

  • Colonel
  • ****
  • Posts: 484
    • View Profile
Re: Weapon Discussion Thread
« Reply #72 on: August 24, 2016, 03:14:14 am »
This is all true only if we agree that accuracy model and armor model are 100% accurate, which they aren't. In normal conditions, there is no such a thing as absolute accuracy, or armor free of any weak spots. How does it relate to damage? Bear with me. While the bullet damage in real cases follows gaussian distribution, accuracy of a shot is not linear (it only can approach 1.0, never achieve it), as a gust of wind, sudden target movement, or a twitch of shooter's muscle causes an otherwise sure shot miss. Likewise, a lucky shot can score damage on an otherwise impervious target; maybe not against powered armor, which is fully enclosed, but certainly against non-powered ones. Since neither case is accounted for in the engine, flattening of the Gaussian distro of bullet damage seems to be the only way to account for them. Most no-damage shots account for actual misses, while most of hi-damage shots account for hitting a particularly weak spot.
Systems which account for this (and therefore should have Gaussian distro of projectile damage) use various tables of critical hits and critical misses, which XCom doesn't have; and, if critical hit/miss mechanics are linked directly to skill/accuracy, some sort of accuracy cap is employed (like 95% cap in Fallout).

I'd say the Gaussian model vis a vis flat distribution accounts for the outliers of a otherwise sure shot glancing/grazing/missing or a lucky bullet through the brainpan pretty well; they're possible but _rare_ which is how it should be. Even if you want to argue that accuracy should be capped, fine, put it at 95%, 99%, 99.99%, etc... but in no way does theoretical 100%s (which aren't really) justify a flat distribution. You'll never achieve a perfect representation of reality, but Gaussian in my view cleaves closer to IRL than flat distributions, even relative to the context of the other mechanics.
« Last Edit: August 24, 2016, 03:18:44 am by Surrealistik »

Offline Dioxine

  • Commander
  • *****
  • Posts: 5412
  • punk not dead
    • View Profile
    • Nocturnal Productions
Re: Weapon Discussion Thread
« Reply #73 on: August 24, 2016, 03:55:32 am »
You'll never achieve a perfect representation of reality, but Gaussian in my view cleaves closer to IRL than flat distributions, even relative to the context of the other mechanics.

In contrast, I think the critical fail/success elements weigh on the model enough for gritty linear damage distro to be closer to reality than idealized Gaussian damage distro.
Since the mod was supposed to be rather gritty than idealized, there's no counterweight to support the idealized approach. However I'm not telling anyone how to play - I can only tell how the mod is supposed to be played.

Offline Surrealistik

  • Colonel
  • ****
  • Posts: 484
    • View Profile
Re: Weapon Discussion Thread
« Reply #74 on: August 24, 2016, 04:00:26 am »
In contrast, I think the critical fail/success elements weigh on the model enough for gritty linear damage distro to be closer to reality than idealized Gaussian damage distro.
Since the mod was supposed to be rather gritty than idealized, there's no counterweight to support the idealized approach. However I'm not telling anyone how to play - I can only tell how the mod is supposed to be played.

What's idealized about it exactly? And how is a model which says that a brain buster is as likely as a body shot closer to reality than a model which says the former is significantly less probable than the latter? Even if your issue is with perfect accuracy not being a thing, that's a separate issue with a separate solution.