|
Coolguye posted:hrm. i'm actually curious how the og xcom handled dispersion. i doubt seriously that had any loving around with arcminutes or whatever, but it worked great for the most part and was quick enough for freaking 486 computers. Wouldn't be surprised if they used arcminutes or miliradians or some other angular measure. The math seems super simple since you can just multiply by the distance from target and a random number to get the size of the miss.
|
#
?
Sep 15, 2017 18:01
|
|
|
|
| # ? Apr 20, 2024 03:28 |
|
|
esquilax posted:Wouldn't be surprised if they used arcminutes or miliradians or some other angular measure. The math seems super simple since you can just multiply by the distance from target and a random number to get the size of the miss. looking it up, it seems like they basically had two more or less standard dispersal cones (which almost certainly did involve a little angular math, but it was a single calculation), one of which was a 'hit' cone that deviated only slightly from the enemy's middle, and another of which was a 'miss' cone that could basically go anywhere in a 90 degree arc in front of the shooter. the listed accuracy was merely the chance that the 'hit' cone would be selected, and from there the game just rolled some dice on what the bullet's actual heading was when leaving the weapon and drew a vector from there. this makes a ton of sense on a ton of levels but is likely not terribly appropriate for DDA.
|
|
#
?
Sep 15, 2017 18:06
|
|
|
Strudel Man posted:I have to say, as a personal thing, I really hate feeling that every single use of a weapon is a step towards using it up completely. Having occasional damage randomly strike, and which can be made vanishingly rare by having sufficient skill, feels a lot more bearable. More like an unfortunate occurrence than the relentless march of entropy. the idea would be that you have the same breakpoints and math, just normalized to a per swing. so some examples... take a survivor with 12 str, 12 dex, 4 melee skill swinging a katana your stat factor = 1/2 dex + melee skill + (64 / (whichever's bigger, str or 4)) = 6 + 4 + 5.333 material factor is just the chip resistance, or 20 for steel (iirc this also scales with wear, so reinforced items have a higher resist and damaged ones have a lower, but set that aside for now) wear multiplier looks like it's always 1? the chance to be damaged is one in (stat factor * material factor / wear multiplier) with a durable_melee only breaking a quarter of the time that works out to about a 1 in 1,226.66 chance of being damaged to marked so take that number and multiply by five, and that means you get 6,133 swings until the weapon is broken and destroyed. Normalizing would mean each swing would burn up only .016% durability. a survivor with 10 str, 10 dex, 0 melee skill swinging a cudgel would work out to stat factor of 5 + 6.4 = 11.4 material factor of 10 for wood chance to be damaged to scratched of 1 in 114, with 570 swings until the weapon is broken and destroyed so each swing would burn up .18% durability shifting it to the flat amount every single swing would change nothing, statistically, about how quickly weapons break, since i still want the performance reductions to only kick in at the 20% breakpoints. it would do two things, though: first, it makes the actual important end result to the player obvious and displayable, since you can see this item costs .02% per swing while this other is .2%. people can dig into the wiki or whatever to see WHY the chance is what it is (and there's discussion to be had about the formula itself), but if you could just look at something and see exactly how rugged it is, there'd be a lot less mystery about things like how multiple materials play into things. second, it shifts the repair system from repairing broken things to maintaining things so they don't break. your katana doesn't suddenly chip and get bad, you swing with it a thousand times and bring it down to 85% durability, then get back to your base and spend a bit of time/materials maintaining it and bringing it back up to 95% durability without it ever dropping in performance.
|
|
#
?
Sep 15, 2017 18:41
|
|
|
I'm a big fan of the game and what you guys are doing with it! I have no coding experience to offer, but I'm a fair hand at proofreading and language cleanup if you guys have any copy you need reviewed. That, or casual playtsting. If not, cheers! I'm looking forward to the new version. I'm so spooked on the lovely vehicle changes that I've been playing 0.C for a loooong time.
|
|
#
?
Sep 15, 2017 19:46
|
|
|
the worst of the vehicle changes in the CleverRaven mainline were eventually backed out. in the current experimental you can do pretty much whatever you want and it's fine. certain designs no longer work (like i used a 5-wide warbike setup where the front wheels were all road rollers, in the current version this makes your vehicle untenably slow) but in general it's fine. this is one of the only fits of intelligence that have come out of that repo though.
|
|
#
?
Sep 15, 2017 20:07
|
|
|
Anticheese posted:You can charge your bionics from any kind of car battery if you take it out, examine it, and 'E'at it. The functionality is there, but the way to get to it could sure be streamlined! The one I coded let you plug into a car and drained the battery to recharge yours, shutting if you weren't directly next to the car. I might have to dig it up. Killer-of-Lawyers posted:Actually there's a bionic for charging directly from cars now. It's pretty cool, you basically have a port on your hip you plug into with jumper cable leads on the far end. I found it in a game but never got a chance to use it before stupidly trying to take out a zombie grenadier. You only really need to fail at killing a C4 hack once before you're SoL. oh lol someone else did it anyway
|
|
#
?
Sep 15, 2017 20:20
|
|
|
Solid Poopsnake posted:I have nothing helpful to contribute because I am a worthless piece of poo poo but I want to say that this all makes me very excited. Same. Except I don't really hate the thirst mechanic for what that's worth.
|
|
#
?
Sep 15, 2017 22:38
|
|
|
I typed up a thing when it comes to Cataclysm core combat mechanics. IT's caffeeine-fueled and not nearly as smart as I imagined when I was at work, but have a look.
|
|
#
?
Sep 15, 2017 23:57
|
|
|
mormonpartyboat posted:the idea would be that you have the same breakpoints and math, just normalized to a per swing. so some examples... Strudel Man fucked around with this message at 00:25 on Sep 16, 2017 |
|
#
?
Sep 16, 2017 00:20
|
|
|
Grenaders are fine as long as you run when the grenade hack beeps at you. The problem is that the message is easy to miss and its not intuitive that a arming beep woukd be based on your move speed.
|
|
#
?
Sep 16, 2017 00:29
|
|
|
I'd just like to see the game respect the player's time, too. I loved the game when I discovered it and subsequently binged, but I just realized that I was probably spending 50% of my time (real world) managing and juggling inventory, etc. I don't think there's a single easy fix for that kind of thing, but even if you kept food and water but made the buffer larger - not even making food/water more efficient, but just have to do the requisite steps / victory dance less often that'd be good. Don't get me started on how much time I wasted making a farm. In-game-time I'd say looking what skills scale well and which ones don't for batch jobs. Cooking definitely should. I think it'd be complicated, but I'd like to see cooking recipes be a little less wooden. Right now the variety is large, but half of the recipes go away if you're missing a single ingredient. Maybe if individual food types were in 'categories' like "starch" "protein" and the actual requirements could take from any of these. This would be a bit like the crafting in Wayward for example. Seconding whoever suggested gunpowder (and presumably casings) beoing able to be made or crafted on some level without mods. Just because it'd be more fun. Balance accordingly.
|
|
#
?
Sep 16, 2017 01:49
|
|
|
yeah, the number one way to handle inventory better would be allowing you to put stuff in actual containers instead of just on your person, so you could, for example, carry around a sherpa bag of poo poo and then just drop it and grab it as needed when things get hot. but that's been a target for a long time at this point. there isn't a good way to allow that without making inventory management more annoying up front. that said, there's a couple things you can/should do. automatic eating and drinking as mentioned should happen during extended crafting, but if that happens then you should just provide a one touch button to drink until slaked/eat until full with the stuff you have on hand, avoiding 'dangerous' comestibles (unpurified water that could ostensibly poison you, food you're allergic to, rotten poo poo). in addition to some smarter stacking that you could make the situation better, though you're not going to really fix it without a more complete approach, imo.
|
|
#
?
Sep 16, 2017 06:09
|
|
|
I don't think there's necessarily anything wrong with calculating dipersion in arcminutes but it could do with a more useful display to the player about how accurate they can expect to be.
|
|
#
?
Sep 16, 2017 06:21
|
|
|
Yeah, the arcminutes is a purely internal thing. I don't exactly know that I love a purely dispersion-based mechanic, but however one feels about them, the units it's measured in are surely irrelevant.
|
|
#
?
Sep 16, 2017 06:32
|
|
|
Coolguye posted:yeah, the number one way to handle inventory better would be allowing you to put stuff in actual containers instead of just on your person, so you could, for example, carry around a sherpa bag of poo poo and then just drop it and grab it as needed when things get hot. but that's been a target for a long time at this point. there isn't a good way to allow that without making inventory management more annoying up front. Separate inventories for containers would be neat but it would also make the current inventory UI even harder to navigate, it would have to come with a full UI overhaul really. You could get a similar effect with two simple changes though; the ability to mark items as important so they're dropped last when carry capacity goes down and make it so if you pick up a piece of clothing with volume picking up other items takes no time up to the capacity of that clothing. OwlFancier posted:I don't think there's necessarily anything wrong with calculating dipersion in arcminutes but it could do with a more useful display to the player about how accurate they can expect to be. It's definitely possible to calculate a % chance to hit from dispersion. The real problem with dispersion is that it makes range pretty much trump every other concern, if you try to make a gun that's decent at medium range it becomes pretty much pinpoint accurate at short range which is really hard to balance. The current builds try to help this with aiming time which is a decent idea, too bad they hosed it up by obfuscating any actual usable information from the player. If I remember right you have like a stability meter and a confidence meter or something which means ???
|
|
#
?
Sep 16, 2017 06:40
|
|
|
it isn't irrelevant from n engineering perspective. a dispersion-based firing formula is most 'real', but the problem is that it's difficult to model it on actual data. there's no decent datasets on, for example, the dispersion vectors from rookie, intermediate, and advanced marksmen in various firearms to provide a good way to provide test data for the system. as a result, you make poo poo up as you go and the entire thing becomes a big massive "ehhh, that looks close enough" problem that feels weird to use and never quite plays right. this is the problem with enumerating the system in arcminutes, you end up making it so granular that there's no way to program the system correctly. it is useless to make a system that accurately models reality if you don't have a corpus of data that allows you to configure what reality looks like. not that i know a better way to do it right now mind you, but the disgust is not wholly misplaced. Dire Lemming posted:Separate inventories for containers would be neat but it would also make the current inventory UI even harder to navigate, it would have to come with a full UI overhaul really. You could get a similar effect with two simple changes though; the ability to mark items as important so they're dropped last when carry capacity goes down and make it so if you pick up a piece of clothing with volume picking up other items takes no time up to the capacity of that clothing my larger thought would be to somewhat automate the question of what goes where by a priority list of various headings mapped to lower or higher encumbrance gear. so for example you could say that weapons belong in the lowest encumbrance spot you can fit them in, medicine is next, all the way down to food, water, and junk explicitly needing to be in the highest encumbrance container you have. this is a pretty coherent way to say something like 'my guns, ammo, knives, and bandages should go on my tactical vest and duster, but my 20 lbs of canned food and 10 lbs of clean water should go in my backpack where i can shrug it off quickly.' but even that sort of thing is just a sane default, you'd still have to let people manually move junk around if they wanted to, and tbh the minute they have to move junk manually you've lost any benefit you might have gained from making individual containers matter. Coolguye fucked around with this message at 06:46 on Sep 16, 2017 |
|
#
?
Sep 16, 2017 06:41
|
|
|
Coolguye posted:it isn't irrelevant from n engineering perspective. a dispersion-based firing formula is most 'real', but the problem is that it's difficult to model it on actual data. Dire Lemming posted:It's definitely possible to calculate a % chance to hit from dispersion. The real problem with dispersion is that it makes range pretty much trump every other concern, if you try to make a gun that's decent at medium range it becomes pretty much pinpoint accurate at short range which is really hard to balance. The current builds try to help this with aiming time which is a decent idea, too bad they hosed it up by obfuscating any actual usable information from the player. If I remember right you have like a stability meter and a confidence meter or something which means ??? Strudel Man fucked around with this message at 06:56 on Sep 16, 2017 |
|
#
?
Sep 16, 2017 06:51
|
|
|
a larger unit of measure (like a degree) causes so much more swing in the overall model generated by the system that it is much, much easier to fake something reasonable simply because dispersion algorithms that make presumptions different by a substantially large unit of measure. i could set up an in game test with a bunch of different degree assumptions and fake a reasonable model fairly quickly. with a very small UoM that's much harder (to the point where it becomes a futile exercise eventually, such as in this case). you can't just hamfistedly convert the small units into larger ones either because angular calculations are affected by being enumerated in the smaller units, so you still get a lot of leftover mathematical effects that end up being significant to the operation. this is precisely why financial systems arbitrarily say they're not interested in anything below a certain denomination (say, a tenth of a cent) - the calculations get hosed over multiple trials if you don't stop the the presumptions somewhere. this shouldn't be read as me saying we should change it to something bigger like degrees though, that makes long-range combat really weird and it likely wouldn't work at all for rifles; you'd just not be able to hit poo poo with a rifle until one magical moment you suddenly became a stone cold deadeye.
|
|
#
?
Sep 16, 2017 07:03
|
|
|
Well, obviously with the larger units you'd have to have decimals for it to work at all sanely. 
|
|
#
?
Sep 16, 2017 07:08
|
|
|
Really I don't know about how accurate to reality dispersion models are. I'm sure people decided on a dispersion model using arcminutes for Cataclysm because arcminutes are what gun companies use to advertise the accuracy of their guns and that's data you can just plug into the game but arcminutes don't really describe the human element of shooting accuracy well. A person who misses by 2 arcminutes at 50 metres won't necessarily miss by 2 arcminutes at 100 metres or 10 metres.
|
|
#
?
Sep 16, 2017 07:09
|
|
|
That'd be arguable, I think. At least for stationary targets, it sorta seems like that should be relatively consistent, since angular distances are what we most fundamentally perceive...it does ignore motion though, as well as more psychological effects of being up close and personal with a target vs sighting them from a hundred meters away.
|
|
#
?
Sep 16, 2017 07:12
|
|
|
Strudel Man posted:Well, obviously with the larger units you'd have to have decimals for it to work at all sanely. which basically means you should be using a smaller denomination to begin with to avoid floating-point arithmetic, which is slow, sometimes unreliable, and generally awful on computers. even in the one tenth of one cent example, the way you'd model that is to just go int centiCents; //this represents one tenth of one cent per unit and go from there. Dire Lemming posted:Really I don't know about how accurate to reality dispersion models are. I'm sure people decided on a dispersion model using arcminutes for Cataclysm because arcminutes are what gun companies use to advertise the accuracy of their guns and that's data you can just plug into the game but arcminutes don't really describe the human element of shooting accuracy well. A person who misses by 2 arcminutes at 50 metres won't necessarily miss by 2 arcminutes at 100 metres or 10 metres.
|
|
#
?
Sep 16, 2017 07:17
|
|
|
Coolguye posted:yeah it's hard to say without a corpus of data, which to my knowledge doesn't exist, but it models the bullet trajectory perfectly which is honestly the problem you're trying to solve for here. the human element of anything becomes such a clusterfuck that it's not worth modeling correctly, see also my ridiculously dumb posts itt about actually modeling nutrition and body mechanics with regard to microdamage from the very hard life you're leading in DDA. the same sorts of questions come up when considering the influences of the human element in shooting. I'm not trying to say that we should try to model the human element. My point is that the reason dispersion was chosen over a simple % chance to hit with modifiers was because it's more accurate/realistic. If it happens to not actually be realistic why not just go back to % chances to hit and maybe have a simple dispersion model for misses or something. Plus % chances are easier to balance and understand, realism be damned.
|
|
#
?
Sep 16, 2017 07:30
|
|
|
well, mostly because you end up right back at the bullet trajectory problem the second a miss happens. like what happens on a miss when you're throwing bullets at a deadhead? does it just pass through the target, and presumably have some sort of % chance to hit everything behind it? does it just disappear? what happens to it? there isn't really a good answer to this question without actually modeling the bullet and letting it go wherever its momentum takes it, which...is an angular equation that results in a vector, right where we started. :/
|
|
#
?
Sep 16, 2017 07:35
|
|
|
But that's fine, misses don't have to be balanced, they don't have to be inherently understood by a player. Don't remove dispersion, just don't make the player have to make decisions based on it. Say I'm shooting a zombie 10 tiles away which I'll just assume is 20 metres, the game should tell me that because of my 5 rifles and 4 marksman skill I have a 60% chance to hit, because I'm 10 tiles away I get +10% and because I took a turn to aim I get another +10% giving me an 80% chance to hit. If I shoot and roll a miss, then the dispersion calculation happens and it works out that to miss a 1 metre target at 20 metres I have to miss by ~2.5 degrees so it calculates the trajectory with a 7.5 degree dispersion. Maybe I hit the zombie anyway but I probably don't and it really doesn't matter, it's just a neat thing that the game does, not something core to it's balance which should be easy to understand as a player and adjust as a developer. Edit: Actually I just realised I'm basically saying do what OG Xcom does, if you roll a hit calculate a hit trajectory, if you roll a miss calculate a miss trajectory, don't calculate a trajectory and decide hit or miss from that. Dire Lemming fucked around with this message at 07:53 on Sep 16, 2017 |
|
#
?
Sep 16, 2017 07:50
|
|
|
I'm not convinced we need to 'realistically' figure out where a bullet goes at all. 90% of the time if you miss, you don't really care what happens to it - the other 10% you're either firing into a target-rich enviroment or using something with AoEs. What I'd go for is a more conventional RPG-style percentage roll for hitting your target, like the new XCOM uses. This is what we balance for combat effectiveness. I typed up a big gdoc last night, fueled by irritation and caffeeine, but honestly, the specific calculation doesn't matter as long as it's easy to tweak for the difficulty/balance/theme we want. If you miss? You could code in a simple scatter algorithm. Say, make it roll for a random direction. Then roll a random distance up to, say, 10% of the distance between the shooter and the target. That's where the shot actually ends up going for, and if it passes thorugh/ends up in a cell with an obstacle or creature in it, it rolls to see if it hits them. Done. The more simple this stuff is, the less time we'll need to spend tweaking the poo poo out of them.
|
|
#
?
Sep 16, 2017 12:06
|
|
|
Strudel Man posted:No, I get that you can make them on average mathematically identical, but the implementation still "feels" different, and in a way I find to be considerably worse. Like how it feels different to lose a guaranteed $5 than to have a 50/50 chance of either winning $10 or losing $20. to me it always felt more like you're betting $5 and you've got a tiny chance to lose it and a huge chance to get your $5 back. and then when you try to repair your thing you're betting nothing, with a tiny chance to make back the $5 you lost earlier it just seems super not great either way, so i was at least hoping to change repairing into maintenance
|
|
#
?
Sep 16, 2017 14:35
|
|
|
Coolguye posted:which basically means you should be using a smaller denomination to begin with to avoid floating-point arithmetic, which is slow, sometimes unreliable, and generally awful on computers. even in the one tenth of one cent example, the way you'd model that is to just go int centiCents; //this represents one tenth of one cent per unit and go from there. Good lord no this is just an argument to use double precision. This code is running on a big boy computer it can do big boy math.
|
|
#
?
Sep 16, 2017 14:50
|
|
|
Dire Lemming posted:I'm not trying to say that we should try to model the human element. My point is that the reason dispersion was chosen over a simple % chance to hit with modifiers was because it's more accurate/realistic. If it happens to not actually be realistic why not just go back to % chances to hit and maybe have a simple dispersion model for misses or something. Plus % chances are easier to balance and understand, realism be damned. that's kinda the problem, it's not accurate/realistic. you can get hard Actual Real World Dispersion Numbers in MOA, which will often come with a gun's paperwork. but absolutely most importantly, that number assumes a completely fixed weapon. you fix a 1 MOA gun to a rigid firing platform and put 5 bullets through it, it'll average a 1 inch group at 100 yards, which is way farther than you can see in game. but once you put it into a person's hands, their personal skill is so dominant that it really doesn't matter how accurate the gun is until the person has enough training to be like a competitive shooter. and then, there's all kinds of milsperg bullshit that would actually have relevance before a 1 MOA gun and a 1.5 MOA gun are apparently different. and all that's before considering the game itself, which has survivors who are usually idiot children who have never seen a car before and struggle to boil water without burning it picking up guns and shooting giant hulk zombies from 20-30 feet away. i think building something from the ground up is probably the way to fixed ranged combat, i'll try to throw something together as an aside: it shouldn't be possible to miss in melee, only for the target to dodge or block
|
|
#
?
Sep 16, 2017 15:09
|
|
|
Would it not be possible to keep the current dispersion system and add some modifiers on top based on shooter skill and what you're shooting? Like if you're bad at shooting it throws off your aim slightly or you get worse recoil control, if the target is something fast or insubstantial it gets a penalty to hit, if it's at point blank and aware of you it gets a penalty which is bigger the heavier the gun is, that sort of thing? Like the current system seems fine for modelling the physical accuracy of the weapon, but if it doesn't properly take into account shooter skill then a shooter skill modifier layer would seem to be the solution unless I'm misunderstanding? I mean that's how it actually works in real life so I would have thought that should give you an intuitive result as long as the modifiers are clear to the player. OwlFancier fucked around with this message at 17:21 on Sep 16, 2017 |
|
#
?
Sep 16, 2017 17:19
|
|
|
just as a note, neu-xcom does still use a dispersal equation for its misses and stuff, this is why a soldier can shoot at and enemy in cover, miss, and occasionally destroy the target's cover.
|
|
#
?
Sep 16, 2017 17:28
|
|
|
It's still an abstracted dice roll though, it doesn't really model projectile trajectory. You can never miss an enemy and hit another enemy (or any other object) by accident for example. AFAIK anyway.
|
|
#
?
Sep 16, 2017 17:40
|
|
|
you will never have damage register on another actor but it is completely possible to whiff someone and hit an explodable environment object well behind them. hell i did this just last night during a Lost mission and it caused another damned mess of the zombies to spawn. took a risky shot at a sectoid behind low cover, he ducked, and i tagged a car a good 5 squares behind the x-ray, which caused an explosion and whoops more lost (gently caress)
|
|
#
?
Sep 16, 2017 17:46
|
|
|
Huh. I haven't played for ages, I must've remembered wrong.
|
|
#
?
Sep 16, 2017 17:58
|
|
|
My background is C++ but it's been several years since I did it professionally. Once i get a chance, I'll download and set up the project and have a poke around, maybe I can help out. I'd certainly like to help do QoL improvements to the UI as it is one area that has had significant improvements over time compared to how it was when I started playing it, but could still stand some improvements.
|
|
#
?
Sep 16, 2017 18:31
|
|
|
Coolguye posted:just as a note, neu-xcom does still use a dispersal equation for its misses and stuff, this is why a soldier can shoot at and enemy in cover, miss, and occasionally destroy the target's cover. Yeah I really think personally that the biggest issue with firearms usage is that it's not at all clear how effective you can expect to be on target. A percentage to hit indicator and perhaps a series of coloured dots along the fire path to show accuracy on that target if it was closer or further away would help you to get a better feel for how your gun performs in various situations, and that'd be super helpful in making them more intuitive.
|
|
#
?
Sep 16, 2017 20:18
|
|
|
One of the problems is that there's information in the game when you aim, but it's not explained very well. The bar fills up, but there's no explanation of what it means. There's accuracy stuff, but you gotta look up to see if bigger numbers are better or worse. There's probably a more intuitive way to present all this. Even just putting something like (Lower is better) by the arcminutes would be a step up. There's a lot of stuff that should be presented better. The text log needs an effect for those important little tells, that let you know if you're infected with parasites, or when a grenade hack beeps at you. Making it pop up on the main screen like the damage amounts in red text would probably help.
|
|
#
?
Sep 17, 2017 17:32
|
|
|
Honestly long-term hazards and inconveniences like parasites, bites becoming infected, catching a cold, and the like should all bring up an interrupt dialogue box that tells you what just happened and, if applicable, what you should do about it.
|
|
#
?
Sep 17, 2017 18:45
|
|
|
The (original) Xcom method of gun firing is so venerable as to be derived from the actual literal grognard standard for dispersal. It can be bullet science if you want it, and in simulation engines like ARMA or whatever should be more complicated since there is an aspect of animating out the bullet. But its turn based, you get to cheat and trace paths beforehand and poo poo. The preponderance on first principles is total sim chaff and well into the territory of funner to design and program than to play with.
|
|
#
?
Sep 17, 2017 19:02
|
|
|
|
| # ? Apr 20, 2024 03:28 |
|
|
Angry Diplomat posted:Honestly long-term hazards and inconveniences like parasites, bites becoming infected, catching a cold, and the like should all bring up an interrupt dialogue box that tells you what just happened and, if applicable, what you should do about it.
|
|
#
?
Sep 17, 2017 19:18
|
|