Nodrak

I showed in this post that shooting into smoke, shows that: Smoke Reduces 'Stopping Chance Component' The 2 roll system works as I expect it to in many cases, otherwise 3 tests of 10 shots would be 2 success, not 4. After that, I searched for a better test case, which resulted in the 'edge of the map' issues. Then I later tested that the reduction in Hit Chance for smoke from misses, only applies if the unit hit is inside the smoke. (The 7% re-roll was calculated NOT the 7% main hit, but I seem to be messing up the calc in my head right now, I get a slightly different, number, but its not important) So, then shooting out of the smoke, we encounter the 'Target uses Hit% instead of Miss%' issue, even with 100 Stopping Chance.: We can just shoot anything in-front of it to kill it, since we have 100% miss hit chance: (or we can shoot behind it because Accurate Misses will prioritize the highest cover) This shows that 'Real Misses' use stopping chance, probably, but only when the target is not the target. It seems the game wants us to get to the bottom of this:

Another example came up in normal play, I opened a door and saw a cyber-drone. All 3 Marines took a 35%+ shot at it (expected 3.5 hits), and all 3 marines hit the shield at the door, killing him: Shot 1: The tile next to the doorway took a hit alien took 2 hits, cover in-front of it took 1+, wall behind took 2+ Second shot: Essentially the same result, less hits on enemy than expected (3 expected min), my solder ducked the wall eats another shot (25% normal setting): These shots in testing vs no nearby colliders were laserbeams, but here my solder gets hit again! (2/10 of the bullets aimed at this soldier in just the picture below) (25% chance?) These are almost 80 accuracy soldiers, with range bonus, one was crouched. The first shot was a laser machine-gun, other 2 were Gauss. I have seen too many of these for it to be 'chance' (poor aliens): 6-7 soldiers in the burst fire, and I can't see any tiles hit anywhere or soldiers The actual laser beams we saw in testing, happen still:

I tried 100.0 setting, it stopped any tests vs aliens. That was the main reason I didn't use it. I can try 100.0 vs soldiers, but I think my next batch will be on 0.0 based on recent findings. I already ruled this out. The 'Missed Shot Hit % is rolled using the actual Shot Hit %', which I believe is a custom function over riding the math? So then I tried killing the front soldier and missing that way, which led to the discovery of 'Accurate Misses'. The current line of thought is that, all my 'Testing' is on the map edge for room (with almost no colliders behind the targets, even in the 51 tile shots), and this is the actual cause. You have to miss and hit a non-target behind the target. This is all aligning very reasonable with my experiences so far. There are some times where I was sure I was going to lose several soldiers to burst fire, but they all missed. This, ironically, would make the shield very strong, as aliens will artificially miss it unless it is off by itself. I was also wondering if the Burst Fire Overkill is related to 'delta t' / tickrate. I have a fairly powerful computer running at high refresh rates usually. I might try forcing 60fps or something later as well.

I only really replied here because I had the exact same thought process in my run, but with 3x generators. I was expecting more power, and ran short, so I built the Workshop upgrade instead of a 4th generator since it was much quicker. I am not sure if it was expectations from other games, or a quirk of English regional languages, but I also assumed I would get (40 + 20 + 20) and not (40 + 20/2 + 20/2), but it was just me on cruise control again, even though I build my Labs and Workshops factoring it correctly.

Knowing all this, one might think "just smoke yourself and test that": Any addition of smoke, makes the shot unable to be taken (0% vs 0.001% issue). If we cover our soldiers as best we can with overlapping smokes, and then try to take the shot, we still can't (we know this from many posts back): We can shoot in-front of the Alien, but out of the smoke, and force hits from a 0.001%. Also the friendlies are safe-ish. Red cover dies, shots turn into accurate misses, and hit the alien, which was not the target.

You can, you probably did not have enough time units.

Switch back to Laser Weapons (Rifle/MachineGun/Shotgun) but keep the Gauss Sniper, and try again. If I am correct, it will be much easier. I believe this is related to a bug.

I took these shots about 40-100+ times i think so far? (400-1000 rolls) At 100% it wont let me take the shot due to the 'AntiHack/0% system', so I use 99. I will try 99.99. I cannot tell between deaths and non-death misses. I cannot get a non-burst shot that low other than pistols, and I only have laser pistols, which stop at 32.5%. So I have to use Smoke (which works 'properly') or Range, which exacerbates deviation unless I grind out a 100 accuracy soldier. (I have a 91 acc soldier I think? so 91+8 is almost there, but all these guys have armor on). So to test means I have to run another mission and all my saves are right before a story mission I have not played yet. Attempting to make this 'situation', is where I realized with the help of the 6.24.0 change, that its related to "Soldier" vs "Rock": [ALL SHOTS are taken AT Soldiers my mouse just moved because of key-binds] The result is the people with rocks and trees behind them die: So lets try to confirm that: Now I already KNOW that a 30-50% shot will prioritize cover, so I know the front guy will die, and the target will live (5x Accurate Miss, 1x Miss that hits, 1x Deviated Miss Hit Sniper) I think 1 hit the trees, Its hard to tell sometimes, and one hit the rock at least: If we remove the cover, the shield will get incinerated: 186 Damage, 2 Misses. Inconclusive. vs We can test this many ways: Did they die and it chose a new target? Did they deviate and die? Why is death related to proximity to rock?: Conclusive proof of "Cover Bias": Every single thing other than what we aim at is pulverized. Another undeniable example of Accurate Misses Another example of cover bias (its a bad one, but the train lived several shots): Deviation and Number of Proximal 'Rocks/Walls/Etc' seem to play a factor: Save: All of this post was on this save:user_5cent-26.json Smoke I forgot to get pictures of the smoke interaction, but this is it:

29+56+45+12+49+15+57 = 263 damage on 1 soldier, through armor. Yes. I was having issues with getting screenshots because of the keybindings, so I rebound it. This is the sniper, he hits 100% pretty much everywhere in the green and orange ranges. So any misses in those 100% zones, will be Accurate Misses due to cover (or smoke, covered in other posts). I forgot I turned the setting back to 25%, so this will be 'real numbers'. At (99% StoppingChance) the shot would hit the cloaked sniper easier than shooting him directly (Ctrl Click accuracy hack). Ergo anytime shooting someone is harder than 25%, shooting behind them is a hack (I think we both know this?). The hack is possible because of the 2 rolls, and the bugs. We have 100% hit on the tile. I took the first shot at 77%, and it hit the cloaked sniper in front, but I failed to get a screenshot of it because of the key-rebinding, lol. The zip file should have some examples vs aliens (i could get those pictures easier).

Its more of a discrepancy between the wording semantics and the math. The 'Adjacency Bonus' is the "+" symbol, so each bonus is shared between 2 generators (and all other buildings work like this). A 2x2 square gets 4 bonuses, not 8. The Generator in the Corner with 2 bonuses = 130 (80 + 50/2 + 50/2) The others = 105 (80 + 50/2)

I just wanted to 2nd this. I have seen this behavior before, just not in my current run. Seems to just be missing a state update or something when falling into smoke or falling with existing stun damage.

I was miss-understanding the damage randomness. I thought it would only choose those values, not that it could chose a floating point result between 0.5 and 2.0. I will keep an eye out for any shots that might show otherwise, but in the meantime it should be safe to assume it is working correct (I literally don't even pay attention to how much damage my shots do in normal play, I only remembered there was damage randomization when I rolled over 120 on a sniper). Essentially, it has to or my evolved play-style would be different. I am having difficulties generating a test case that proves this conclusively and not just via statistics, but it seemed like the Deviation chance in 6.24.0 had made this even harder. Then it occurred to me that I have been using soldiers as stand-ins for 'cover/obstacles' every time I was trying to test this (other than smoke). I will have to setup a new test situation to confirm this, but in the meantime these are examples of 'Accurate Misses' with the sniper rifle taking 100% shots against a target behind 99% 'cover'. The shot cannot hit the cloaked soldier as a "miss" because we have pushed misses off the 'combat table' with 100% hit. The 'miss roll' never even gets rolled here because we are using an Accurate Shot, but the game will still show the Miss over Soldiers/Aliens/Civilians (not cover/objects). This also shows that accurate misses don't deviate, they will hit the 'red cover', and then the next red cover as that gets destroyed or becomes invalid (suppression ducking, etc). Misses Below is the type of shot that I was expecting to be able to generate with ease using 99% StoppingChance to show missed shots hitting, but would run into several issues (Target would die before taking 8+ hits, and deviation seemed to start working more reliably as the 6.24.0 fix). This picture should have been easy to make at 99% and it should be a 1 in 500+ chance at 0% StoppingChance, but it showed up on my first 0% shot (~44% Displayed Accuracy): While at the same time, these picture have 99% StoppingChance, and things I expected to die did not die again: The shield in-front dying is expected (accurate hits), and the sniper in the back getting hit is expected (accurate misses), but the shots missing the soldier next to the cloaked sniper should be impossible (unless I lost track of my variable settings, but I don't think I did). My current suspicion is that 'inanimate objects' behind the target are artificially increasing hit chance, regardless of 'StoppingComponent' value, I believe this shot was so accurate because the rock (and a tree) was behind the target: And that 'live targets' behind the target might be reducing hit chance: These results would explain my 'High Accuracy Burst Reaction Fire' that I have been seeing from my Machine Gun soldiers. Post 6.24.0 shows the 'force the miss to hit something else, even if we spin' system IS present like you said, I had just not noticed before due to deviation and high accuracy soldier. Further Testing Required Still I am unable to currently determine why sometimes it seems like I can do massive overkill to a target, but sometimes they die on the 3rd shot (when 4-6 should hit) and the others miss (I expected 8-10 hits here, but did the target die early?): This is also causing troubles with testing, as I am not sure if the target is just dying or if they are being missed normally. I might have to try to generate a save with a high-hp alien to test, but that will take some considerable time to arrange. Cover taking Priority The issue with cover becoming relatively more powerful at Accuracies under 100%, and less powerful over 100% still remains, and will be a large factor in balancing the Shotgun/MG/Rifle against the Sniper/Pistol. The 'simple' fix is to give the Shotgun/MG accuracies over 100%, take away their range bonus, and lower MG range slightly (their range will stop it from being an issue). Adding a -2 or even -5 range bonus on the sniper would probably make it feel more correct (its cover penetrating ability would scale with range, where it does not currently). The pistol and rifles can essentially choose to shoot fairly (100% shot) or shoot at cover (snap shots/burst). This is why lasers are busted. The sniper can choose to shoot fairly (100% shot) or invalidate most cover (160% shot ignores 37.5% cover), making the choice a non-choice unless the target is in the open and you moved to get the shot. The rifles being able to choose between, very low single shot, very low burst shot, 100%+ shot, make them the best weapons, but mostly Laser/Fusion. The Cleaner weapon was a specialty 'close range sniper' rifle, and the alien weapons function similar. Their different range bonus actually helped this 'role distinction' by narrowing the accuracy band where it was powerful, but not making it bad elsewhere. It did not vaporize cover like the overpowered Laser Rifle (one of the 4 best weapons in the game pre-fusion) This Cover issue is a separate issue to the 'missed shots' issue, and is solely due to the math used. Save Files Apologies, my save files are getting a bit messy and I have reloaded some saves like 70+ times. I made the thread before I had ironed out all the 'smoking guns' because I had hoped to get some insight from others' experiences as well. Sniper Shot Tests (also has the guy with 25% bugged StoppingChance): user_shottest-19.jsonuser_shotmiss-24.json Machine Gun Tests user_deviationtest-22.json Smoke Sniper Test shows Forced Miss Spin and Missed Shots outside of smoke not modified by smoke to hit (only the 'miss target' being inside smoke gives protection from misses) user_shotsmokemiss-23.json

Not a problem, take your time. FYI in 6.24.0 it is still the same, but the deviation change helps. Getting close still unfairly prioritizes cover under 100% 'Modified Combined Accuracy', and under 50% it Misses start having a large factor (but the deviation will help this): Main consequence is that weapons with modifiers under 100% are better at range than up close, since they will hit cover up close and miss more at range. Play with this, should be accurate to 6.24.0: https://www.desmos.com/calculator/nlvvpyjcr9

The game will not let me take these shots, they come up every map. So I cannot prove them, because I am literally stopped from proving it. I can only take 0% guaranteed hits against friendly, or with grenades.

Its not working fine, you are saying this should have -40% damage, and I have shown that it applies to all hits. The Laser does 23/34.5/46/92 damage, with 8 armor destruction. We are shooting vs 36 armor. All shots did 0 damage. The first target took 16 armor damage, meaning 2x8, meaning no Falloff, and No Damage Randomization. 46-36=10dmg, not 0dmg. So HP Damage and Armor Damage use different falloff. I suspected this, but only confirmed it with those screenshots. The Solder at the back who took 10 armor damage, had 40% Reduction. If each shot applies armor damage before the next one, it makes this even more bugged. (Picture of it being -40): I am not blaming the dev team, at all. I have no idea where you got that idea from. I am trying to explain that a bug with the "StoppingChanceComponent" variable and its derivatives is influencing the entire game: ALL shots roll to hit TWICE, not once. Missed Shots have 25% Accuracy Missed Shots roll cover sequentially, Accurate Shots roll cover once. Accurate shots that punch through cover, will be specially modified to hit the 'target' with the 'hit%' and not the 'miss%'. [Attempt to counter the bug/hack] Accurate shots that kill the target will not deviate as missed shots. Missed shots deviate via 'Combined Accuracy' in terms of Tiles (not fully confirmed). Smoke is the only thing that modifies 'Missed Shot Hit %' (other than modifying it directly via crouch, etc) Due to all of this, the Shot Indicator is hilariously wrong most of the time, and weapon balance is all over the place. This setting is one of the single biggest balance variables in the game right now. It is the sole reason why most weapons need range bonus. It is the only thing that causes soldier death for me. This is a 0% shot that is actually a 0% shot, but I cannot take the shot to 'prove it' because of the 'system': You can make it into a non-0 shot by getting closer, but this 'abuses the bug' to a limited degree (this is a bad shot because I did not have enough TU for a 10x): If missed shots have 0% hit, this shot can expect 1 hit if we shot a 10 round burst at it: If 'missed shots hit 25% of the time', this is actually 3.3 hits: Now its not quite 3.3, there is some smoke and other factors we can get to. So lets look at a 7% shot into 80% smoke: How many shots will hit (I have 99% hit on missed shots on for this shot)? If smoke worked like the other systems, it would hit 9/10 shots: But smoke is the only thing that works right: So when does a missed shot get better than a normal shot? Unless the shot indicator says 20% or higher, missing is better before cover. So with cover, it gets messy. An Accurate Shot, that Misses due to Cover, will hit that Cover. This is why I only throw 100% grenades. Grenades effectively ONLY miss due to cover (not really, but in practice this is the case, we will get to it). These Shots have been programmed to not use the 25% miss-hit chance, thus they will usually hit cover behind the target. Missed Shots are not subjected to this arbitration, but have to roll through 'each cover', not just the 'highest cover'. This is why Heavy Weapons other than the Laser Machine gun are bad. Cover absorbs their shots like mad, the laser does this too, but it destroyed the offending cover, in both Accurate Misses and Misses. You probably think I am crazy talking about 'accurate misses', but they exist. These are accurate misses killing the cover in-front of and behind an enemy in smoke, but missing the enemy every time. Every target here has the same 25% hit chance: We can duck the front soldiers to make them non longer valid targets for "Accurate Misses" and only valid targets for "Misses", thus they will live and the wall will get shot instead: Now this all gets really messy really quickly. This value determines so much, and it is not even behaving consistently enough for me to test it properly. This is a picture, where my setting is at 50%, but this solider has 25%, he did not move or duck or anything. I even ended the turn and it stayed the same: So ignoring why this value is so inconsistent even in controlled tests, lets look at grenades. This will always land on the square, which is expected: This one is 50% behind the solider, 50% on the solider. I think many people would assume this grenade can land somewhere other than the 2 red squares, they would be wrong. This grenade will only land in the red square, or the orange area: Basically, there are no 'missed' grenades. They will always be "accurate misses", and thus will ALWAYS hit the red cover (which is not always red). "missed" grenades will miss due to deviation, but do not deviate far enough to cause FF issues. So lets talk about Deviation. It seems to be directly related to 'Combined Accuracy'. This solider has 76 accuracy, making my estimate ~2 tiles deviation: If you increase accuracy, you lower 'Missed Shots', and thus you lower your ability to suppress since 'Accurate Misses dont deviate': If you get closer, you increase accuracy due to 'Range Bonus', again lowering missed shots, and lowering suppression: Since we are using 'Accurate Shots', they will prioritize any cover instead of the target: This is why this image exists: (I just miss-understood the root cause) This is why the gauss machine guns feel off (not bad), it shoots AT COVER, not at enemies the closer you get. Taking all this into account, this shot has a very high chance of killing your own soldier (if the wall is breakable, I don't even really fire burst inside UFOs). If you shoot enough burst to break all the cover, in this case 5 tiles of cover, but we only hit each one 40% of the time, so 10 bullets might do it. Please take the time to read this thread in its entirety once more. I have explained everything from multiple angles.

Even damage falloff is 'bugged' and is based on target and not location, because of attempts to 'counter' the 'hack': (I miss-understood the damage randomness being float vs int). These guys getting suppressed, and ducking works like a laser destroying cover, and the result = dead target. The ducked guys are 20% instead of 99% now: Lets check armor damage falloff: Solider inside Max Range = 2 Hits = 16 Broken = ok Soldier outside Max Range = 2 hits = 10 Broken = ok So already, Armor Damage Falloff and Damage Falloff work differently, and we also established that Terrain Damage also works differently. Crouching and Un-crouching has a measured, proven, tested, verifiable effect on missed Grenades, and Missed Shots. (Regardless of my setting) The game will let me take 0% shots that will kill my soldiers, accidentally or not, but it will not let me do the same against enemies, because you think its a hack and not a bug I specifically avoided mentioning Ctrl Click, because I knew this is exactly where this would lead once I encountered the 'You cant take 0% shots system'. This is a system I don't even use, but I know how it works, and thus I never grenade myself.

Missed shots ignore accuracy. Weather that is a bug, a feature, a hack, an exploit. It is present. It effects grenades, burst weapons, weapon balance, weapon feel, AI behavior, and so much more (soldier survival, meta decisions, etc). You guys have programed at least 2 Systems, and done at least 2 patches, specifically to stop this issue from being 'used' one by good players, and one by bad players. I don't have a 100 Accuracy Soldier yet to test with, this 86 accuracy one can get close at 51 tiles. Its hard to even get the camera to follow the shot: At 48 tiles, same thing: At Exactly -160 (where the 100 accuracy sniper would go from 1% to 0%): The miss on the right was actually a 0 damage hit the game thought was a miss. Now, there was still a deviated shot at this range, so my next assumption is (86 + 8 -3)*1.60 -x= 0, which is 40.56 tiles: There is some deviation here still. So closer. 0 Damage hit again, going through 3x 99% cover. Still some deviation though: Not much though: This is madness: Why does crouching and uncrouching change missed shot accuracy?

Setting at 99 Setting at 0 auto_groundcombat_turn_18_start-739.json The Sniper, The Assault, and The Rifle have the highest Reactions if you want recreate this.

This setting IS the bug. I could play this game at any value, that is what I am trying to say.

Why only throw 100% nades now:

99 -> Red Guy dies 0 -> Red guy dies 5 -> Red guy dies Oh my god, its even worse than I thought. Crouching and Uncrouching does whatever this is. ^

I have ONLY done that for the purposes of getting these pictures. I have not used it ever in my current run, I drain all my TU's so reaction fire doesn't do it. I limit lasers, so they don't do it. The AI still runs. Sadly the pre-mission save I have is pre-map gen, so you get a different map than I did.

What will this alien do?user_thebug-18.json Hes trying to run away because it knows, any non-burst, non grenade will get it killed. It still did not attack, and tried to run again, and died to reactions: If you remove reaction fire, it will burst fire to kill someone, then find cover. It gets reaction punched: Twice: Trying to run and hide from my better accuracy, better armor, superior numbers. We got Rocky III over here, eye of the tiger stuff: It stopped here and took bleed damage: