X-COM OpenXcom Thread

[Average Manatee]

I don't think superhuman alters the number of missions that spawn, only adds some more guys. So you still need to do the same amount of missions, they are just slightly quicker. You still need to explore the whole area and that's what takes the most time. It's not too much a problem of missions being difficult as much as it is them being too numerous. Doing like 10 per month while suffering with 5 scientists who never get things done is torture.

I guess the "land and leave" thing works but it seems so silly.

 

[lightbane]

I don't think superhuman alters the number of missions that spawn, only adds some more guys. So you still need to do the same amount of missions, they are just slightly quicker.

IIRC it does alter the mission spawn to a degree. It makes certain enemies that should show up late appear earlier than normal. Ie: Once the invasion begins, you get Ethereal encounters instead of Sectoids. Adding more enemies does not make missions quicker IMO, as the bad guys hit harder and are more numerous, which may be okay for a cultist mission if you know what you're doing (since the enemy will surrender at some point), but God save you if you're doing an animal/zombie mission.

You still need to explore the whole area and that's what takes the most time.

Theoretically there's the Bughunt Mode that should trigger once one or two enemies remain, which supposedly reveals their location in the minimap. IMO it's just faster to use Debug mode.

It's not too much a problem of missions being difficult as much as it is them being too numerous. Doing like 10 per month while suffering with 5 scientists who never get things done is torture.

The start is slow indeed, but nothing to be done unless you cheat to get more money and build more bases, so that you can have more projects researched at once, or edit your saves to instantly research stuff. You also can do one mission or two of each type and forget the rest. As long as you keep researching and doing the terror missions, you should have enough points to win the month.

I guess the "land and leave" thing works but it seems so silly.

It was a standard tactic in OG XCOM when encountering say Chrisalids in a Terror mission.

 

[eXalted]

Update:
XPiratez v.L4 15-Sep-2020 Mad Melee Massacre

- OXCE Upgrade to 6.6.1+ 12-Sep-2020
- Fixes to psiSkill overflow (again)
- Fixes to enemy morale regen (again)
- Fixes to Mutant Blessings (again)
- Fixed EMP weapons (erroneously ignoring Shock Immunity)
- Fixed Chort ranks - Fixed X-Vampiress recovery
- Fixes to Recon Expeditions
- New Missions: Haunted Forest, Isolated Valley
- New Facility: Bee Hive - New Armor: Ghostbuster (PEA)
- Experimental: extended melee reactions enabled
- Experimental: off-centre shooting by Karadoc (enabled by default)
- Basescape Gfx update (WIP)
- Dogs can bark now which reduces enemy TU, Morale
- Battle Tank can now be used as craft in missions (battlemap, turret unit & weapon added)
- Battle Tank manufacturing time x2, M1A1 version removed
- Basic Mecha Armor weapon changed to Aircraft HMG (fixes no-rotation and no-reload issue)
- Zombie damage type changed to Daze
- Chryssalids and other zombifying units now need to actually cause damage to zombify
- More Pedia stuff
- Tweaks, Improovements, Adjustments, Map Fixes, Other minor fixes

No more "why I can see the enemy but can't shoot him?".

 

[agris]

Does it have a de-weeb mode yet?

 

[Horvatii]

- Experimental: off-centre shooting by Karadoc (enabled by default)

DISGUSTING CAUSALIZATION! STREAMFAGGIN!
if you dont have a window open with fpsview preivew, which autorefreshes, YOU R A CASUAL!

 

[geno]

Can someone explain what is that off-centre shooting stuff?

 

[Average Manatee]

Guessing its removing the issue where you could see an enemy but not shoot them because your sight came from a different section of the model than your weapon's firing point.

 

[Horvatii]

Yeah, what Manatee said.
Stand 1 tile to a side of a door and look at it at 45 degrees and you wont be able to shoot the purple alien.

 

[geno]

Should that fix the issue with trying to shoot throw a window but the game says you haven't a line of fire?

 

[Horvatii]

Yeah, depending on how off-center they go.

There is an interesting problem with this new behaviour though, which existed before but now is gonna get a bit worse.
The accuracy displayed ingame is actually an estimate of the simulated spread of a shot that depends on distance, hieght level and funnily on direction of shot, eg left of right. Its not true accuracy.
And the estimates fits best (if I remember my math correctly) for center mass of an enemy the size of 8x16 voxels at ~10 tiles distance.
While in fact, you can shoot at something if you see a single voxel at max view range.
Cases like that but for long ass range are now more often shootable and will give more misses and make newbies rage.

So say you are shooting with 100% acc onto a single visible voxel at 25 tiles distance.
The max deviation on that shot is 20 voxels horizontally. So your chance to hit a fully exposed alien is 8/20.
You chance to hit a single voxel is 1/20. But the game displays 100% acc.

Maximum effective accuracy btw is 110 or 111%. At that level your max deviation is 1 voxel. You are guaranteed to hit the target voxel or the one next to it, coin flip. Independent of distance.

 

[Average Manatee]

The max deviation on that shot is 20 voxels horizontally. So your chance to hit a fully exposed alien is 8/20.

That doesn't sound right. I'm pretty sure there's some kind of probabilistic factor with the center targetted voxel having a massively higher chance than the edges. Some kind of normal distribution. I can definitely say that if you shoot at something with a 100% chance to hit you have something like a 95% hit rate. It's incredibly rare not to bullseye enemies across the map.

 

[Horvatii]

I dont think its normal. RNG::generate(min,max) should be uniform.
https://github.com/OpenXcom/OpenXcom/blob/master/src/Battlescape/Projectile.cpp#L392

 

[Horvatii]

now the note on zshift for a full y dir shot: the zshift would be 400, so the final devi would be 20 and go from -10 to 10 around the target voxel.
So I gave the worst case example
-edit-
target.x does not matter when shooting in x direction we are shooting but target.y is the same in calc so whatevs, goes from -5 to 5 aswell

The hilarious thing about this is that its actual gameplay relevant knowledge.
Its a difference between 8/20 or 8/10.
Imagine sniping across the base or you are sneaking up to enemies in a corner of a map.
The direction of approach matters.
Kinda.
You hit 110 acc too fast too easy.

 

[Average Manatee]

I think I see your problem

deviation = std:max(1, zShift * deviation / 200)
which with zShift = 200 reduces to
deviation = std:max(1, deviation)
with the input deviation being -90 to 10.

You can correctly identify that the output deviation is between 1 and 10, but it is not evenly distributed. Instead, 91% of the time it will be a deviation of 1, with 9% of the time being 2-10 deviation. If we go with your assumption of a target that is 8 voxels wide, we'd have a 91% chance to hit it dead center or one off (whatever that means, obviously there's no center voxel in an 8 voxel wide target), an 8% or 9% (depending on rounding) to hit within the target, and a 2% or 1% chance to by 5 off and barely miss.

I really don't get the

if (xDist / 2 <= yDist) //yes, we need to add some x/y non-uniformity
xyShift = xDist / 4 + yDist; //and don't ask why, please. it's The Commandment
else
xyShift = (xDist + yDist) / 2; //that's uniform part of spreading

thing though. Is that replicating some kind of original x-com code error?

In cases where xdist / 2 <= ydist then you'd still have a 90% chance to have a final deviation of 1 (bullseye) with a 10% of 2-20 (leading to a ~95% chance to hit the target overall).

So high-accuracy shots are still pretty damn accurate even in this weird case where 25% of the time the distance is up to doubled. Its the low-accuracy shots (deviation +50) that suddenly really suffer from a doubled final deviation.

 

[Horvatii]

You r right!
Did not expect clamping to cut into the distr. I thought its there just to remove the negatives...

Who knows about The Commandment... I'd guess its something dumb the coder came up with. Otherwise I'd expect a comment like "old xcom behaviour"



Hmm, another numeric example, 10% acc, point blanc 1 tile distance:
MISSES 0.9 of cases:
original devi rolls between 50 and 140
zshift 16
final voxel devi is between 16*50/200==4 and 16*140/200==11.2
(devi of 4,5,6,7 are hits,
devi 8 has a 1/9 chance to miss, eg devi rolls are: -4,-3,-2,-1,0,1,2,3,4, but the alien is only 8 voxels wide, so 1 shot goes past
devi 9 ~ 2/10,
devi 10 ~ 3/11,
devi 11 ~ 4/12)
That means about half of the misses actually hit the 8 voxel alien.
The other devi's hit at about 89% 80% 72% 66% => sum of 3.07, so (4+3.07) / 8 == 7/8 devis hit of all the misses.
That means that 87.5 percent of the misses actually land on the alien, if shooting 10% shots point blanc.

HITS 0.1 of cases:
original devi between 0 and 10
zshift 16
final voxel devi 1 to 1

Final tally is 0.1+0.9*7/8=89% acc

So grab a 10% autofire SMG and run very close!
Try not to activate Xpiratez melee evasion...

 

[Average Manatee]

I'm guessing by "The Commandment" that someone fixed it to work properly, then an order came From On High that the accuracy equation had to 100% match the original and it got changed back. I'm pretty sure the openxcom devs have been quoted as saying that accuracy is a 100% replication of the original.

Not sure about OpenXcomEx though, perhaps they fixed it?

EDIT: Nope, looks like its still there.

 

[Horvatii]

turns out all the commits since the very beginning of dev have been in a git repo
git blame says its vanilla
https://github.com/OpenXcom/OpenXcom/commit/bb43f7a439d903840c8f375debfb1399d5f59282

looks like it had 3 versions over the lifetime, first ufoextender like rangebased, then deviation in radiants derived from experimental data, and now its probably based on vanilla disasm
https://github.com/OpenXcom/OpenXcom/commit/4b6b86d3bd068a582672933573bea5328e0ca1db

 

[CaffeineConquistador]

Yeah, what Manatee said.
Stand 1 tile to a side of a door and look at it at 45 degrees and you wont be able to shoot the purple alien.

whoa, apologies, I haven’t followed this thread in awhile, but is this some new 3D engine for openxcom? Or an example you’re giving for clarity? Cool, either way.

 

[eXalted]

whoa, apologies, I haven’t followed this thread in awhile, but is this some new 3D engine for openxcom? Or an example you’re giving for clarity? Cool, either way.

X-Com engine (or rather its calculations) are actually 3D, the graphical representation is 2D. In OpenXcom, you can make a first-person 3D screenshot during a battle with F11.

 

[CaffeineConquistador]

I’m embarrassed to say I never knew X-com worked that way nor knew openxcom had that feature.

 

[agris]

3D vector math; does anyone know their Mythos lore? Who on the dev team had an engineering background?

Color me not surprised that one of the classics was developed by a dev house staffed with people who have backgrounds outside an insular world of video games.

 

[Endemic]

3D vector math; does anyone know their Mythos lore? Who on the dev team had an engineering background

Sorry to disappoint you, but the programming was done by the Gollop brothers, who were already experienced with video games at that point.

Not exclusive to X-COM either; flight simulators were also tracking projectiles in 3 dimensions, and they were en vogue during the 1990s. The battlescape is (IIRC) 10 stories high, I'm surprised you didn't notice shot deviations can happen in the vertical plane as well.

X-Com engine (or rather its calculations) are actually 3D, the graphical representation is 2D. In OpenXcom, you can make a first-person 3D screenshot during a battle with F11.

Explosions don't have any height unless you activate the option for it in Openxcom, strangely enough.

While I'm here, we ought to be thanking Microprose for there being a geoscape, that wasn't in Gollop's plans originally.

 

[Average Manatee]

Most programmers (as in actual programmers, not shitty "I do web design" idiots) can do math pretty well and usually have math interests. I was doing (simplistic) stuff for a school assignment calculating projectile trajectories in 3D space in the 7th grade using basically the same kind of math. Except AFAIK I didn't fuck it up with something weird like X-Com apparently did. I'd assume that probably ~95% of lead programmers on any game could whip up something similar.

The battlescape is (IIRC) 10 stories high, I'm surprised you didn't notice shot deviations can happen in the vertical plane as well.

Vanilla X-Com had 4 height levels. I think OpenXcom supports up to 12?

 

[Cael]

Most programmers (as in actual programmers, not shitty "I do web design" idiots) can do math pretty well and usually have math interests. I was doing (simplistic) stuff for a school assignment calculating projectile trajectories in 3D space in the 7th grade using basically the same kind of math. Except AFAIK I didn't fuck it up with something weird like X-Com apparently did. I'd assume that probably ~95% of lead programmers on any game could whip up something similar.

The battlescape is (IIRC) 10 stories high, I'm surprised you didn't notice shot deviations can happen in the vertical plane as well.

Vanilla X-Com had 4 height levels. I think OpenXcom supports up to 12?

Apoc is the one with 9

 

[prengle]

3D vector math; does anyone know their Mythos lore? Who on the dev team had an engineering background?

Color me not surprised that one of the classics was developed by a dev house staffed with people who have backgrounds outside an insular world of video games.

i have no idea who exactly added the 3d voxel collision shit, but julian and nick gollop are the only two programmers credited for working on dos ufo. unless some poor bastard designed that entire system and went uncredited, chances are it was one of them. i'll shit on julian for days, but the sheer amount of systems he and nick managed to pack into x-com is still impressive

also out of curiosity, does anyone know if nick gollop's ever done an interview? julian seems to love basking in the limelight whereas nick seemingly hasn't worked on a game since rebelstar on gba