[SOURCEMOD] Brutal-OXCE 7.12.1

[Juku121]

Why not 200 enemies? Or 1000? This is nuts, dude. How do I suppose to make accuracy mechanics which works similar in terms of fun both for original game with 15 missions per month / 15-30 enemies vs this nonsense?)

I think there are missions with 200, too, at least on SH.

The whole point of my remarks so far has been to say that you do it by not introducing severe rebalancing. 'Realistic accuracy', on the face of it, sounds like an attempt to remove the 'accuracy is a value and not a probability' issue from the game. But your rework goes far, far deeper. 

Better in many ways is still the goal tho.

As long as your changes completely invalidate anything even resembling vanilla tactics in any moderately cluttered map, promote aimed shots (and grenades and, I suppose, melee, due to running being a thing for quite a while now) over snap/auto to the tune of two orders of magnitude, well, that's going to be a tough sell.

Nothing confusing for players here. You get highly covered enemy?

The voxel/sprite issue (and others, like asymmetric vision due to only some unit voxels counting for visibility) means it's hard to predict cover by just looking at the map. And Xilmi's AI gets a tremendeous leg up here, since it can test all firing positions, something the player cannot do without insane savescumming.

In my experience, even one new variable which seems to be clear and have predictable effects - turns out as unpredictable...

Yeah, that is very much true and a solid argument in itself. But since the whole model is going to behave very different anyway, giving some people (not all!) the ability to tweak your model to achieve results more in line with their mods, or the player base in general, sounds like a better approach than 'my way or the highway'. Well, if you want people to actually play with your option turned on and not look at it, exclaim "5% chance to hit the alien in the bushes 10 tiles away?! What even is this...?" and then turn it off again. JA2's nCtH, for whatever faults I may be finding with it, at least gave people a laundry list of options to tweak to their liking.



Now, at the end of the day, it's all about what you find fun to do yourself. So don't take me too seriously. But I would like an option that makes the accuracy value an actual percentage and nothing more. Or at least nothing radically different, accuracy-wise. Whether there are others who would appreciate that, or whether that's something you'd like to do, well...

[Xilmi]

The bushes are also troublesome for AI. They block tile-vision but are leaky in terms voxel-vision. So you often can see units but not the tile they are standing on. The AI relies on tile-visibility to detect cover. So the bushes often are identified as valid cover because of that.

I personally suggested that no tile-vision should also mean no unit vision. But that caused all sorts of other issues. So to this day there is no real solution. It just means the AI will perform worse on jungle-maps.

[jnarical]

[mention]Juku121 [/mention] don’t get me wrong, I’m not fond of current state of the game, so… “my way or highway” - we aren’t there yet)) And I take into account other people and their opinions what’s fun and what’s not

[Juku121]

Sure, one can find plenty wrong with the current state of the game. Of course, all the mods mean that there is no longer a singular state, anyway, but there's some commonality.

I don't think I've seen anyone except you express an opinion like "hit chances are two orders of magnitude too high", though.


Of course, I don't have much of a leg to stand on regarding all this, since I also frequently change things to my own liking and have even done a limited amount of shoving it down other people's throats (though not here on this forum or under this moniker). So do as you wish, and ultimately it'll come down to whether people like what you do enough to adopt it, or not.

[zee_ra]

The only issue accuracy-wise is the fact that a powerful weapon cannot shoot through weaker obstacles (like e.g. bushes and wooden walls).  I think that a good sniper should be able to shoot through the bushes just fine, especially with sniper-spotter setup.  What do you think about this?  Would implementing this break too many model assumptions behind the logical setup of tactical battlescape?

[jnarical]

I don't think I've seen anyone except you express an opinion like "hit chances are two orders of magnitude too high", though.

That’s not my opinion. I wonder where did I say something like that…

[jnarical]

The only issue accuracy-wise is the fact that a powerful weapon cannot shoot through weaker obstacles (like e.g. bushes and wooden walls).  I think that a good sniper should be able to shoot through the bushes just fine, especially with sniper-spotter setup.  What do you think about this?  Would implementing this break too many model assumptions behind the logical setup of tactical battlescape?

Oh no what have you done… now I want this)

upd: chopping 50m long firebreaks with heavy plasma…

[Abyss]

That’s what I was talking about. Why not 200 enemies? Or 1000? This is nuts, dude. How do I suppose to make accuracy mechanics which works similar in terms of fun both for original game with 15 missions per month / 15-30 enemies vs this nonsense?)

This is part of SH difficulty. Lower difficulty means 50-40-30-20 enemies by step.
With vanilla OXCE it was the only challenge option. Same to ironman, because otherwise you can beat the game with no casualties at all. 
With some advanced AI most challenge comes in balance:
- you have to сorrelate losses with the training speed of new units. The later is very low speed process, and yet forces you to take fresh low-stats units to missions, because on-field training goes much faster.
- you have to train 4x units more.
- you have to skip missions which mean significant loses.

IMO, the perfect game design is made so player
-  allocates resources in tight connection with game progress, which, in it's turn, takes them from him for challenge & drama. First is mod design, last is AI.
- with random and gambling, so joy from better result in same action makes player engaged.
It is random missions spawn, with better or worse loot/research topics vs effort. Lootboxes - there are a lot of them. Weapon boxes, ammo boxes, alien stuff packages. Even gambling itself - in X-Piratez you can build casino and spend money in it - and the prizes are everything from weapons and bananas to slaves and stuff.
But also it is gambling with gunfire: whether you hit or miss, whether enemy hits or misses.
And, no less important is gambling with reaction shots - whether this enemy does something or not. In some cases, when enemy has no spare TU's, it's much easier to beat.
- gives player a reward worth all spent resources at certain points
- punishes player for all bad decisions, actually teaching player how to play: everywhere player could have made a better decision - he understands it.
- overall, has good story plot and dynamic
     
RA affects gambling factor and dynamic. Here's another video with RA enabled:

There you can see that player will almost never try a 5% shot. Gambling is affected in negative way. Potential joy is wasted.

[Abyss]

Oh no what have you done… now I want this)

Not only sniper shots, but also heavy machine gun, gauss, plasma, etc.
Bullet pre-roll damage should make these bushes less bulletproof, yet should be in balance with heavy covers, as result affects all covers. You won't code so only bushes are affected, right? Or you will, but this will take too much effort.

This is part of suggestion to make powerful bullets more accurate.

bullet pre-roll damage 20% (0 = 0%, 100+ = 20%)

Even though 20% seems too high: pistol with 20 dmg will have +4%, HMG with 50 dmg will have +10%, sniper rifle with 75 dmg will have + 15%. Plus overshoot. Everything else - to the balance of mods, already tuning hitchances.

Or even can make 20 = +0%, 100 = +15%

Modificators apply for shooting into covered target, so only to compensate missing probabilities, not surplus the shots on open terrain.

a bunch of variables with totally arbitrary values and hope for the best

This is how I see to make it balanced in good and joyful way.
We shouldn't exclude some types of attacks and weapons from the equation, making others supreme.

And, for god's sake, please don't forget that hit/miss mechanics are already emulated by 0-200% dmg roll. 

[Xilmi]

Not only sniper shots, but also heavy machine gun, gauss, plasma, etc.
Bullet pre-roll damage should make these bushes less bulletproof, yet should be in balance with heavy covers, as result affects all covers. You won't code so only bushes are affected, right? Or you will, but this will take too much effort.

Each object has an armor-value. Currently this only determines whether the object will survive the hit of a bullet or not based on the bullet's power.

Something that would be quite feasible is if instead of just destroying the object when it's armor is below the damage-roll but the bullet disappearing, the bullet keeps flying but it's power is reduced by the armor of the object it just destroyed.

This would imho be the most logical and consistent approach.

Your plasma-bolt might pierce through 5 plants easily as their armor is like 12. But it can only penetrate one stone-wall with an armor-value of 50.

And also: The destroyed cover would still have helped you in the sense that it lowers the damage of the shot.

[Abyss]

the bullet keeps flying but it's power is reduced by the armor of the object it just destroyed.

This approach requires revision of all objects below 60 armor.
And also, there's fork:
- whether object will be destroyed completely
- whether object stays intact, but lets bullet through it
- whether it will demote bullet power overall
- whether it will have HP or not
- etc. etc.

This seems like whole another mod.

Then, also How BAI will deal with all that?

For previous suggestions, everything was in regard to current gameplay.

To Joynarical:
You have two enemies in 10 tiles from you. One is totally open and is within 50% hitchance. Other is covered in bushes and has 2% hitchance.
If you shot into second target, will 48% of shots statistically hit the bushes?
It's not obvious yet. Something makes me not believe that. Seems like most shots go into the air/not line of the hit.

[Xilmi]

IMO, the perfect game design is made so player
-  allocates resources in tight connection with game progress, which, in it's turn, takes them from him for challenge & drama. First is mod design, last is AI.

One of the biggest issues I have with the game-design is that there is a contradiction between individual missions being exciting and the overall game-progress.

A mission that I barely win is exciting. But the campaign isn't balanced around barely winning and sometimes losing missions. You'd fall behind in score and soldier-development.

Vanilla actually still kinda works like that. There is some leeway to lose missions and still go on.

The issue is much exacerbated in mods where there is a big discrepancy between a fresh recruit and someone who is well-trained. The missions must not be exciting as losing well-trained-people is too much of a set-back.

Basically: The Geoscape-difficulty has to be low so the difficulty for individual missions can be higher or vice versa.

[Xilmi]

This approach requires revision of all objects below 60 armor.
And also, there's fork:
- whether object will be destroyed completely
- whether object stays intact, but lets bullet through it
- whether it will demote bullet power overall
- whether it will have HP or not
- etc. etc.

This seems like whole another mod.

Not sure why this would require a revision of all sorts of objects.
Staying intact while letting a bullet through currently isn't possible either. This could be an independent second step but not a necessity for first implementing this.
Same goes for demoting bullet-power. Always destroying the object and always demoting power would be the most consistent approach I can think of.
Objects currently generally don't have HP. They just have armor. You can't damage an object over several attacks and have it eventually break. Each attack has it's individual chance based on whether it rolls more or less damage than the object's armor. Making a change like that would have a big impact on the engine and is once again not necessary for implementing it the way I suggested.

Then, also How BAI will deal with all that?

It would have to be adapted. It needs some sort of new canTargetUnit-code that takes armor-piercing into account. If not it would act as if the feature didn't exist and only benefit from piercing projecties by chance.

You have two enemies in 10 tiles from you. One is totally open and is within 50% hitchance. Other is covered in bushes and has 2% hitchance.
If you shot into second target, will 48% of shots statistically hit the bushes?
It's not obvious yet. Something makes me not believe that. Seems like most shots go into the air/not line of the hit.

I'd also kinda like to know. I haven't tested this in-depth but it would be a big difference if the 48% discrepancy will likely hit and potentially destroy the cover or just miss in some unrelated way.

If I imagine aiming at the center of an enemy and having some sort of hit-cone around the center, then a lot of the off-center-shots should still hit the enemy.

But if I aim at his finger because the finger is the only thing exposed, the cone around of where I aim for would be much worse, especially if the cover is easily breakable.

So I guess the question is: Does lower exposure change where the soldiers aim at? Do they still aim at the center of the enemy or do they aim for only the exposed voxels?

[Abyss]

Xilmi,
I start to record the 7.11.1 (which loads as 7.11.0 in cmd console, though)
Now I have even more questions how things would interfere:
- statistical randomization, when 40% units are VAI
- weighted randomization, for other 60% BAI units, which still supposed to do some side stuff, randomly. Will it be VAI stuff? Or strange BAI stuff? How many percent of them will do so?
- dynamic aggressiveness will be starting from "don't bother take cover" or "careful advancement"?
- just got shot from other side of the map, while no sight contact is made. Regarding this: is vanilla sniper-spotter mechanism working for these 40% VAI units?
- and if so, are only VAI units eligible to see and shot into "is spotted" units?

If last two questions are "yes", then this is amazingly joyful and cool. 

[Abyss]

To Joynarical:

"Why not 200 enemies? Or 1000? This is nuts, dude. How do I suppose to make accuracy mechanics which works similar in terms of fun both for original game with 15 missions per month / 15-30 enemies vs this nonsense?)"

Fair balance, other is up to modders. Right now, with EA it feels natural and balanced.
Even with RA it is already sort of fair right now, except too low hitchances overall in several cases.