[v22.8] GC - Line of Fire Issues

[kabill]

Was playing around looking at the new line of fire system and there's definitely an issue to do with how the game is calculating LoF in some instances. Here's an example:

The soldier at the bottom of this image has LoF on the soldier at the top:

However, the soldier at the top doesn't have LoF on the soldier at the bottom:

This is in spite of the fact that there is a clear LoF right though the centre of that soldier's tile, as demonstrated if I aim a few tiles behind the target:

I'd thought this might have been due to asymmetrical LoF tracing, but that seems to be wrong. In these images, the LoF is identical for both soldiers:

[legit1337]

Just because one firepath is symetrical, does not mean all of them are.

Also, in your second frame, the soldier is facing the opposite direction. That might have affected the LOF calculation?

Edited May 15, 2014 by legit1337

[kabill]

Just because one firepath is symetrical, does not mean all of them are.

I checked it out in a few instances, but you can only upload five pictures per post. (I may also not have taken pictures after the first example. Definitely tried out a few positions, though, all of which had symmetrical LoF).

You've still got a point though. Two in fact! Firstly, there could be some relative positionings that don't produce symmetrical lines of fire. And secondly, just because shots in the open produce symmetrical lines of fire doesn't mean the game is trying to do that where leaning is involved (it's impossible to know this for certain since if the game thinks it can't avoid a prop, it doesn't show the 'best effort' so it's not possible to tell what trajectories it was actually trying).

Nevertheless, either of these situations would imply a situational bug as the cause rather than the general design of the LoF system (which is what I thought was the case). Also, while it's still not 100% equivalent because it lacks cover, the example I posted has units in the same relative positions as Skitso had in his example posted on the other thread (which suffered from asymmetry).

Also, in your second frame, the soldier is facing the opposite direction. That might have affected the LOF calculation.

Yeah, I spotted that after I took the picture. I was still in the game, though, so I was able to check and it was still the same even when I changed the soldier's facing.

[legit1337]

Yeah, I spotted that after I took the picture. I was still in the game, though, so I was able to check and it was still the same even when I changed the soldier's facing.

Ok, just making sure.

I checked it out in a few instances, but you can only upload five pictures per post. (I may also not have taken pictures after the first example. Definitely tried out a few positions, though, all of which had symmetrical LoF).

You've still got a point though. Two in fact! Firstly, there could be some relative positionings that don't produce symmetrical lines of fire. And secondly, just because shots in the open produce symmetrical lines of fire doesn't mean the game is trying to do that where leaning is involved (it's impossible to know this for certain since if the game thinks it can't avoid a prop, it doesn't show the 'best effort' so it's not possible to tell what trajectories it was actually trying).

Nevertheless, either of these situations would imply a situational bug as the cause rather than the general design of the LoF system (which is what I thought was the case). Also, while it's still not 100% equivalent because it lacks cover, the example I posted has units in the same relative positions as Skitso had in his example posted on the other thread (which suffered from asymmetry).

Perhaps. It still may be a bug with the entire LOF system, if the problem logic is buried so far in the code that you would need to rewrite the entire system to fix it...

[StellarRat]

Perhaps. It still may be a bug with the entire LOF system, if the problem logic is buried so far in the code that you would need to rewrite the entire system to fix it...

I doubt it. We had to work this problem in our game. It was just some math using the X and Y coordinates of the target, intervening objects, and the shooter. I don't remember it being overly complex. It could be tricky with irregularly shaped objects as I think you'd have to break them down into points and analyze them one by one (there might be an easier way though.) It could be a rounding problem though. Also, if you rely on the engine to determine intercepts it's quite possible IT has a problem. We only worked with our own graphics system, so we had complete control over everything. Edited May 15, 2014 by StellarRat

[kabill]

Ok, just making sure.

It's a good idea to work through these things, so it's no trouble.

Perhaps. It still may be a bug with the entire LOF system, if the problem logic is buried so far in the code that you would need to rewrite the entire system to fix it...

May well be. Just wanted to discount it as a systematic error that was happening all the time as a result of how the system was designed ('cause I got it into my head that this was the likely cause).

[Chris]

The save would be handy there tbh, it seems quite hard to generate these issues artificially myself.

[Chris]

Also, having looked at Skitso's example image again (http://i.imgur.com/GlndAZV.jpg) I don't believe that is a bug either, as the two units are not directly aligned. If they were, neither unit would be able to hit the other.

As it is, the Caesan is one tile to left and therefore can shoot while the Xenonaut cannot. Were the Caesan one tile to the right of center instead, the Xenonaut would have the advantage instead (his wall would cover relatively more of his body if they were shooting at one another in that situation).

[Skitso]

@Chris: Umm... I'm not sure if I understand your point correctly, but either units' direction shouldn't matter if the LoF starts from the best corner of the shooters tile.

Also, if you look carefully, xenonauts LoF starts from wrong location: you can see the small green square in the middle of the SE side of the xenonauts tile, so all I could hit is the wall.

To try and make my point more clear, I modified the picture for more clarity and as you can see, LoF's are perfectly symmetrical.

... and that fact doesn't change, no matter how and in which direction you look it from:

Edited May 17, 2014 by Skitso

[Chris]

I'll post a more detailed explanation in another post shortly. Kabill, if you have a save for this example then that would be really handy as this does genuinely look like a bug.

[kabill]

Ok, so I don't have a save for the example above, but here's some others.

(Ugh, no. These won't work as they're from a modded game. You'd have to install my hull breaching mod to get them to work, I think, as there's some soldiers with a custom weapon. I'll upload them anyway, though, just in case there's anything you can do with them.)

Example 1:

The two soldiers here are 6 tiles NE-SW and 8 tiles NW-SE apart (in case it's not clear from the picture). The soldier at the bottom can target the one at the top:

However, the reverse is not true:

Example 2:

This example's actually possibly more useful in terms of demonstrating what's happening. Here, both soldiers can target one another. However, shooting top to bottom avoids the corner of the other side of the hedge:

However, shooting the opposite way clips the corner of the hedge:

The LoF in these pictures is slightly different. Actually, it's mirrored, since the pattern going from shooter to target in each one is the same. That's fine, but because the full pattern doesn't repeat in the last iteration it makes them non-symmetrical. (Not sure I've articulated that very well; hope it's clear from the picture).

As it happens, this is precisely what I thought was happening in the first place: because the shots are traced always from shooter to target, in some instances it will produce asymmetrical results with the effect that shots will sometimes be interdicted one way but not the other. Don't know if that's causing all these issues, but it's definitely have some effect.

EDIT 1: Can't tell for sure, but I think that explains the problem in my first post as well. Reversing the LoF by eye, it looks like it would be slightly asymmetrical and cause the LoF to clip the edge of the rock prop from the soldier in the top right.

Hope that's useful, and sorry about the saves probably not being helpful.

EDIT 2: Kind of related exploit: you need to have a shot on the centre of a target in order to shoot at them, but a LoF which passes over *any* part of a tile can be blocked by it. That means that, in instances where you can't target a unit directly, you can deliberately aim behind them such that the LoF clips the target you wanted to hit. It's not as good as shooting them directly, but it's better than not being able to shoot them at all.

Don't know how you'd cure that without some fairly hefty changes, though, and since it's only something the player will be able to do, it's probably not that important.

Edited May 17, 2014 by kabill

[kabill]

Sorry for double-post, but I've got some more screenshots and a save file which should be compatible with vanilla if that's useful at all? From what I can tell, they point towards the same conclusion as above.

[Chris]

Skitso - it's explained in the other thread. Simply: one of them has their body exposed, the other doesn't.

Kabill - yeah, a save for vanilla would be more useful.

[kabill]

Here it is: https://www.dropbox.com/s/sxqfo45xr210uqp/LoF%20Issue.sav

(It still might not work - it's not from a vanilla game, but there's nothing in it that should cause a problem so far as I know).

Pictures of the situation:

I thought I had an instance of it happening with non-full cover props as well, but I've not been able to reproduce it. I'd not been checking carefully to make sure that units were symmetrical with one another, however, so I think I'd accidentally mistaken a legitimate asymmetrical cover situation for an illigitimate one.

But in the example above, the cover which is protecting the soldiers is definitely symmetrical in terms of its coverage. And again, if you reverse the LoF and apply it manually for the blocked unit, it would be forward by a single sub-tile and therefore would clip the edge of the rock before it can strike the unit.

EDIT: Here's the reverse LoF drawn on in pink. The orange tile is the point where the difference is; because of the slight asymmetry, the shot hits the corner of the rock prop and therefore is blocked:

Edited May 17, 2014 by kabill

[Chris]

That's working as intended, though. If those mesas were actually 2x2 buildings (which would fill an identical footprint), you'd see the top soldier doesn't have his body hidden by the corner but the bottom soldier would.

[Skitso]

Intended or not. It's too unintuitive and messy if a developer needs to explain how it works in the forums. While I get the core principle how the system works and I'll probably get used to it and can live with it, it just feels wrong. I'm just worried that people who hasn't played this game for years and discussed it's mechanics here daily for god knows how long will be confused with it. It's guaranteed, and I feel it's a shame as it's one of the most (if not THE most) important core gameplay mechanic in this game.

I'm not trying to play GH decisions down or anything. I respect Chris as a game designer and I trust his vision. I just feel this amazing game deserves a better LoF system as this feels broken and wrong.

Edited May 17, 2014 by Skitso

[kabill]

That's working as intended, though. If those mesas were actually 2x2 buildings (which would fill an identical footprint), you'd see the top soldier doesn't have his body hidden by the corner but the bottom soldier would.

I'm not sure I understand. Are you saying the unit at the top is out of cover? It's not - that unit is behind the corner of the top rock just like the one at the bottom. If you drew a diagram, it would look like this (the thick lines represent the edges of the rocks):

If the unit at the top kind of looks like it's out of cover, that's simply because of how the prop is drawn (the edge of the prop in-game extends beyond the edge of the image). If you happen to be able to get the save game to work, you can check this for yourself if you want.

Incidentally, I've added onto the diagram arrows which show what the 'true' trajectory would be like if it wasn't constrained to the sub-tiles used in game. As you can see, according to the game mechanics (ignoring sub-tiles) neither soldier should be able to target the other. It's only because the sub-tiles result in a somewhat rough approximation of the 'true' LoF that a shot is possible at all.

EDIT 1: Actually, having just rechecked, the reverse LoF should be able to target the soldier at the bottom too:

All I've done is traced the same pattern in reverse, as I did in the image below. But, here, it shows that it should still indicate a hit from the soldier at the top. Odd.

Also, as an aside, I note that the LoF traced by the game seems to aim at the first target tile, rather than the centre. In the image I've just posted above, the 'true' LoF differs somewhat near the end. In contrast, if the arrow points towards the first target sub-tile struck...

... then you get an exact match. (This is true in reverse for the shot that was actually possible in-game, too). So it looks like the game isn't tracing all the way to the centre, for whatever reason.

Edited May 17, 2014 by kabill

[Skitso]

Yeah, that example is exactly like Chris' first picture (example where neither has LoF) in that LOS explanation -thread.

[kabill]

Er, yes, that would have been a much more succinct way of putting it!

[SpikeOnUr6]

I'm just worried that people who hasn't played this game for years and discussed it's mechanics here daily for god knows how long will be confused with it.

Exactly. And so the game would get bad reviews by the many who just don't feel like trying to understand.

[Nethris]

That's working as intended, though. If those mesas were actually 2x2 buildings (which would fill an identical footprint), you'd see the top soldier doesn't have his body hidden by the corner but the bottom soldier would.

For this to be true, the mesas/buildings would have to not occupy their entire tiles (and not be symmetrical in what parts of the tile they cover) - is this the case? Otherwise that looks like it might be an inconsistency with how paths are being rounded onto subtiles, but it's hard to tell as the shot selection seems to be not ideal when there's no shot above 0%:

[ATTACH=CONFIG]3734[/ATTACH]

The selection of which of the 0% options is displayed, or presumably used if a 0% shot is attempted anyway, is important - shooting the cover your target is behind rather than your cover is much more effective for suppressing them or attempting to destroy their cover. I'm guessing there's no metric for ranking 0% shots against each other currently implemented, ranking them by how close they get to the target before being blocked would be pretty decent most of the time. Would also make it clearer for checking to see if there is an actual bug/inconsistency here, as the "best" 0% shot by that metric would usually correspond to the shot that works in the other direction.

An oddity is certainly present in this situation though:

I recreated something similar, and the full corner the top guy is next to blocks shots from the bottom guy, but if you take out the wall section of that that's farther away from the top guy (that shouldn't be able to block anything that the wall directly next to him wouldn't) he can be hit like the bottom guy can be, at least in the setup I ended up with.

Saves: In both of these, Rolf Brown (#4) can hit Lucas Hansen (#6), but 6 can only hit 4 in the second save, where one of the walls has been removed by shooting it.

[ATTACH]3761[/ATTACH]

[ATTACH]3762[/ATTACH]

Pictures of the relevant area (entire shot doesn't all fit on one screen):

one way.sav

two way.sav

one way.sav

two way.sav

[UmberSkies]

That's working as intended, though. If those mesas were actually 2x2 buildings (which would fill an identical footprint), you'd see the top soldier doesn't have his body hidden by the corner but the bottom soldier would.

I'm afraid that this makes no sense. Both soldier are behind identical cover if you look at the situation from a top down view, and thus should be protected by their cover. The only reason it looks like one is more exposed is because of the angle of the camera in the game.

Currently the line of fire system is making the game not fun in the slightest, due to the computer knowing exactly what spot will give them a shot, while we have to guess, wasting TUs in an attempt to find a shot due to the unintuitive LOF system.

In a game like this, the LOF system has to be the most clear and easy to understand thing in the game, that way the player knows when he will be safe, and when he wont. Currently, using cover is hit and miss at best, and I still find myself having to move out from cover to shoot, despite playing for a great deal and being active on the forums. I can only imagine the frustration a new player with no understanding of the mechanics would feel when even the beta bug testers cant figure it out.