Cliff Visibility

[Cetacean]

Hi, I've noticed some major inconsistencies when looking uphill at structures that are resting on cliff edges. Essentially, you can you can only view the buildings from the ground when they rest on northerly or easterly cliff edges. The map I've provided illustrates this pretty clearly . I suspect it may be somehow linked to the building facing constant, but I am by no means certain of this. Does anyone know if there might be a constant or unit property I've missed? I'd really love to achieve 360 degree visibility without workarounds, as they can be less than ideal, aesthetically.

edit: Oops, I forgot to attach the map before!
View attachment cliff test.w3x

 

[GFreak45]

check this

 

[Cetacean]

Yes, that page has been a great reference . Unfortunately, it doesn't really help here, as the issue likely cannot be fixed through unit properties alone (barring some sort of hidden bug/interaction). It was those more arcane uses that I was interested in when I mentioned object fields. Anyways, did you have a look at the map to observe the behavior in question?

 

[GFreak45]

i generally post on here from work so i didnt get a chance to, i can take a look at it tomorow night

 

[Cetacean]

Ok, I've experimented a bit more and it appears that buildings placed on northerly and easterly cliff edges are considered ground level, while those on southerly or westerly edges behave as if they are on high ground. This affects miss chance, vision, visibility, etc. My guess is that this is an engine quirk they either overlooked or simply deemed acceptable, as buildings aren't typically placed this way under normal circumstances.

View attachment cliff test 2.w3x

Workarounds for the sight issues aren't too tricky to manage, but the misses are going to be eminently more challenging. I would love to find an elegant solution that doesn't involve dummy attackers!

 

[Cetacean]

Comments/insights anyone?

Right now I'm working on a few systems and custom models for which this bug is quite relevant, so before moving ahead any further with them I'd really like to get a bit more information. If someone could run the map and take a look around, even if it's just to say "Hmm . . that's pretty strange, but I have no clue what's causing it!" that would at least confirm my findings. Any help would be greatly appreciated!

 

[GFreak45]

honestly i think its one of those qwerks about wc3 that never got fixxed, like how entering regions from the north doesnt work, locust cant be properly removed, etc. I think its something that you just have to deal with

 

[merlinds]

the only way i can imagine to fix this is through visibility modifiers.
Like, every 0.5 sec pick all units of each playet and chek the clif height on the position of each unit, and deppending on the value of it, activate/deactivate the visibility modifiers.
Have in mind, that for each lvl of terrain the cliff height increase by 128 points.
The hard part is that you will have to create a visibility modifier for each building.
But i have no other idea, and i think no one else does have one

Edited:

i`am trying to implement my idea into a map, but i think it does not work . I have to go to bed so i left my map for if you want to see my idea:
http://www.epicwar.com/maps/download/188482/e937631ae1e2e01d7e88d4cf3fca797f9da5cd164eab6e2b/cliff%20test.w3x

 

[GFreak45]

the problem is it registers that it is on the same height, misses and all that dont have an effect, so that means the game is registering its on equal elevation

 

[merlinds]

But, do you think it may work?? or the visibility modifiers are useles in this case?

 

[GFreak45]

they are useless, he is worried about how the game recognizes the height of the unit, not just visibility, misses work incorrectly, all of that

EDIT:zomg zomg zomg i just found something perfect for you, you can combine merlind's idea with this:

• Custom Script: set Integer_variable = GetLocationZ(location_variable)

essentially you can use this and check the height of the unit and the nearby areas, adjust the visibility, and trigger a miss chance

 

[Cetacean]

Hello again! I was away for nearly two weeks w/o internet, but now I've finally had a chance to take a look at your ideas. So before I get to the comments, let me say thanks very much for the input/help!

@GFreak45
Ok, this opens up some really cool possibilities! Since GetLocationZ takes into account hill height, it would be possible to do a system that allows units on hilltops to see into low-lying cliff regions and, conversely, to obscure vision to the the hilltop from the cliff region. I'm going to try and work on it a bit and see how computationally viable this might be, but in terms of flavor and tactics/strategy, even if it's limited to specific areas of a map, this could be an awesome feature. Oh yes, and before I forget, could you give me some more specifics on the bug that messes with entering regions from the north? I've not yet encountered it to my knowledge, but would certainly like to know how to avoid or work around it if possible.

@merlinds
I tinkered with your test map a bit, and I think the major reason it wasn't working for you, is that visibility modifiers are very sketchy when done by region, particularly if the regions are small. If you've ever tried to use regions for environmental effects, you've probably seen this lack of granularity. Fortunately, using circular vis modifiers works flawlessly. So your solution is quite solid, and incidentally pretty consistent with some concepts I was tossing around. Though I'm generally hesitant to implement computationally intensive periodic triggers unless strictly necessary, I think an efficiently coded vjass system on a ~1 sec. timer would be pretty reasonable in this case. Also, the cliff z value you mentioned is good info to have if I end up going for a GetLocationZ-based system.

 

[GFreak45]

Oh yes, and before I forget, could you give me some more specifics on the bug that messes with entering regions from the north? I've not yet encountered it to my knowledge, but would certainly like to know how to avoid or work around it if possible.

its an error that is possible to encounter in gui where the event a unit enters regionX does not fire when the unit enters the region from the north, eventhough they are in the region now and you can check if they are the trigger would not fire, it only happens with some regions and im not sure exactly what instance of triggering/map size and other things like that that directly affect it