LOS / LOF problems

[Battlefront.com]

Cpl Steiner,

If you have a save game you can send me I would be VERY interested ot see it. I've not seen something like that in more than a year. IIRC the one time I saw something like that was because a buildnig was put down in the Editor and then deleted, but some of its data remained embedded in the map. So I'd really like to see that one.

Thanks,

Steve

[Thomm]

Originally posted by Battlefront.com:

Strykers are outfitted with optics that are far superior to what some joe can see with his bare eyes, or with binoculars. The thermals help out a huge amount, allowing them to see at night and through smoke. They can also use the zoom in the camera to get a fairly high resolution closeup of things of interest. Vehicles equipped with FS3 can see things AND have them identified by heat sig out to 5km at night and 10km during the day. It's amazing stuff and it does give an edge.

Without being able to support hard data, I got the feeling that the (unbuttoned) recon Strykers consistently spot worse than their dismounted infantry. Thus, I find them (the recon Strykers) rather unfit for their intended role?!

Best regards,

Thomm

[Sergei]

Thomm, are you using your recon vehicles buttoned or unbuttoned? The Stryker RV has no remote weapons system, so to use that FS3 (or the weapon) one needs to be unbuttoned.

[Thomm]

Originally posted by Rollstoy:

... I got the feeling that the (unbuttoned) recon Strykers ...

Best regards,

Thomm

[Sergei]

Right. I don't have the same experience. Especially at ranges over 500 meters the RV has been unsurpassable. At shorter ranges I'm not sure (then it tends to get dangerous to expose the commander).

[Cpl Steiner]

Originally posted by Battlefront.com:

If you have a save game you can send me I would be VERY interested to see it.

Steve,

I have reproduced the problem in another QB using the same parameters. The weather was set to hazy. Perhaps this has something to do with it. If the haze is so bad it blocks LOS a few metres away then I should see a thick fog but I don't see that, as is obvious from my previous screenshots.

[Edit]Just checked using the same parameters except setting weather to "Clear" and it made no difference, so "Haze" is NOT the problem.

Here's my saved QB:

Saved QB

Using the file above I played on and created the following screenshots.

Screenshot 1: Shows a Stryker unable to identify a friendly unit less than 100m in front of it.

Screenshot 2: Shows the scene with no units selected. The front vehicle should easily be visible and identifiable by the friendlies behind.

I should add that the enemy units further down the road ARE visible to some units. I think the left-hand side of the road is blocking LOS but the right-hand side isn't.

[ August 13, 2007, 03:55 AM: Message edited by: Cpl Steiner ]

[panzermartin]

Originally posted by Battlefront.com:

</font><blockquote>quote:</font><hr />How can a buttoned up stryker see better than a human?

Money CAN sometimes buy stuff

Strykers are outfitted with optics that are far superior to what some joe can see with his bare eyes, or with binoculars. The thermals help out a huge amount, allowing them to see at night and through smoke. They can also use the zoom in the camera to get a fairly high resolution closeup of things of interest. Vehicles equipped with FS3 can see things AND have them identified by heat sig out to 5km at night and 10km during the day. It's amazing stuff and it does give an edge.

The Syrian stuff, on the other hand, is generally outfitted with nothing but WWII style vision blocks. On top of that, Soviet design philosophies over the years have consistently downplayed the importance of crew comfort and crew situational awareness. There are blind spots sometimes, ergonomically difficult to use devices in others. In other words, they not only lack fancy gizmos, but they also lack the basics for the old Mk 1 Eyeball.

Steve </font>

[Cpl Steiner]

Originally posted by Cpl Steiner:

</font><blockquote>quote:</font><hr />Originally posted by Battlefront.com:

If you have a save game you can send me I would be VERY interested to see it.

Steve,

I have reproduced the problem in another QB using the same parameters. The weather was set to hazy. Perhaps this has something to do with it. If the haze is so bad it blocks LOS a few metres away then I should see a thick fog but I don't see that, as is obvious from my previous screenshots.

[Edit]Just checked using the same parameters except setting weather to "Clear" and it made no difference, so "Haze" is NOT the problem.

Here's my saved QB:

Saved QB

Using the file above I played on and created the following screenshots.

Screenshot 1: Shows a Stryker unable to identify a friendly unit less than 100m in front of it.

Screenshot 2: Shows the scene with no units selected. The front vehicle should easily be visible and identifiable by the friendlies behind.

I should add that the enemy units further down the road ARE visible to some units. I think the left-hand side of the road is blocking LOS but the right-hand side isn't. </font>

[Bigduke6]

Double post, how about that.

[ August 13, 2007, 08:35 AM: Message edited by: Bigduke6 ]

[Bigduke6]

Originally posted by Bigduke6:

MAJ-

I don't own CMSF so I can't comment on what is modeled in game. I can tell you that the Stryker has a VERY good thermal imager - most of its capabilities are classified - making it realistic for a Stryker commander to see that target before enemy engaged. Also, a .50 round has no problem penetrating the concrete blocks used in construction of most mid east buildings. As such, I think the firefight in your videa was realistic.

Last I heard, thermals do not see through concrete walls.

Thus, a concrete wall would logically offer concealment for ground troopies trying to hide from the Stryker's and its nasty .50, if not cover.

So I would say it is not so realistic for CMSF to issue a virtual Stryker thermals in a computer game, so that the Stryker can see through concrete walls.

Supposedly the idea is to replicate Strykers manufactured by General Dynamics, not ones that run on Kryptonite.

Of course, it could just be a dumb bug.

Edited to fix a typo or two.

[panzermartin]

Here is the LOS blocking effect, perhaps generated by nearby walls.

How does the Target tool work? I had to move the whole squad away from the walls and in the middle of the street to target the tank. They got slaughtered as a result. How can you sneak up from corners if it is required to expose the whole squad to get LOS?

[Battlefront.com]

BigDuke6,

Of course, it could just be a dumb bug.

Uhm... of course it is. You obviously haven't read much of this thread yet, eh? I've already clarified this here and in probably another half dozen threads. It's a prime focus for v1.03.

BTW, the US Army deployed a device to Iraq last year that can see through some types of walls. I don't remember the specifics, but it is a pretty blunt tool from what I do remember. It basically can inform the user if the building is occupied.

Haze could be a reason for the strange LOS conditions. I'm looking into it.

Steve

[Cpl Steiner]

Steve,

If you set up a QB with the following parameters...

Meeting Engagement

Small

Village

Clear

Day

...you will probably get the same map I got. I don't think haze is the problem. It could be the map itself. LOS down one side of the road is almost always blocked. LOS down the other is almost always clear. LOS from one side of the road to the other is also sometimes blocked.

Hope you find the answer.

[JasonC]

Sights on various BMP models

BMP-3

"The gunner has a 1K13-2 sight with a x 8 day channel and x 5.5 night channel, a 1V539 ballistic computer and 1D16-2 laser rangefinder. He also has a secondary PPB-2 day sight with a magnification of x 2.6 and the commander has a dual 1P3-10 sight with x 1.2 and x 4 magnification."

BMP-2

"All three have a number of fixed day periscopes and the gunner has a binocular sight (BPK-1-42) and a TNPT-1 designator. The commander's day binocular sight (Model 1PZ-3) has x 1, x 2 and x 4 magnification and he is also provided with a TNP-165A designator with a TKN-3B binocular sight with a day magnification of x 4.75 and a night one of x 4. The infrared searchlight is mounted co-axially..."

BMP-1

"The driver sits in the hull to the left side and has a single piece hatch as well as three day periscopes the centre one of which can be replaced with a periscope that can be extended upwards for amphibious operations. The commander is seated behind the driver and has a cupola with three day periscopes, the centre one of which can be replaced with a binocular one or a variable magnification periscope... The gunner has a single piece hatch and a 1PN22M1 dual mode monocular periscope sight and a stadiametric range finder."

While the BMP-1 is relatively primitive in the sensor-n-optics department, it is still not mark I eyeballs, and the later models definitely have serious optical and sensors improvements.

[panzermartin]

Originally posted by Battlefront.com:

Haze could be a reason for the strange LOS conditions. I'm looking into it.

Steve

No no, it happens in perfectly clear weather too. Theses caps are from "Strength and Faith" scenario. Its like invisible objects blocking LOS on parts of a street, possibly related to nearby walls, or it could be that bug with data left from objects already removed in the editor you mentioned. It not like los is restricted to 20m radius, one unit can have perfectly los for 200m all along the middle of the street but at some certain spots with no apparent reason the target tool is greyed out. Maybe LOS grids with wall tiles in them are affected.

[Cpl Steiner]

Hi,

I am just replaying "Dawn Patrol" and one of the Uncon pickup trucks was destroyed by the .50 MG of one of my Strykers. However, the LOF went right through 4 solid buildings - i.e 8 solid walls in total. Now I know the .50 MG is a powerful weapon but it isn't that powerful! This makes cover a bit meaningless until it is fixed.

Screenshot below:

[Bigduke6]

Steve,

I have read this, and practically every other thread concerning your new game very closely.

If the crappy Syrian sighting issue is just a bug, then great.

But until the problem is fixed, I can't rule out another possibility.

After all, sometimes removal of a bug does not repair all problems. There is always the chance of a design judgement error as well.

It would be a shame if your new game makes the Americans inaccurately "ueber", and the Syrians inaccurately "untermensch".

But no sense in going down that line of discussion, before you guys get the game patched and running properly.

[Cpl Steiner]

Thinking about it now I'm wondering if in my previous example the Stryker was somehow firing through the windows of the buildings, which seem to all line up. However, if this is the case, I think the "aperture" of each window is too big and needs to be reduced.

[dalem]

Originally posted by Battlefront.com:

There are some things with LOS/LOF that might not look right, but are perhaps not bugs. The underlying game is based on 8x8 meter tiles, which we call Action Spots. Without these it would be impssible for CM to run on a computer that mere mortals have at home. This, unfortunately, produces some situations where the visuals don't quite hold up to "reality". Some of this is unavoidable, unfortunately. The most common cases have to do with extreme angles and terrain that is juuuuust barely cutting into LOS/LOF. I think I know where Normal Dude is talking about and I *think* that might be just such a case.

On the other hand, it could be a bug. The environment is so complex it is entirely possible that certain combinations of elements are producing incorrect results.

A 200x200m CM:SF map probably, on average, has more terrain packed into it than the largest CMx1 map!! This required us to approximate some things based on the underlying 8x8m grid. LOS/LOF, unit placement, basic terrain types, and AI pathing all use the grid for basic calculations and then refine from there. Sometimes the refinements don't look as "correct" as they should.

CMx1 had the opposite issue. The vastly simplified terrain system that used 20x20m terrain grid, usually having a single type of terrain on each square, allowed for very finely resolved LOS/LOF. So you had greatly abstracted and limited terrain, but a very precise way of measuring form A to B.

Units in CM:SF are also infinitely more complex, with each individal soldier represented. These visual references (usually) have meaning and are not abstracted. CMx1 had no such thing. Everything was "head of a pin" and its visual represetnation was purely visual.

Meaning, in CM:SF you get a vastly richer environment, visually and gamewise, but have a small amount of abstraction to contend with. In CMx1 you had a rather simplistic (especially compared to CM:SF!!) environment, with all the abstractions that came with it, BUT a very exact way to trace LOS/LOF and to position units.

What we'll do is continue to hear about things and then decide if there is something specific to explore. I know of one combo with a single type of balcony in a special circumstance that appears to be a problem, for example. That just turned up a few days ago and we haven't had a chance to even double check it.

Steve

Hm. Just re-read this. So... you basically went to hexes after all?

-dale

[SlowMotion]

Here's another example. I was wondering why my tank wasn't shooting at enemy about 200 meters away. T62's commander seemed to have clear view yet nothing was happening.

LOS straight ahead from the T62 looks clear.

So try targeting a Bradley that is clearly visible. Suddenly LOS is blocked.

The same thing from above:

[ August 14, 2007, 11:25 AM: Message edited by: SlowMotion ]

[dalem]

I can't reconcile my earlier posts to anti vehicle fire. Are vehicles assumed to be on a particular "Action Point" (i.e. the closest) at any given time? If accurate and precise ballistic tracking is going on, is it assumed to originate from one action point and arrive at another? I assume that when shot arrives at vehicle model, it is tracked to a very precise point - more precise than it was in CMx1, for example.

Is it a case where each 8x8 tile looks like this:

________

|********|

|********|

|********|

|********|

~~~~~~~

with each * being one of these Action Points, and every modeled object (soldier, doodad, vehicle) is assumed to be at one of these APs within the tile, and the tile's terrain affects all such objects equally with a standard "10%, 15%, 18%, etc." reduction in LOS/FP? And LOS is drawn from one AP to another?

All that makes sense to me based on what's been said but I dunno if it's even close to the truth.

-dale

[bitchen frizzy]

Originally posted by Battlefront.com:

Cpl Steiner,

If you have a save game you can send me I would be VERY interested ot see it. I've not seen something like that in more than a year. IIRC the one time I saw something like that was because a buildnig was put down in the Editor and then deleted, but some of its data remained embedded in the map. So I'd really like to see that one.

Thanks,

Steve

Could the LOS problems and some of the pathfinding problems both be due to something like this? Phantom obstacles would explain a lot, maybe. Or something.

[Battlefront.com]

BigDuke6,

It would be a shame if your new game makes the Americans inaccurately "ueber", and the Syrians inaccurately "untermensch".

Sure would be since we never have, and never will, bias a game to suit some sort of stereotype. Many of the reviewers, in fact, are saying that we've made the Syrian stuff too good. Reminds me of the Sherman vs. Tiger debates in CMBO days... one group convinced we were tipping it in favor of the Germans, another adamant that we were tipping it in favor of the Americans. Pretty hard for us to be doing both, but that's what we accused of doing. Guess we're more creative than we thought

Dale,

Hm. Just re-read this. So... you basically went to hexes after all?

I think you need to reread that again, because besides the attempt at poking fun at us (again), I don't understand why you wrote what you wrote. I can guess, and based on that guess I'll try to clarify something for you since you appear to have missed the point of what I wrote...

A computer game is nothing more than a grid or data. The shape of the grid, the fidelity of it, the complexity, etc. are all variable based on the design. The one thing that is not variable is the amount of hardware that is required to handle it. The more irregular the grid, the more refined it is, the more complex, etc. the more hardware you need to do nothing more than have the environment for things to work within.

When you add units in things get worse, always. The more complex the unit capabilities are, the more of them there are, the greater the range of attributes, etc. the more hardware is required to simulate them. The more complex the environment, the worse all of this becomes.

Therefore, the sum of the game's underlying semi-static environment (Terrain), the objects that can move about (Units), the various game elements (Parameters), and the visual representation (Graphics) determine the hardware requirements. The more refined and complex each is, the more impact it has on the overall hardware needs. But the sum of the parts is always less than the impact of the whole due to the interactions between them. Meaning, each part feeds off the other's complexity and that means taking more juice to run each feature.

In short, there always has to be compromise, there always has to be abstractions. But not all compromises and abstractions are equal.

In CMx1 we GROSSLY abstracted the Terrain. Not only in terms of the variety of types and sophistication of them, but also in terms of their placement in proximity to each other. You want a bunch of trees around a house in CMx1? Can't do it. You want a wall to run right next to a road? Can't do it. You want a small building abutting a large one? Can't do it. So on and so forth. All terrain in CMx1 was simply eye candy because trees, walls, buildings, etc. were not being paid attention to for their actual physical presence. They were just some numbers in the computer and it said "unit 1234 is on square 22-56. That's got a house on it. Is the unit in the house or not. If not it is on grass, if so it is in the house." That's as sophisticated as it is.

CMx1 also had massively abstracted units. Sure, tanks did have a lot of details that other games do not, but they were still essentially abstract bits of data that were tracked in a massive spreadsheet (if you will) and had no actual interaction with the environment. The tanks and infantry were just a plotted point in a data grid. Nothing more than that.

Since all of this data was so highly abstracted, we were able to do things like exact LOS. The terrain mesh and elevation possibilities plus the very limited number of abstracted terrain features allowed this to work. Drawing clear LOS/LOF from one data point to another was therefore quick and easy, but also abstract to the extreme.

Since LOF was combined into LOS we had more savings. We also had anomalies, such as Tank A seeing moving Tank B, firing, and hitting Tank B as it moved behind a house. This was because at the time the unit drew LOS it had LOF. There was no further check to make sure that LOF was maintained because that would have involved a much more sophisticated LOF system and we felt the hardware wouldn't like it.

Now... CM:SF is fundamentally different. The units are represented in visual form as they are in data. Ballistics are therefore also based on that. When a round is fired in CM:SF it is simulated in full from the time it leaves the barrel until it strikes something. The striking is now based on the intersection of the flightpath of the round and the polygonal representation. If something gets in the way, such as a tree or a even an intervening enemy unit, the round hits there instead of the target. If the round gets to where it was aimed and the target has moved, the round impacts where its flight determined it should be, not where LOS traced to.

Terrain is still a grid, as CMx1 was, but it is almost 4 times finer. It is also far more flexible in that you can have lots of different types of terrain in a single tile compared to CMx1's max limit of 2 (I don't think there were any with 3 types mixed). You can now have a road, a wall, some bushes, grass, and trees all simulated in one spot, for example. And although we still have limitations on the variety of road shapes, you can put them down anywhere in any terrain as opposed to CMx1's fixed choices. The buildings in particular are arguably more detailed than all of CMx1's terrain features combined. Floors, doors, windows, balconies... different types, positions, and other variable points are now considered. And, for the most part, they are simulated 1:1. Yes, there are still abstractions in the interior of a building. Yes sometimes infantry are allowed visually fire through a wall instead of a window, but nothing can be simulated perfectly in 1:1 so some compromises must exist.

The important thing to remember is that the abstractions in CMx2 are all less abstracted than CMx1. Therefore, if you think the abstractions in CMx2 are unacceptable, then you might as well remove all CMx1 games from your harddrive because they are far, far more abstracted than CMx2 is. Or put another way, if you think CMx2 is a bunch of crap because of its abstractions, then CMx2 is a pile of excrement that was pooped out by a rather large mammal. Not as big of a dung heap as other wargames, true enough, but compared to CMx1... definitely inferior.

Steve

[ August 14, 2007, 03:54 PM: Message edited by: Battlefront.com ]

[dalem]

Steve-

First, thanks for that explanation - it helps me get my head around what is going on.

Second, my only point with the "hexes" comment was literally to draw a parallel to an abstraction mechanic. I guess where I'm having a hard time (and maybe it's none of my business) is how to reconcile where the LOS and LOF points originate and terminate.

-dale

[Philistine]

Steve,

One of the things on abstraction that has been boethering me a bit (that I think is Dale's point)is mismatch of abstraction between graphics and underlying combat mechanics.

AIUI, CM1's level of graphics abstraction was higher then the combat system. It seems with CMSF, it goes the other way.

What I'm taking from your comments here (and in other threads) is along the lines of: In CM1, what could be done with the terrain was much less--but it allowed for more of a WYSIWYG in relation to the combat engine--e.g. if you appeared to be covered by terrain or structures, you generally were.

It seems this is less the case in CMSF--in that individuals (and possibly vehicles, too?) may appear to be either in or out of cover but in certain circumstances may actually not be for purposes of LOS/LOF.

This is all my distallation of what I'm understanding from what's been said so far (and from playing CM1 and the demo of CMSF).

Abstraction is necessary, both graphically and in the mechanics, and it's really more or less a question of preferences, but I think I generally prefer a higher level of abstraction in the graphics then in the mechanics (if they can't be equal), as it can become frustrating when things are happening that look like they shouldn't.

Once some of the LOS/LOF issues get ironed out in a patch, it may turn out the mismatch isn't as great as it seems.

My $.02, anyway.

--Philistine