LOS and Foliage

[Roter Stern]

For an updated v1.06 version, please see this post on page 2. Post below unchanged for continuity.

Prompted by this seemingly outlandish claim by a fellow forum dweller, who shall remain nameless …

You tried to simulate a forest environment very well, unfortunately the engine in its current state does not care much about trees, bushes and that kind of stuff... too bad, I hope at some point this will be fixed/included. This resulted in having a good LOS early in the game and during it, and as result a good situational awareness (not much surprises).

… I set myself a mission to find out once and for all whether investing my life savings*1 into BFC was a grave mistake or a stroke of genius.

The Experiment: To determine the effect of foliage (commonly referred to as “trees”) on the Line of Sight between two given points.

The Setup: A plot of perfectly flat terrain, measured 280m by 1200m, is to be divided lengthwise into five equal 56m-wide lanes. Each lane is to have a position marker at each end marking the center position of each lane. A pair of units is then placed at either ends of each lane – a Red AT-14 ATGM team at one end, and a Blue Stryker ICV vehicle at the other. Each and every one of the ten units involved is to have identical experience and other vital stats. The first four lanes are then filled with the same type of foliage of the varying densities – “1-tree”, “2-trees”, “3-trees”, and “grid” - the fifth lane is left empty as control. See pic.1

The Execution: The experimental scenario is then loaded up as the Red side and each of the five ATGM units is made to face the Blue Strykers 1200m away, but also given short enough arcs of fire to prevent unwanted engagements. Part one of the experiment is to observe if and when the ATGM teams will be able to spot their respective Stryker vehicles. Part two of the experiment is to use the Target Command as an improvised LOS Tool to determine how far the ATGM teams are able to see along the length-axis of their respective lanes. The experiment is then to be reset using a different type of foliage.

The Findings: Part one immediately validated the experiment setup – the control lane #5 spotted its Stryker within a split second, followed by “Grid” lane #4 at 15 seconds, however none of the other lanes were able to spot anything even after 5min.

Part two findings were as following – first using “Tree A” foliage:

</font>

• Lane #1, 3-tree density – max LOS distance of 165m</font>
• Lane #2, 2-tree density – max LOS distance of 225m</font>
• Lane #3, 1-tree density – max LOS distance of 450m</font>
• Lane #4, Grid density – max LOS distance is infinite, along a very narrow FOV</font>
• Lane #5, Control – max LOS distance is infinite, along the entire width of the lane</font>

Then using “Tree F” foliage:</font>

• Lane #1, 3-tree density – max LOS distance of 230m</font>
• Lane #2, 2-tree density – max LOS distance of 395m</font>
• Lane #3, 1-tree density – max LOS distance of 580m</font>
• Lane #4, Grid density – max LOS distance is infinite, along a very narrow FOV</font>
• Lane #5, Control – max LOS distance is infinite, along the entire width of the lane</font>

The Conclusion: The empirical data is very definitive and undeniable – not only is foliage as a whole taken into consideration when calculating the LOS, but density and more importantly (and frankly stunningly) the type of foliage also have a significant effect on the LOS depth.

Given the complexity of the engine involved and the sheer number of foliage+brush+terrain combinations possible we can only speculate about the exact effects each and every one of those factors have on Cover and Concealment, but one thing is for sure – my life savings are safe, my heartfelt thanks BFC!

pic.1

*1 Figure of speech, No actual cash value, Not valid in Quebec

[ February 08, 2008, 12:21 PM: Message edited by: The Louch ]

[SlowMotion]

"not only is foliage as a whole taken into consideration when calculating the LOS, but density and more importantly (and frankly stunningly) the type of foliage also have a significant effect on the LOS depth."

It's really stunning if you must have 165 meters of trees between two units in order to have cover. In CMx1 I think two Woods tiles was enough, 2x20m.

[jomni]

It seems that two woods tiles in CMx1 contains more trees than 165 meters of 3-tree tiles. 1:1 representation also applies to tress as well in CMx2 unlike before.

[athkatla]

Originally posted by SlowMotion:

"not only is foliage as a whole taken into consideration when calculating the LOS, but density and more importantly (and frankly stunningly) the type of foliage also have a significant effect on the LOS depth."

It's really stunning if you must have 165 meters of trees between two units in order to have cover. In CMx1 I think two Woods tiles was enough, 2x20m.

This isn't CMx1!

[KNac]

Ok thanks for the education, but check it out this:

It's really stunning if you must have 165 meters of trees between two units in order to have cover. In CMx1 I think two Woods tiles was enough, 2x20m.

Without doing intensive testing as yours, the first (and second) impression one gets is this, after playing several scenarios.

More important though, what's the real level of concealment & COVER these tiles provide? Specially to infantry? Yes, is very fine they block LOS to X meters (with the stats you wrote, is possible to simulate dense forest? Where LOS shouldn't be more than 50 or 75m?), but do they provide any real cover while infantry is there? I know is an other topic.

P.S: as much as it could seem that my intention was to irrationally rant about the game, nothing farther away from reality it could be, indeed I've oen of these who have defended the game in the main forum, but if I seem to find limitations I don't hesistate to talk about them. If soemone proofs me wrong, excellent, more happy I will be that the game is better.

[ August 15, 2007, 02:36 AM: Message edited by: KNac ]

[SlowMotion]

Now that we have several kinds of trees and also you can pick 1-3 trees per tile IMO it should be possible to somehow simulate more dense forests also. Maybe one of the tree types could block LOS enough that it's possible to get closer to real thick forest. Like KNac I think it should be clearly below 100m.

I tried to find some examples of such forest from Syria and found quite a few. Not very high resolution pictures, but if you compare some of these pictures to what we get using the most dense forest in Scenario Editor at the moment, there's a big difference.

http://www.souria.com/syriaphotos/index.asp?oo=20

[Roter Stern]

Well first all of don’t confuse having an LOS on a particular spot and actually being able to identify enemy units at that spot. For example I used vehicles for the first part of the test, where as initially I wanted to use infantry squads, because even the empty Control Lane was not able to identify it’s respective enemy squad at that distance, despite having a 100% clear LOS and being able to spot a vehicle in under a few seconds. And I don’t even want to go into LOF and how each round can now be potentially blocked or deflected by foliage… there is enough material there to write a PhD thesis on.

Secondly this is not an absolute test designed to find out how well the in-game foliage compares to that of real life or even to that of CMx1. As for foliage seemingly not providing any cover in most scenarios, you have to keep in mind that we’re dealing with a mostly arid environment – basically a desert. So it is hardly a surprise that 30m of arid foliage doesn’t provide as much cover as 30m of lush forests of North-western Germany. That is something that can be debated endlessly, and is not my intent. Instead that was a relative test with-in the controlled parameters of the game, designed to demonstrate that foliage is now modelled in utmost detail and sophistication.

Essentially you can argue that instead of CMx1’s three types of foliage tiles, we’re now looking at least 24 different types of foliage – and that’s not even counting in grass and underbrush, which would swell the number of combinations well into the hundreds.

To me that is a beautiful thing, and whether a 3-density tree-D tile with tall yellow grass and a sprinkling of brush is supposed to block LOS at 30m or 530m is up to the developers to figure out and for us to accept.

[ August 15, 2007, 07:59 AM: Message edited by: The Louch ]

[SlowMotion]

If you look at some of the pictures from my previous post, it's clear that Syria is not just desert. There are also areas where you have dense forest.

Some more pictures in this thread:

http://www.battlefront.com/discuss/ultimatebb.php?ubb=get_topic;f=52;t=002476

IMO this foliage LOS differs from most game related issues discussed in these forums. This is not secret knowledge about some latest army technology that only few people know or understand. I'm sure all readers have walked in thick forest many times and know that just one tree can block LOS entirely. So based on this I hope some way of simulating really thick forest would be added. Whether the developers will add such option and when, that I can surely accept.

[hoolaman]

Originally posted by The Louch:

To me that is a beautiful thing, and whether a 3-density tree-D tile with tall yellow grass and a sprinkling of brush is supposed to block LOS at 30m or 530m is up to the developers to figure out and for us to accept.

I thank you for your efforts in compiling this test. I don't really think the results are that satisfactory as far as accuracy of the simulation, and it is not our only option to be happy with this.

As SlowMotion posted, we have all been in real life woods, and seen thick growth that blocks sight of anything beyond 30-50m. This is really in line with what we accept from CM1 ie. thick woods blocks LOS at about 40m.

Simply CM1 meets what we expect from reality and CMSF does not. Syria's terrain resembles somewhere like Italy in many areas, where forests and thick scrub exist as well as more arid areas. If the game cannot simulate these it cannot even simulated its chosen theatre, and does not bode well for WW2.

[jomni]

I guess the way to simulate thick woods is to create tiles with more than 3 trees in them (which is not currently done). I believe an abstracted terrain modifier for LOS and protection is not in line with the current engine (1:1 representation). And if they ever make it possible to stuff 4 to 5 trees in a tile, then it may adversely affect game performance / frame rate.

I'd really like to have thick forests for some variety... But I enjoy urban battles the most so fixing this is not a priority for me.

[PrezCartman]

Louch, excellent testing, certainly valuable to scenario devs.

I was wondering what the underlying terrain you were using was?

I think this is definitely something that more testing can be done on. Particularly creating a ranking system of which foliage types create the most cover and also what the effect is with various terrain types. Who knows for all of those who want to simulate dense forests if you use the right trees and terrain it may be possible even in the current engine because it appears to me that Louch was using flat, open terrain for this test.

[SlowMotion]

I did some own testing on this and it looks like we already have solution for creating thick forest. From the original test I simply assumed that trees A and F had been selected because they were the most and least LOS blocking tree types available. It's not like that.

Especially tree D blocks LOS very nicely. Don't have results as accurate as in the first post, but it seems to be the Woods tile of CMx2.

So sorry for my original assumption. Before making any further feature requests, I'll do more testing from now on.

[KNac]

It would be ncie if between all we could do some sort of chart for effects on LOS, cover etc.

[Mishga]

Excellent thread, learned a lot from it. Thanks

[Roter Stern]

Originally posted by SlowMotion:

I did some own testing on this and it looks like we already have solution for creating thick forest. From the original test I simply assumed that trees A and F had been selected because they were the most and least LOS blocking tree types available. It's not like that.

Especially tree D blocks LOS very nicely. Don't have results as accurate as in the first post, but it seems to be the Woods tile of CMx2.

Actually I picked A because it was the first one on the list and I picked F (palm trees) because it was the most distinct looking from the rest of the tree-tiles. Since I was just trying to establish if there was a relative difference between the tree types, I felt that finding the absolute most and least LOS blocking tiles was unnecessary.

That said, I’d live to see some actual number from your experiments.

Originally posted by SlowMotion:

So sorry for my original assumption. Before making any further feature requests, I'll do more testing from now on.

I’m glad you came to that conclusion – this is exactly what I’m trying to prove with this post – CMSF engine is so complex that to make any ‘blanket’ statements about it without doing at least some form of basic objective testing is simply inadequate.

Originally posted by PrezCartman:

I was wondering what the underlying terrain you were using was?

Who knows for all of those who want to simulate dense forests if you use the right trees and terrain it may be possible even in the current engine because it appears to me that Louch was using flat, open terrain for this test.

I was using just the basic “dirt” with no other layers other than the trees, just to keep the number of variables influencing the result as low as possible. But of course if you wanted to simulate a dense forest you would also add tall grass and lots of brush, which would unoubtfully have a significant effect on the LOS.

Originally posted by Hoolaman:

Simply CM1 meets what we expect from reality and CMSF does not.

I volunteer Hoolaman to perform tests on all 7992 Ground-Foliage-Brush-Road combinations (no, this number is not made up) to determine absolute LOS blocking distances. I’ll expect a conclusive result with in 7 days, at which point we’ll be judging whether or not CMx2 is accurate to what we expect in real life. Have fun mate!

[ August 21, 2007, 07:20 AM: Message edited by: The Louch ]

[SlowMotion]

The Louch:

"I’d live to see some actual number from your experiments"

Well, I just tried to see if a thick forest is possible and thus didn't make any notes. I won't make any more testing about this before 1.03, because there are many LOS problems at the moment. We might get new kinds of numbers soon.

[Roter Stern]

Originally posted by SlowMotion:

Well, I just tried to see if a thick forest is possible and thus didn't make any notes. I won't make any more testing about this before 1.03, because there are many LOS problems at the moment. We might get new kinds of numbers soon.

Good point about 1.03, forgot all about it … *crosses fingers for this Friday*

None-the-less, do you recall what was the biggest LOS decrease you were able to achieve?

[SlowMotion]

It was close to CMx1 Woods numbers, 40 meters.

[Roter Stern]

I rest my case

[z1812]

Hi All,

Very informative topic. I am building a scenario and I want my opfor troops to hide in a glade of trees.

The tile type I used was 3 tree palm. Even though I had the opfor hiding at the back edge of the trees the blue force could still see them.

Now I know to use the "D" tree type for better density. I will try this tonight. Thanks for a useful topic.

Regards John

[Roter Stern]

Expanding a little bit on the topic of least and most LOS blocking tiles – a comparison between LOS depth through 3-tree density of the foliage tiles:

Tree-D, apx 25m

Tree-C, apx 55m

Tree-A, apx 185m

Tree-F, apx 215m

Tree-E, apx 225m

The Rest: The Tree-B type and the three brush-types of the foliage tile, the brush tile, and the grass tiles, all seem to provide an infinite LOS depth.

Keep in mind that being able to see x-meters into a tree line does not equal being able to actually spot units – at best that means you will be able to order an area fire that deep into the tree line.

For those screaming "murder", notice how the LOS depth of 3-Tree-D tile is identical to that of CMAK’s Woods tile.

[ August 22, 2007, 09:21 PM: Message edited by: The Louch ]

[SlowMotion]

Those numbers give scenario makers approximate idea of how many tiles is needed to create cover. Things are a bit trickier for the player though. Many of the tree types look very similar especially when looking down from a high camera position, not from ground level.

Once modding possibilities of this game are opened, maybe someone will make a mod that makes it easier to notice different tree types. Like there were those gridded grass mods that made it easier to see ground elevations.

[Chelco]

The Louch,

Awesome job. You da man!

Thanks for doing this.

[aka_tom_w]

its an OLD thread but its worth a bump

as the game engine even after 5 patches still works the same way..

[Webwing]

--

I didn't see this before! Brillant!

Thanks aka_tom_w for bumping it.

--