Abrams CITV/primary gunner sight limitations

[Grey_Fox]

Following a round of testing done primarily by @Millien on the unofficial CM discord server (link in signature I believe), it appears that the CITV on an abrams is not drawn from the CROWS viewer:

 

Nor is the Primary Gunner's Sight used to draw line of sight from:

 

Instead, it appears that the line of sight for both the PGS and the CITV is drawn from the main gun barrel.

Additionally, the CITV does not provide a 270 degree field of view - instead it would appear to effectively act as a repeater for the gunner's thermal imager. Unless the turret is facing towards an enemy, the CITV does not appear to be used. Testing instructions are to use non-multispectral smoke between tank and hostiles, hostiles should be at a 90 degree angle, and wait.

As such, this would appear to throw into doubt the perceived wisdom that Abrams should be used turned-in to take advantage of the tank's thermal imagers.

Thanks to @Millien for performing the tests.

Edited May 11, 2023 by Grey_Fox

[Millien]

Thank you for posting some of this on the forum, although I want to add a bit more context and information about my own thoughts on this.

First, I am not going to make any conclusion on whether it is better to be turned out or not at this point in the M1. None of my tests indicated anything that would appear to indicate one way or another. However, neither the CITV nor CROWs system appear to scan around the vehicle as would likely be expected. Instead they seem to remain fixed forward in line with the turret front like the gunner's main sight. This isn't seen by the tests above though but with some smoke tests I had done prior:

These M1A2 SEPs are facing away from the smoke screen, one facing at approximately 90*, the other 180* away. Both vehicles are turned in, where the commander is presumably using the CITV or CROWs system to spot targets. Neither is able to identify any partial contacts on the BMP-2 platoon sitting on the far side of the smoke screen about 400 meters away though. These BMP-2s drove up about two minutes after the smoke screen started and have been sitting here for about a minute with nothing from the M1s, this will continue as well:

It isn't until nearly 5 minutes that anything happens when a BMP-2 that was facing towards the smoke screen managed to spot and engage one of the M1A2 in a gap in the smoke after the mission ended. As can be seen here, the vehicle that is being hit still hasn't ID'd any partial contacts even though it's taking fire at this point, and the subsystems and soft factors are not at fault here as far as I can tell.

And in case anyone is wondering that for some reason that these tanks are unable to spot targets through a smoke screen this intense, let's run it again and allow me to turn one of the tanks toward the smoke screen and see the results:

The M1 spotted the BMP platoon basically instantly as it rolled over the ridge and was engaged and destroyed in less than a minute, the BMP-2s never saw a thing.

So in my conclusion, the CITV and CROWs system does not scan around the vehicle, these vehicles can still get spots on targets coming up behind them without a smoke screen so the commander or loader is inherently keeping their head on a swivel but the moment the smoke screen is up they are entirely blind until the turret faces the smoke screen, or the smoke clears enough to see through it with the Mk 1 eyeball.

What about the second part with the line of sight though? Well, that's more interesting and it makes me wonder about some of the deeper parts of the game as a result.
We saw above the M1 was unable to ID targets over the wall, is that a problem with the wall though?

Simply put, no. The stryker here is quite close in terms of height to the M1, the CROWs system is about the same height too, but the stryker can not only ID but engage targets on the other side of the wall while the M1 remains oblivious apart from possible horizontal sharing of contacts. The M2A3 here is quite tall and can look over the wall with the 25mm bushmaster, so it can happily engage the BMP platoon as seen here.

So why does this happen? This is more speculation on my part and I can't infer beyond these three vehicles. But what I think is happening is that the LOS checks on the turret are tied to the main gun. The big difference here is that the M2A3 and the Stryker have main weapon systems that are taller than the wall, so they can draw LOS. While the M1 main gun is blocked by the wall, so it is unable to draw any LOS beyond it, even if the sensors, CROWS weapon system, and the commander are able to physically see over the wall.

Now it's possible that this is due to the testing conditions, the wall might work differently than other terrain and I didn't test that. It's also possible this is something specific to the M1A2 SEP in black sea. But here's my theory: IF the main weapon system on a turret is unable to draw direct LOS to a target, then it will be unable to see that target regardless of any other factors.

Now I hope both of these results are specific to the M1A2. Cause if either is more universal that would actually be somewhat troubling. Especially the second as I have done a lot of turret-down positions with commanders peering above walls or hills but if that actually doesn't do anything that's wasted effort and misleading. Regardless I welcome additional perspectives and tests here, and if anyone is interested you can try the scenarios I created to test this out here:

wall test v3.btt smoke test.btt

Edited May 11, 2023 by Millien

[Artkin]

Good stuff. IIRC there are 5 vertical points where the TACAI can see from. Maybe the CITV is just too tall for that 5th point. I think CMSF or CMA was the first cmx2 right? So maybe this should have been included from the start. 

[Millien]

8 minutes ago, Artkin said:

Good stuff. IIRC there are 5 vertical points where the TACAI can see from. Maybe the CITV is just too tall for that 5th point. I think CMSF or CMA was the first cmx2 right? So maybe this should have been included from the start. 

That is possible, yeah. Although I think it would be pretty difficult to test ourselves. I think a possible alternative would be testing additional vehicles. See if the sight systems for those follow the same criteria as seen here with the M1A2. M2A3, T-90AM, and Oplot-M would be pretty good examples. They all have pretty large CITVs at different heights. If those work while the M1A2 doesn't, then it's specific to the M1A2. If not, then it's a larger-scale problem.

[Pelican Pal]

I'm fairly sure Artkin is right. This stuff was discussed back in 2008/2009 pretty extensively and you could probably track down forum threads from that time period. Steve and Charles participated more in technical discussions so the few years around the CM:BN launch is a treasure trove. Now this is from my memory so I'll have to double check later if I can find the threads again but....

CM calculates LOS from 5 calculated heights. Essentially its a big look up table saying "From this action square what other squares can be seen and at what heights". So while spotting is done on the fly LOS drawing isn't. Those 5 height levels are:

• Prone
• Kneeling
• Standing/Small Vehicle
• Tall Vehicle
• Very Tall Vehicle

Prone and Kneeling are only for infantry obviously while vehicles have access to the remaining three. So for an Abrams it would be

Driver: Small vehicle
Gunner, TC: tall vehicle

And this means that the CITV can't spot from its Very Tall Vehicle slot and gets rolled into Tall Vehicle where the TC is. Since spotting is tied to a person and a person can't be within two different heights.

A note for clarity. The content in blue is my working theory for what is going on here. So once you get to that I'm not 100%. I'm putting stuff together from some notes and memory.

Edited May 12, 2023 by Pelican Pal

[Pelican Pal]

So here is probably the most pertinent post on the issue:

 

Quote

 

The big issue, however, is the requirement for two points of LOS generation (mast and vehicle crew height). This would need a massive change to the game engine, which will likely have a hit to the framerate.

 

In addition to two points of LOS we'd also need the vehicle to have two different chances of being spotted, depending on circumstances. That's because the enemy should have a chance of spotting the sensor box even if it can't see the rest of the vehicle. But the chance of spotting the sensor box should not be the same as the chance of spotting the whole vehicle. And if the sensor box is retracted that's quite different. Again, major coding effort needed here.

 

This is a ton of work for one or two vehicles in a single game Family. We really can't get ourselves bogged down by things like this.

 

There are some additional posts in the thread so I recommend you read it. but they key points seem to be:

- Having multiple points of LOS generation (per crew member) is problematic

- The game has issues with spotting routines when it comes to just sensors (i'd describe the CITV as a sensor)

My guess is that the Brad and Stryker are just getting the hull down concealment bonus applied to them but there isn't a system for a turret down concealment bonus so you can't do that. FWIW this used to be pretty common knowledge, or at least I recall it being fairly well known about a decade ago. There aren't any vehicles (that I know of) that you can go turret down in.

[Grey_Fox]

51 minutes ago, Pelican Pal said:

So here is probably the most pertinent post on the issue:

 

There are some additional posts in the thread so I recommend you read it. but they key points seem to be:

- Having multiple points of LOS generation (per crew member) is problematic

- The game has issues with spotting routines when it comes to just sensors (i'd describe the CITV as a sensor)

My guess is that the Brad and Stryker are just getting the hull down concealment bonus applied to them but there isn't a system for a turret down concealment bonus so you can't do that. FWIW this used to be pretty common knowledge, or at least I recall it being fairly well known about a decade ago. There aren't any vehicles (that I know of) that you can go turret down in.

Good to hear that nothing has changed in 13 years, and that it's now 3 games that have the same issue not just 1.

[Pelican Pal]

If I'm not mistaken this would apply to every game in CM but modern obviously suffers the most. I don't think any vehicle can go gun down in CM

[Brille]

On 5/14/2023 at 9:58 PM, Pelican Pal said:

If I'm not mistaken this would apply to every game in CM but modern obviously suffers the most. I don't think any vehicle can go gun down in CM

I had it multiple times now that my commander could spot something over a crest or behind a wall if he is looking out his hatch, while I couldn't actually shoot the target and the target couldn't shoot at me. 

 

So it is possible though a bit difficult to pull of sometimes (and sometimes bad for the health of the commander) . I would rather use proper inf scouts for this. 

 

But it is sad to hear that the roof mounted optics are somewhat broken. 

But thank you for pointing this out @Grey_Fox and @Millien. 

 

[domfluff]

CM does not calculate LOS from five heights.

The Target tool calculates LOS from five heights, and populates a look-up table at the time the map is built. This is why the Target tool can, and has always, operated without lag, since it's only checking a pre-filled table.

This is also why the Target tool isn't the best measure for when a unit actually has line of sight. It's usually pretty close, but it's not actually correct.

Actual LOS is tracked from eyes/sensors. Where those are defined on the model is the question in this thread.

[Pelican Pal]

Calculating LOS on the fly would be immensely expensive for the game to do and the whole point of the look up table it to improve that so it can reasonably run on home computers circa 2010. The targeting tool being snappy is a happy accident of this system. You wouldn't build a look up table just for the targeting tool.

Here is the loop as far as I can detail it:

LOS: This is a look up table and it lets the game quickly calculate who can possibly see who. (So I think you were inadvertently correct in your post. CM doesn't calculate LOS from five heights but from a single height).

Once LOS has been achieved it drops down into these next systems.

Spotting: This seems to be a combination of dice roll and the ELOS system. Its drawn from the eyes/sensors but the eyes/sensors seem to be in predetermined spots on the AS height map. They don't seem to be truly dynamic.

LOF: This is drawn directly from the gun barrel to the target which is also seems to use the ELOS system.

 

For this particular issue there seems to be a secondary problem that a single individual cannot draw LOS from two points simultaneously.
 

[Millien]

I decided to expand on my tests a little bit. I was slightly less scientific in terms of methodology this time around but I think the results are still fairly effective and overall a bit more widely applicable. I decided to take four vehicles with CITV systems from black sea. Besides the M1A2 this includes the M2A3, T-90AM, and T-84 Oplot. For this, I wanted to see whether or not turret down was possible with other vehicles. The vehicles would move up a reverse slope, spot the targets, and then reverse back down till they lost the spots.

Assuming that turret down is possible on these vehicles they should still have full contacts or at the very least partial contacts that do not degrade once they are unable to engage the targets with their main weaponry. The target is just a typical BTR-82 platoon in a column about 400-500 meters away. This example below is what things looked like after a ceasefire, so everything is visible. Additionally, If turret down is not possible, what is the determining factor as to when a vehicle is in LOS of a target? Is it following a height map with different predefined heights shared between multiple vehicles, or does each vehicle have its own point of origin for LOS?

Now if the LOS system for each vehicle was predetermined on a specific height table, that means each vehicle has a LOS height that is based on that table (Short, Tall, Very Tall Vehicles). This means that following the height map the T-84 and T-90AM should have the same LOS height origin (either short or tall). The Bradley is known to be able to see over objects the M1 was unable to, so we know this has to have a taller LOS point than any of the other vehicles (Very Tall, or possibly Tall if the M1A2 is short). The M1A2, on the other hand, will either be in its own category or follow the same height as the T-90AM and T-84 for drawing LOS to targets (Either Tall or Short if it shares the same height as the T-90 and T-84).

If the LOS system is instead linked to a specific point on each vehicle, then that means that there will be a lot more minor fluctuations in terms of being able to draw LOS and spot targets, it may even be possible to discern differences between the T-90AM and T-84. It will also mean that any cut-off point for LOS will likely be linked to a physical feature on the vehicle which should theoretically be identifiable. At least in theory.
So what do the results show?

Above You see the M1A2 we have used previously. When testing this I drove the vehicles up so they saw the BTR platoon, then I backed them up slowly until they lost the spots. Here the spots for the M1A2 was lost just as the main gun was unable to draw LOS to the vehicles anymore, despite the turret sensors, commander, and 50 cal being able to clearly look over the slope.


Now the M2A3, Unsurprisingly as the tallest vehicle the M2A3 is the farthest back on the slope when it lost the spots, placing it within its own category assuming a height map. What is interesting though is that it also lost the spots the moment the 25mm was unable to draw LOS to the vehicles, even though the CITV system, GPS, and TOW are still visible and still capable of engaging the target in real life.

Now things are starting to get weird, Now we are looking at the T-84. This lost the spots it had before it was able to even have the gun barrel hidden by the terrain! Despite having partial LOS on the terrain according to the target order, it lost the solid contacts on the BTRs and they never came back. I'll talk in more detail on this in a moment but I'll just finish up the results first.

Finally, the 90AM. This was able to back up and lose the spots just after the gun barrel had lost direct LOS to the targets, the CITV, and maybe the GPS, could theoretically see the targets in real life as well. It was noticeably further back than the T-84 but it was still short of the M1A2 and M2A3. So this is also within its own category.

So for the most obvious conclusion and most widely applicable.
Turret Down does not seem to be an applicable tactic with any of the above vehicles, they will be unable to draw LOS and spot targets unless they are able to engage them at the very least. Possibly even more. While I didn't include any pictures, I tried this with commanders both turned out and buttoned up and the results were virtually identical. My suspicion is that this may ring true for most if not every vehicle in CM, but I'll get to the why of that in a moment.

But what about the way that LOS is able to be drawn? is it a predetermined height shared between multiple vehicles, or is it specific to each vehicle? While I am unable to be entirely conclusive, the evidence seems to favour the vehicle LOS points are unique to each vehicle rather than pulled from a height map with three distinct choices.With three out of four vehicles losing LOS the moment the main gun disappears behind the slope and never returning. The only exception was the T-84, which for some reason lost the spots on target prior to being fully out of LOS.

Realistically the reason that the T-84 is so different from the T-90AM is within all likelihood a bug. Although it offers interesting implications as to why its happened, it could have happened with either model. If it is the height table, the T-84 may have been given an even shorter height on accident, like kneeling height for instance. Alternatively, the height map for the T-90 and M1A2 is the same and I just didn't notice how close the vehicles were together in the test. If it's following a unique LOS point for each vehicle then that would mean the origin point for LOS on the T-84 is significantly misplaced resulting in the difference. 

I think I will redo these tests with a bit more care into how far exactly each vehicle backs up till they lose LOS, and add a couple more vehicle types to increase confidence. But from casual observation, it seems more likely to me that the LOS of many vehicles will be tied to the main gun in some fashion, and the main gun alone, rather than a height map or other physical points on the vehicle (sensors and optics, turned out crew, periscopes, etc). This is also an explanation as to why turret down failed to function for any of the above vehicles as well.

There are probably exceptions to this if it was the case, vehicles like the BRDM-2 ATGM carriers definitely do not trace LOS from the ATGM tubes, they need to have some of the hull exposed to be able to see and engage targets. Other systems like the Wirbelwind with multiple barrels, and even bow machine guns on ww2 tanks will likely have their own specific quirks as well. That will only be answered through more testing though. I am also curious about Cold War as well, the M60 is quite tall compared to most other tanks in the series, and the M901 and M150 offer some unique opportunities to test as well, which I or someone else can touch on in the future.

Additionally, @Brille, if it wouldn't be too much trouble could you let me know what vehicles you used which had turret down work with them that would be great, thank you!

Edited May 24, 2023 by Millien

[Pelican Pal]

@Millien

My comments are all drawn from some old posts I have saved mostly from Steve commenting about a decade+ ago. So incidentally around the release time period of CM:BN and CM:BS. You can probably search for these sections in the forums to find the full threads. But they lead me to believe that there are preset heights of some kind although its possible I'm misinterpreting them or that the original comments were incorrect.



LOS: 
 

Quote

LOS is, by far and away, the most taxing element on the game's use of RAM and CPU time. In order to allow CM to run and run well on an average computer there must be an underlying grid to avoid RAM and CPU limitations. The only way a grid works is to have ONE point represent each square (Action Spot) of that grid for purposes of LOS. Therefore, when a unit can not see the center of an Action Spot it is presumed LOS doesn't exist and the system does not attempt to proactively refine LOS. Even with that the system is stressed out pretty hard to deal with CM sized maps and average unit loads. Your computer quite simply can't handle the alternative.

LOF:

 

Quote

When LOF (Line of Fire) is drawn it is between two specific points, not between two Action Spot centers. A unit either occupies a specific spot (soldier) or a range of spots (vehicles, which includes big guns). Each spot has a height associated with it depending on stance (soldiers) or the portion of the unit in question (vehicles). To keep the CPU and RAM from being overloaded there are a sent number of predefined heights, which we refer to as ELOS (Enhanced LOS). When a shot is fired it travels a literal ballistics path between two points. If it collides with something the physics of that collision are dealt with and the effects applied

Spotting:

Quote

Spotting basically works the same way once a unit spots another unit. At that point tracking is point to point with the Action Spot system offering ways to shortcut otherwise massively "expensive" calculations.

Most games have a single height with "modifiers" to increase/decrease exposure in certain equations. CMx1 was like this. CMx2 (as of last year) has what we call Enhanced LOS which simulates 6 different heights; 3 for soldiers, 3 for vehicles, 1 for aircraft. One of the vehicle heights is the same as the tallest soldiers, which from the game's standpoint means it's 1 height instead of 2 different ones. Also, aircraft height is variable while the others are not, so that one is unique.

 

Quote

An entity's height is always a specific point above the underlying ground depending on which of the 6 heights it is at. Vehicle heights are assigned per vehicle and never change due to circumstances. Soldiers, however, have three stances (prone, kneeling, standing) depending on what they actually are doing graphically. But it is just that... the soldier is either prone, kneeling, or standing. It isn't ever "mostly prone, but a little bit sitting up" or "hunched over so not quite kneeling or standing". As computers get more powerful we might include more heights, but it's unlikely since the existing system works well and we'd rather use the computer's resources for other things.

 

[Grey_Fox]

1 hour ago, Pelican Pal said:

Vehicle heights are assigned per vehicle

I think this is the key, if what Steve said is accurate. It may correspond with driver/bow gunner, gunner, and commander. As it is vehicle-specific it would explain the differences in fields of view of the different vehicles.

It would also indicate that the T-84 may in fact have a bug, as @Millien has suggested.

Edited May 24, 2023 by Grey_Fox

[Brille]

11 hours ago, Millien said:

Additionally, @Brille, if it wouldn't be too much trouble could you let me know what vehicles you used which had turret down work with them that would be great, thank you!

Phew... Don't know which vehicles that was honestly but I would say I had it with ww2 style and modern tanks too. 

 

I can surely test it but not before next Monday. Though it would be like your tests... just with the commander sticking his head out.