reported CBU 87/103 Damage issues.

[Tippis]

1 hour ago, NineLine said:

Hey guys, I appreciate a good bug thread, but I am not sure where we are going with this one, what are we expecting?

We are expecting CEM submunition (and possibly HE, but that's a different set of weaponry) to be much more effective on indirect (frag and blast) hits on soft targets, and possibly — but that's more open to discussion — slightly more effective on direct hits on hardened targets, up to a point.

 

The problem with this is that, as far as anyone has been able to tell, frags aren't simulated at all, and it's all cooked into a general blast effect that is itself too weak in relation to the death limits of those targets to reliably incapacitate them, and that the whole “up to a point” bit means the bast effect can't just be boosted willy-nilly. Thus…

 

1 hour ago, NineLine said:

The damage model in its current state to do more than it was ever designed to do with out upgrading it?

…this would be a good interim solution until we get component damage on ground vehicles. See the post above about shuffling around the damage effect limits to create more sensible distinctions and triggers for the different damage states as one suggestion — I'm sure there are others.

 

It feels like the only possible solution because, unless you update the damage model to also calculate frag patterns (which arguably isn't that complicated — it's inverse-square-law vs. a known outline area), the only other solution available at the moment is to increase the CEM blast damage by adding some compensating explosive mass to them. The problem with that solution is that it has a very sharp upper limit, after which each CEM will suddenly be able to overcome the damage mitigation of armour on units that by all rights should be fully protected. At thee same time, everything else in the area would just face mass-extinction, when in reality they stand at least some small probability of survival. Therefore, just increasing damage won't actually solve the problem in a useful way.

 

So while the more sensible solution would indeed be to adjust the damage model such that it has a more sensible distribution of the damage states, the question becomes: is this at all supported, or does this actually mean that we're in practice talking about a completely new damage model anyway, in which case it's not an interim solution at all? Put another way, is the only viable interim solution not viable (or interim), so we might as well wait for the full damage model rework?

 

1 hour ago, NineLine said:

Things like this are too vague and have no sense

That's ok. It was a general statement of the overall scope of the problem: that the indirect effects aren't modelled properly enough — and this is a known fact since forever — such that the only time you see any effect is when you get a direct hit on something that dies from a direct hit. A subset of this problem is that only kills — i.e. only a complete reduction of HP to 0 — have an appreciable effect. Yes, in practice, a unit's ability to act is already diminished as it takes damage, but that is very hard to identify from the air.

 

1 hour ago, NineLine said:

I need good reports, or this might as well be in Chit Chat.

One immediate problem is that you've locked off the lua files so we can't easily explore the exact damage; hitpoint, and armour values of munitions and units. There are community tools available to get to some of that, but the accuracy and completeness of those is always questionable. So documentation is locked off. If it's kill probabilities from real-world scenarios you're looking for, they'd quite easily fall afoul of the rules against posting that kind of documentation.

 

Experiments can be done, but they will always be vague and subject to uncontrollable RNG. They also run into the problem of the vague nature of the damage states. I can check on the F10 map that a vehicle is at what looks like half damage… but how do I know? The health bar is tiny; it is not moving, but I can't tell if this is because it's incapable of moving or if it has hit the can't-move-now damage state; it's not firing, but I can't tell if that's because it doesn't see the target, doesn't feel the target is viable, or if it has reached the can't-fire-any-more damage state. While the exact HP% can be found out through (excessive) scripting, that's not even telling half the story.

 

You might be asking for something that can't reliably be delivered here, unfortunately. Moving the whole thing to chit-chat doesn't make much sense, because it is a “problem” with weapons, and thus belongs in the “bugs and problems” subforum, but it also might not be an outright bug because it is working as intended… it's just that “as intended” is exactly where the problem is.

 

So, what, exactly, would be a convincing example of things not working as they should, and to a degree where it could be consider a “valid” bug report? What would tracks show that isn't already known?

[NineLine]

24 minutes ago, Tippis said:

One immediate problem is that you've locked off the lua files so we can't easily explore the exact damage; hitpoint, and armour values of munitions and units.

First off, this has already been made available by a community member.

 

24 minutes ago, Tippis said:

That's ok. It was a general statement of the overall scope of the problem: that the indirect effects aren't modelled properly enough — and this is a known fact since forever

More vague, give me examples of what is not getting damaged as you see fit and I can report it.

 

24 minutes ago, Tippis said:

You might be asking for something that can't reliably be delivered here, unfortunately. Moving the whole thing to chit-chat doesn't make much sense, because it is a “problem” with weapons, and thus belongs in the “bugs and problems” subforum

So give me a proper report with proper information, not vague statements or feelings or guesses.  Right now this is a discussion not a bug report.

 

From the Hind FAQ as it is related to some extent but applies to clusters.

 

Why do my rockets not seem to have an affect on the target?
Splash damage, or the damage effect from a warhead hitting nearby, it composed to two primary elements: the blast shockwave and fragmentation from the weapon’s casing. While the former is modeled, the latter is currently not (it is planned). To help alleviate this, the blast shockwave effect is increased to provide a better slash damage result.
 
The warhead type also needs to be considered for slash damage. A shaped-charged warhead will have a much smaller splash damage radius than a high-explosive or fragmentation warhead.
 
When the warhead detonates on or nearby the target, the blast value is then calculated to determine if the blast can penetrate the armor with a 20-30% randomized value. Only if the armor is first penetrated will a damage affect be calculated and applied. As such, even if multiple weapons hit on or around the target, there will be no affect unless the armor is first penetrated. Think hard outer shell and gooey interior.

It should also be noted that slash damage effects may not be visually obvious because we currently have only three damage states for ground vehicles: undamaged, undamaged model but with fire and smoke, and a destroyed model. So, you may often have damaged a vehicle, but not see any visual results. However, based on the level of overall damage, the unit is affected regarding mobility, reaction time, and weapons accuracy. It is our hope, in the future, to add multiple elements of visual damage (advanced damage modelling for ground units). This will not just have benefits weapon effects, but also for modules like Combined Arms.
 
Because of the lack of detailed, visual damage modeling, the best indication of unit damage is from the F10 map view. From this view, you can enable “health bars” that provide the general status of the unit with resulting performance effects listed above.

 

24 minutes ago, Tippis said:

So, what, exactly, would be a convincing example of things not working as they should, and to a degree where it could be consider a “valid” bug report? What would tracks show that isn't already known?

Beyond what we have stated needs some love, what is wrong, and show me with tracks and if you have it real world data to back it up, I know you all think I am mean or whatever for asking for these things, buts its the very same thing I will be asked for when submitting a bug. Thanks.

[Nealius]

What I find interesting after watching the video, is that the results on the internal build that Wags is using are not reflective of our current consumer OB build. Both in terms of damage dealt to the ground units by the submunitions, and the mobility/reaction time/weapons usage impairment mentioned in the notes at the end of the video. Is this improvement to the damage modeling going to be in the next OB update?

Edited July 12, 2021 by Nealius

[Frederf]

I agree. This is a discussion, not a bug report.

[NineLine]

31 minutes ago, Nealius said:

in terms of damage dealt to the ground units by the submunitions,

Can you show me something different from release?

 

31 minutes ago, Nealius said:

and the mobility/reaction time/weapons usage impairment mentioned in the notes at the end of the video

Its already in the public version, uploading a little video to show this.

[NineLine]

Ok, two low production value vids, just to show some quick testing I did on a Sunday evening, if warranted I can do something more put together, but for now these are good for this thread, first video I set up a little outpost in Syria for something else, so I decided to see what it would look like after 2 JSOW As, now I doubt anyone would generally use this weapon in a town setting like this, but it suits the purpose. You can see the F10 map before and after at the start and end of the video.

 

 

Next video:

This is a very basic video showing what happens to the AI with damage, as stated in the latest Wags video, AI will lose mobility, response time and weapon usage.
 

In the video I have a BMP1 going at top speed, I have him set to weapons hold. I hit him when he comes in range, you can see him slow when he starts to take damage (Watch the BDA window). I have a zone set up to set him to go weapons-free, but the only thing you see from him is that the crew buttons up, he doesn't return fire, no doubt too damaged at this point to do so. This video was done using the latest Public OB Release.

 

 

[Xavven]

2 hours ago, Frederf said:

I agree. This is a discussion, not a bug report.

 

Agreed. I think this thread started out as a discussion and then was moved by a moderator to the bugs section, and now we are being told this isn't a proper bug report. I don't think it was ever meant to be a bug report in the first place.

[Nealius]

4 hours ago, NineLine said:

In the video I have a BMP1 going at top speed, I have him set to weapons hold. I hit him when he comes in range, you can see him slow when he starts to take damage (Watch the BDA window). I have a zone set up to set him to go weapons-free, but the only thing you see from him is that the crew buttons up, he doesn't return fire, no doubt too damaged at this point to do so. This video was done using the latest Public OB Release.

 

Are there any units that have not received this feature, namely AAA or WW2 assets pack? The 88cm flak continues to provide accurate fire at a high rate of fire with the health bar down around 5% or less. I don't have CA so replicating will take a lot of time (and bombing trials) to get hard evidence. Overall it seems to be inconsistent between units. 

Edited July 12, 2021 by Nealius

[nighthawk2174]

Yeah something to consider is that some vehicles like the flak should just stop working completly if the pattern is even near the piece as its likely the crew would get shredded if the are not taking cover.  Same with trucks and other light vehicles being close enough to only damage them in game right now would probably result in the driver getting swiss cheesed irl.

[Lurker]

In Nineline's video, apart from one tank which has lost a sliver of "health", those T55s are completely undamaged. To me this indicates that there is definitely room for improvement in the damage modeling of ground units. 

[Silver_Dragon]

58 minutes ago, Lurker said:

In Nineline's video, apart from one tank which has lost a sliver of "health", those T55s are completely undamaged. To me this indicates that there is definitely room for improvement in the damage modeling of ground units. 

 

Actualy, the life bar has a temporal "placebo" and dont expected a improve over the old system. That only go to change when the vehicles turn to a "Advanced Armour Damage Model", all systems get modeled, the weapons penetrations and effects turn as a actual WW2 aircrafts damage model on all DCS vehicles.

 

 

Other points, the implementation of a correct wheel / track phisics, to get a "inmovilized killl" by a mine, explosion or weapons and the correct crew implementation, to be close hatches when fired, wonded, escape and get panic.
 

Edited July 12, 2021 by Silver_Dragon

[Tippis]

Data, eh…?

Well, ok then…

The problem isn't the JSOW, as such — it's just a delivery vehicle. The problem is the BLU-97/B. The easiest way I found to deliver a bunch of those and quickly iterate on the delivery is by packing a single CBU-87 onto an A-10C II and dumping it over a bunch of targets. The release parameters can be adjusted using LUA, and the targets can be precisely placed and mass-adjusted by doing search-and-replace in the mission file. As such, I built two missions: Cluster.miz and Cluster-Tight.miz — the former using standard settings; the latter using HOF 900 and RPM 1000. The Hog is placed such that it sits exactly 2000m above the targets; by unpausing and immediately active-pausing, the CBU dispersal pattern will hit right in the middle of the target.

521 units are placed in two overlapping grids: one 20×20, each 20m apart; and one 11×11, also 20m apart. This creates a central area with overlapping 10m-offset rows, with 14m between closest units. The centre-point between any quad of units will be no more than 10m away from any one of those units. With very few exceptions, the BLUs will land within this inner area, meaning that within a small margin of error, 265 units — just over half — are within the hit zone. The first version of this uses Infantry AK units — by searching replacing all of them, they were quickly replaced by Ural-375 trucks, BRDM-2 scouts, BMP-1 IFVs, and T-55 MBTs.

 

I made five passes for each unit type, using the F10 map at an appropriate zoom to get a good picture of what happened, and also checking the exact hit and death count in the debrief screen (which, incidentally, showed a bug: infantry can apparently die twice so now this is a legit bug thread ). A “hit” in this case indicates a damage event — it became quite clear that units can be hit, but it doesn't register and show up as such in the debrief unless they actually take damage, however minute. I don't know if it has ever been established which health bar colour corresponds to what damage state, or if such a direct correlation even fully exists, so I'm going to call them “damaged” (green, still working just fine), “crippled” (yellow, probably some reduced capability), “incapacitated” (red, definitely not doing anything useful), and “dead”.

 

As a real-world-data source, I stumbled across this Norwegian Defence Research Establishment report — it is publicly available as FFI is a public government agency and it doesn't contain any classified data. It's mainly just maths, really… But it does offer a pretty hefty analysis of the lethality of the BLU-97, and a few other things that go boom, both per component (frag, shock, incendiary) and in total, against prone and standing human targets. The interesting results for sanity-checking what DCS does come at page 70 and onwards. Perhaps most immediately interesting is an estimated lethality radius of ~20m, irrespective of stance due to the combined effects.

 

So… in-game, against Infantry AK:

Hits	Green	Yellow	Red	Dead
58		32	6	12
58		25	15	9
58		16	12	16
64		19	17	14
59		29	14	8

 

In total, 1010 BLUs thrown at 1,325 units yielded 297 hits (29% of BLUs); 121 crippled units (9%), 66 incapacitated units (5%), and 59 dead units (4%). As few as 1 and as many as 3 hits were applied to targets that died, but some survived having two hits applied to them.

 

This is for BLUs with a calculated 20m lethality radius landing no more than 10m away from any given target inside the dense centre box.

 

I only tried a single run using a tightened-up dispersal pattern (so 202 BLUs against 265 targets), and it yielded 20 hits (10%), 5 crippled units (2%), 3 incapacitated (1%), and 12 dead units (5%). Not really worth the button-fiddling, tbh…

 

Same experiment, this time against Ural-375s:

Hits	Green	Yellow	Red	Dead
26	4			22
26	5			21
33	5			27
35	10	1		23
26	5			21

 

In total, 1010 BLUs thrown at 1,325 units yielded 146 hits (14% of BLUs); 29 damaged units (2%), 1 crippled unit (0.1%), and 114 dead units (9%). Dead units were “hit” 1–2 times each. It is apparently very hard to hit a truck and not kill it, unlike the much sturdier infantry(?!).

 

Same experiment, this time against BRDM-2s:

Hits	Green	Yellow	Red	Dead
10			8	1
13			11	1
10			10
13			9	2
10			10

 

In total, 1010 BLUs thrown at 1,325 units yielded 56 hits (5% of BLUs); 48 incapacitated units (4%), and 4 dead units (0.3%). One hit incapacitates; two hits kill.

 

Tightening up the dispersal pattern yielded slightly better kill chances in a smaller area:

7 dead, 5 incapacitated in two runs, of which 3 died without the double hit that was needed in the large-area runs. This can only be explained by more tightly-grouped overlapping blast effects combining to overcome the armour.

 

Same experiment, this time against BMP-1s:

Hits	Green	Yellow	Red	Dead
24		22	1
18		12	3
20		12	4
14		14
17		13	2

 

In total, 1010 BLUs thrown at 1,325 units yielded 93 hits (9% of BLUs); 73 crippled units (6%), 10 incapacitated units (0.8%), and no dead units. One hit cripples; two hits incapacitate. An undocumented test I did while setting the whole thing up showed that three hits would generally kill.

 

Tightening up the dispersal pattern increased the odds of getting the requisite hits:

The tighter pattern yielded 11 crippled,  3 dead, and one unit that survived a double hit and became incapacitated.

 

Finally, the same experiment against T-55s:

Hits	Green	Yellow	Red	Dead
12	11
5	5
6	6
6	6
7	6

 

The images become almost irrelevant due to how tiny the slivers of damage are:

In total, 1010 BLUs thrown at 1,325 units yielded 36 hits (4% of BLUs); a grand total of 34 damaged targets (3%) and nothing else. Two hits will put a dent that can be seen in the health bar; one hit can only barely do so.

 

Tightening up the pattern does nothing of value:

A total of 7 hits creates 6 damaged targets.

 

Unfortunately, as mentioned, the community tools to data mine do not cover the full set of information avialable in the old Lua files. So we don't know the hit points of these units, for instance. So there's no way of telling exactly why we're seeing the sometimes baffling differences between units and dispersal patterns.

 

With this (and the mission templates) in hand, I leave it to others to expand the data set and offer their analyses and opinions on the reasonableness of these results in relation to any real-world performance data. Changing to a Hornet or Viper and a JSOW would be more in line with the title of the thread, but should otherwise make very little difference in terms of the percentages per BLU. At best, you get slightly tighter or wider patterns, but the limits for killing things should remain the same. If not, we have a really huge bug on our hands…

 

Another option would be to replace the -87s with -103s and throw them at WP1. Again, this should make zero difference, and if it does, we have a problem. The fun-inclined might also want to try to replace the Hog and -87s with some other, non-CEM cluster munition and see what that does (but the starting position will probably have to be adjusted so the bombles land where they should).

Edited July 12, 2021 by Tippis

[Northstar98]

11 minutes ago, Tippis said:

Data, eh…?

Well, ok then…

The problem isn't the JSOW, as such — it's just a delivery vehicle. The problem is the BLU-97/B. The easiest way I found to deliver a bunch of those and quickly iterate on the delivery is by packing a single CBU-87 onto an A-10C II and dumping it over a bunch of targets. The release parameters can be adjusted using LUA, and the targets can be precisely placed and mass-adjusted by doing search-and-replace in the mission file. As such, I built two missions: Cluster.miz and Cluster-Tight.miz — the former using standard settings; the latter using HOF 900 and RPM 1000. The Hog is placed such that it sits exactly 2000m above the targets; by unpausing and immediately active-pausing, the CBU dispersal pattern will hit right in the middle of the target.

521 units are placed in two overlapping grids: one 20×20, each 20m apart; and one 11×11, also 20m apart. This creates a central area with overlapping 10m-offset rows, with 14m between closest units. With very few exceptions, the BLUs will land within this inner area, meaning that within a small margin of error, 265 units — just over half — are within the hit zone. The first version of this uses Infantry AK units — by searching replacing all of them, they were quickly replaced by Ural-375 trucks, BRDM-2 scouts, BMP-1 IFVs, and T-55 MBTs.

 

I made five passes for each unit type, using the F10 map at an appropriate zoom to get a good picture of what happened, and also checking the exact hit and death count in the debrief screen (which, incidentally, showed a bug: infantry can apparently die twice so now this is a legit bug thread ). A “hit” in this case indicates a damage event — it became quite clear that units can be hit, but it doesn't register and show up as such in the debrief unless they actually take damage, however minute. I don't know if it has ever been established which health bar colour corresponds to what damage state, or if such a direct correlation even fully exists, so I'm going to call them “damaged” (green, still working just fine), “crippled” (yellow, probably some reduced capability), “incapacitated” (red, definitely not doing anything useful), and “dead”.

 

As a real-world-data source, I stumbled across this Norwegian Defence Research Establishment report — it is publicly available as FFI is a public government agency and it doesn't contain any classified data. It's mainly just maths, really… But it does offer a pretty hefty analysis of the lethality of the BLU-97, and a few other things that go boom, both per component (frag, shock, incendiary) and in total, against prone and standing human targets. The interesting results for sanity-checking what DCS does come at page 70 and onwards. Perhaps most immediately interesting is an estimated lethality radius of ~20m, irrespective of stance due to the combined effects.

 

So… in-game, against Infantry AK:

Hits	Green	Yellow	Red	Dead
58		32	6	12
58		25	15	9
58		16	12	16
64		19	17	14
59		29	14	8

 

In total, 1010 BLUs thrown at 1,325 units yielded 297 hits (29% of BLUs); 121 crippled units (9%), 66 incapacitated units (5%), and 59 dead units (4%). As few as 1 and as many as 3 hits were applied to targets that died, but some survived having two hits applied to them.

 

This is for BLUs with a calculated 20m lethality radius landing no more than 10m away from any given target inside the dense centre box.

 

I only tried a single run using a tightened-up dispersal pattern (so 202 BLUs against 265 targets), and it yielded 20 hits (10%), 5 crippled units (2%), 3 incapacitated (1%), and 12 dead units (5%). Not really worth the button-fiddling, tbh…

 

Same experiment, this time against Ural-375s:

Hits	Green	Yellow	Red	Dead
26	4			22
26	5			21
33	5			27
35	10	1		23
26	5			21

 

In total, 1010 BLUs thrown at 1,325 units yielded 146 hits (14% of BLUs); 29 damaged units (2%), 1 crippled unit (0.1%), and 114 dead units (9%). Dead units were “hit” 1–2 times each. It is apparently very hard to hit a truck and not kill it, unlike the much sturdier infantry(?!).

 

Same experiment, this time against BRDM-2s:

Hits	Green	Yellow	Red	Dead
10			8	1
13			11	1
10			10
13			9	2
10			10

 

In total, 1010 BLUs thrown at 1,325 units yielded 56 hits (5% of BLUs); 48 incapacitated units (4%), and 4 dead units (0.3%). One hit incapacitates; two hits kill.

 

Tightening up the dispersal pattern yielded slightly better kill chances in a smaller area:

7 dead, 5 incapacitated in two runs, of which 3 died without the double hit that was needed in the large-area runs. This can only be explained by more tightly-grouped overlapping blast effects combining to overcome the armour.

 

Same experiment, this time against BMP-1s:

Hits	Green	Yellow	Red	Dead
24		22	1
18		12	3
20		12	4
14		14
17		13	2

 

In total, 1010 BLUs thrown at 1,325 units yielded 93 hits (9% of BLUs); 73 crippled units (6%), 10 incapacitated units (0.8%), and no dead units. One hit cripples; two hits incapacitate. An undocumented test I did while setting the whole thing up showed that three hits would generally kill.

 

Tightening up the dispersal pattern increased the odds of getting the requisite hits:

The tighter pattern yielded 11 crippled,  3 dead, and one unit that survived a double hit and became incapacitated.

 

Finally, the same experiment against T-55s:

Hits	Green	Yellow	Red	Dead
12	11
5	5
6	6
6	6
7	6

 

The images become almost irrelevant due to how tiny the slivers of damage are:

In total, 1010 BLUs thrown at 1,325 units yielded 36 hits (4% of BLUs); a grand total of 34 damaged targets (3%) and nothing else. Two hits will put a dent that can be seen in the health bar; one hit can only barely do so.

 

Tightening up the pattern does nothing of value:

A toal of 7 hits creates 6 damaged targets.

 

Unfortunately, as mentioned, the community tools to data mine do not cover the full set of information avialable in the old Lua files. So we don't know the hit points of these units, for instance. So there's no way of telling exactly why we're seeing the sometimes baffling differences between units and dispersal patterns.

 

With this (and the mission templates) in hand, I leave it to others to expand the data set and offer their analyses and opinions on the reasonableness of these results in relation to any real-world performance data. Changing to a Hornet or Viper and a JSOW would be more in line with the title of the thread, but should otherwise make very little difference in terms of the percentages per BLU. At best, you get slightly tighter or wider patterns, but the limits for killing things should remain the same. If not, we have a really huge bug on our hands…

 

 

Incredible report! Awesome stuff Tippis 

[Silver_Dragon]

The problem actually has simulate the BLU-97/B 300 shapers...... x 200 without melted a CPU in a event of infantry / light vehicles / tanks. Ir you like a realistic "soldier kiling", ED need move to make the same situation with a vehicle and build a propper "damage model", no get a 100% kills by a CBU.

[Xavven]

2 hours ago, Tippis said:

Data, eh…?

Well, ok then…

 

-snip-

 

 

BRAVO! This is 100% in line with my anecdotal experiences. Thank you for putting in the time to get this info.

 

2 hours ago, Silver_Dragon said:

The problem actually has simulate the BLU-97/B 300 shapers...... x 200 without melted a CPU in a event of infantry / light vehicles / tanks. Ir you like a realistic "soldier kiling", ED need move to make the same situation with a vehicle and build a propper "damage model", no get a 100% kills by a CBU.

 

This is a good point. I'm not a developer but aren't there some tricks one could use to fudge this a bit? If we can't model all 200 bomblets without a CPU resource issue, then currently if I'm not mistaken the game simulates fewer bomblets and then just increases their blast damage to compensate.

 

But instead of just increasing blast damage, why not create a few more explosions nearby in a random pattern at just above ground level, without rendering them but only for the purpose of causing damage? Each bomblet suddenly "becomes" 5 bomblets for example but only had to be spawned once, rendered once, ballistically calculated once and only 1 explosion animation with particle effects. This would increase the number of "direct hits" instead of relying on the blast damage model of which some IFVs seem pretty well armored against.

[Lurker]

10 minutes ago, Xavven said:

 

BRAVO! This is 100% in line with my anecdotal experiences. Thank you for putting in the time to get this info.

 

 

This is a good point. I'm not a developer but aren't there some tricks one could use to fudge this a bit? If we can't model all 200 bomblets without a CPU resource issue, then currently if I'm not mistaken the game simulates fewer bomblets and then just increases their blast damage to compensate.

 

But instead of just increasing blast damage, why not create a few more explosions nearby in a random pattern at just above ground level, without rendering them but only for the purpose of causing damage? Each bomblet suddenly "becomes" 5 bomblets for example but only had to be spawned once, rendered once, ballistically calculated once and only 1 explosion animation with particle effects. This would increase the number of "direct hits" instead of relying on the blast damage model of which some IFVs seem pretty well armored against.

 

You are assuming that it's the rendering that is the cause of the slowdown, in fact I think that the main issue is the actual ballistic calculations. 

Edited July 12, 2021 by Lurker

[Tippis]

9 minutes ago, Lurker said:

You are assuming that it's the rendering that is the cause of the slowdown, in fact I think that the main issue is the actual ballistic calculations.

Especially since DCS is so CPU- and single-thread-bound as it is.

It wouldn't surprise me in the least if some gains could be had by offloading the calculation to the GPU. It's a whole bunch of independent (and thus parallelisable) known-parameter calculations to find an impact point, followed by area-at-a-distance calculations, preferably with some nearest-object occlusion thrown in. Sounds like a pretty GPU-ish task…

Edited July 12, 2021 by Tippis

[Xavven]

4 minutes ago, Lurker said:

 

You are assuming that it's the rendering that is the cause of the slowdown, in fact I think that the main issue is the actual ballistic calculations. 

 

 

Yes, doesn't this solve that issue as well? Don't calculate the ballistics on 200 bomblets, just do 40 of them, and when each one impacts the ground, that's when you create 4 more ground-level explosions.

[Lurker]

1 minute ago, Xavven said:

 

Yes, doesn't this solve that issue as well? Don't calculate the ballistics on 200 bomblets, just do 40 of them, and when each one impacts the ground, that's when you create 4 more ground-level explosions.

 

Aha, okay I see what you mean now. 

[twistking]

i haven't followed the discussion in detail, so please excuse me, if it has been brought up before:

i think all the discussions will always come back to blast/frag damage at some point, because that's the most obvious shortcoming of the health bar system (apart from subsystem- / mobility kills).

 

as an interim solution i think it could help to have a blast damage multiplier for unarmored targets: utilities, infantry, unprotected cars and trucks as well as SAMs (carrier vehicles might be protected but their important bits should be very fragile).

at the moment area damage is assumingly just a function of base damage and distant to target. my solution would add a fixed multiplier (area damage only) per unit type (infantry x2, unarmored car x1.5, SAM x1.5 ... etc.)

this would obviously not be the perfect solution, but it would help with the odd infantryman / ural-truck surviving a severe carpet bombing or cluster-bomb salvo...

Edited July 12, 2021 by twistking

[nighthawk2174]

57 minutes ago, twistking said:

i haven't followed the discussion in detail, so please excuse me, if it has been brought up before:

i think all the discussions will always come back to blast/frag damage at some point, because that's the most obvious shortcoming of the health bar system (apart from subsystem- / mobility kills).

 

as an interim solution i think it could help to have a blast damage multiplier for unarmored targets: utilities, infantry, unprotected cars and trucks as well as SAMs (carrier vehicles might be protected but their important bits should be very fragile).

at the moment area damage is assumingly just a function of base damage and distant to target. my solution would add a fixed multiplier (area damage only) per unit type (infantry x2, unarmored car x1.5, SAM x1.5 ... etc.)

this would obviously not be the perfect solution, but it would help with the odd infantryman / ural-truck surviving a severe carpet bombing or cluster-bomb salvo...

 

IMO this is a reasonable solution and is one that has been proposed before in the past, can't imagine its too hard to do.

[NineLine]

6 hours ago, nighthawk2174 said:

Yeah something to consider is that some vehicles like the flak should just stop working completly if the pattern is even near the piece as its likely the crew would get shredded if the are not taking cover.  Same with trucks and other light vehicles being close enough to only damage them in game right now would probably result in the driver getting swiss cheesed irl.

Please read what I said previously, I stated we need better suppression simulation. 

4 hours ago, Tippis said:

Data, eh…?

No tracks showing it not working though? As I said I need to see where you think its not working, targeting the appropriate units for that weapon. It could be a specific unit that has an issue, it could be something else.

1 hour ago, twistking said:

i haven't followed the discussion in detail, so please excuse me, if it has been brought up before:

i think all the discussions will always come back to blast/frag damage at some point, because that's the most obvious shortcoming of the health bar system (apart from subsystem- / mobility kills).

 

as an interim solution i think it could help to have a blast damage multiplier for unarmored targets: utilities, infantry, unprotected cars and trucks as well as SAMs (carrier vehicles might be protected but their important bits should be very fragile).

at the moment area damage is assumingly just a function of base damage and distant to target. my solution would add a fixed multiplier (area damage only) per unit type (infantry x2, unarmored car x1.5, SAM x1.5 ... etc.)

this would obviously not be the perfect solution, but it would help with the odd infantryman / ural-truck surviving a severe carpet bombing or cluster-bomb salvo...

 

We already have splash damage, and depending on the vehicle type, it would be affected more with less armor, so watching my video, not a single solider survived, some BMPs did, and the T-55s were not hardly damaged at all. Its not because the infantry took direct hits, its splash damage from the cluster.

 

5 hours ago, Lurker said:

In Nineline's video, apart from one tank which has lost a sliver of "health", those T55s are completely undamaged. To me this indicates that there is definitely room for improvement in the damage modeling of ground units. 

This type of weapon isn't suited for heavily armored units, that said, I didnt see any crew or other personnel surviving that attack. (and again, we have said over and over the DM needs some improvements, but in general the weapons are doing what they are supposed to)

 

7 hours ago, Nealius said:

 

Are there any units that have not received this feature, namely AAA or WW2 assets pack? The 88cm flak continues to provide accurate fire at a high rate of fire with the health bar down around 5% or less. I don't have CA so replicating will take a lot of time (and bombing trials) to get hard evidence. Overall it seems to be inconsistent between units. 

 

Not that I know of, but if you come across any, please feel free to report them. Thanks.

[Tippis]

43 minutes ago, NineLine said:

No tracks showing it not working though?

There is no “not working” to show.

 

It does what it does, and the images (and missions) show this pretty well. The problem is that what it does doesn't match up to what the data says it should be doing. It's a weapon with a supposed 5–20m lethal radius (depending on the source – funnily enough, some of the anti-cluster weapon sites I dug through to find stats on that suggest a lower number than some of the… ehm… explosive-fetishistic sites, let's call them, suggest) against infantry that can't reliably kill or incapacitate infantry that is packed in at 10m distance. And why does it kill trucks more easily than infantry? Why can't this anti-armour weapon (and it is an anti-armour weapon) not reliably punch through one of the least armoured vehicles in the game?

 

43 minutes ago, NineLine said:

As I said I need to see where you think its not working, targeting the appropriate units for that weapon.

What are “appropriate units” for a combined anti-infantry, anti-materiel, anti-armour weapon if not infantry, non-armoured trucks, rifle-calibre-protected scout vehicles, and half-century old lightly armoured IFVs (which, by sheer coincidence, were one of the primary targets the whole system was designed to take out)? I could conceivably agree that an MBT might not be its main intended target, except for the previous discussion about its penetration capabilities (120mm) compared to how much overhead armour that ancient MBT has (16–30mm).

 

Going by listed armour stats for the units in this test, a single BLU should reliably be able to kill pretty much all of them, although a direct hit may be needed in the case of the “heavily” (LOL) armoured T-55. In-game, it can only unreliably kill trucks, with infantry as a distant second.

 

 

Edited July 12, 2021 by Tippis

[nighthawk2174]

  

26 minutes ago, NineLine said:

Please read what I said previously, I stated we need better suppression simulation. 

Sure but suppression is suppression not being shredded into small bits.  Where this would occur is at a range where the crew has a chance of not getting hit and dying.  This is highly dependent on both the frag size/weight and its inital velocity and the density of frag.  Higher weights are more lethal and penetrate more armor with less velocity needed, density of frag increases the chance of being hit lethally, and higher velocity increaes the lethal range and armor pen.  Based on the document Tipis posted for the blu-97 that very lethal area is 40m circle radially from the bomblet.  With it being very dangerous probably out to 80m based on the explosive weight and frag weight.  Currently an infantryman 40m away from a bomblet will not die nor the exposed crew on many of the WWII AAA pieces.  Additionally the frag should have just about the same lethal range against stuff such as trucks and even longer ranges against very sensitive items such as radars and missiles.   Currently they have to be very close to the bomblet the frag should be tearing up targets withing 10m however from Tipis's tests they are not.

26 minutes ago, NineLine said:

We already have splash damage, and depending on the vehicle type, it would be affected more with less armor, so watching my video, not a single solider survived, some BMPs did, and the T-55s were not hardly damaged at all. Its not because the infantry took direct hits, its splash damage from the cluster.

Sure although DCS doesn't simulate every single bomblet right?  So if it did that may not be the case as there would be a higher chance of a direct hit.  Where even if the direct hit doesn't kill we should still have effects such as the crew bailing out if it is penetrated (likely injuring or killing crew) or possible starting a fire.  Even if its not enough for a K-kill causing the crew to bail out of fear is a real thing that has and does happen.  Or on softer targets rendering them inert (radars such as the fan song in particular) as either the crew is dead or the radar is damaged.  This should be occuring a lot farther from the weapon and at much higher damage thresholds due to the nature of the target.

26 minutes ago, NineLine said:

This type of weapon isn't suited for heavily armored units, that said, I didnt see any crew or other personnel surviving that attack. (and again, we have said over and over the DM needs some improvements, but in general the weapons are doing what they are supposed to)

Not really

 

14 minutes ago, Tippis said:

...

+1

[twistking]

@NineLine you misunderstood me. i know that blast damage exists, but right now it seems to be only a function of damage and distance. i propose a function of damage, distance and unit-specific multiplier.

direct damage would still be just base damage substracted from the health pool, while splash damage would factor in base damage, range to target as well as a unit specific multiplier.

 

maybe this is already the case but from my experience, i doubt that it is.