AI ignoring speeds - I am going insane.

[NETSCAPE]

I have copy-pasted a formation of 111's from a mission. Luckily the issue can be reproduced this way. It is not as bad as in my actual mission... as far as number of planes it affects. 

https://www.dropbox.com/s/yvmou1sqmbtbwv8/speed%20test.Mission?dl=0

 

All the way points are set to 310kph. But I've tried 300 all the way to 350... in increments of 5. I tried different wind settings ect... I tried to slightly delay the target link to the first set of way points to see if that helped.

 

The issue isn't collision, because the planes that decide to "fly like a turtle" are never the same. And they never swoop around, they just decrease speed as though it is defaulting to 100kph and not registering my 300-ish commands.

 

In my mission it is always 2-4 planes derping out, slowing down while the rest of the formation cruises at the proper way point speed. In this little test mission it's usually 2 planes derping out for me. Out of like 10 times in this test mission a couple times everything worked as planned. But again I must stress in my actual mission it is far worse. 

 

And just a little note about the actual mission, on top of these 9 x He 111's there are 4 x Bf 110's including the player. Other than that there is one AF with AA active, the target area to be bombed. The mission starts in the air so the players AF has not even been activated at all yet. 

 

I am curious if this works perfectly fine for some of you intel users. I have a sneaking suspicion that slow performance of my amd while loading up everything might be causing this issue...that's just bro science though, I have no idea. 

 

Also before anyone asks "why don't you group and use V formations?" - the mission is an in-air start mission. I don't care how carefully you place 111's in the sky, if you set them to a V formation right at the start they will swoop around like drunk retards and take their sweet time to get in a V formation...which when they achieve that, the formation isn't as tight as I'd like. Hence the individual plane method. I've used it before with 111s and Ju 52s for eye candy in my Stuka missions. This issue is new to me. 

 

So in turn I avoided known formation problems with the bombers and traded it in for some random speed issue...

 

One of those days.

 

Thanks for any feedback on this. Player starts in a tank, hit keybind to view friendly plane and increase time compression and you will quickly see what I mean. You can do a handful of tests in a minute.

[Gambit21]

I'll try and test tomorrow morning if I have time.

[unreasonable]

Just had a look at this - had to   although understand your frustration.  

 

It is funny seeing how it is different planes that drop behind each time you start again.  I tried aligning all the WPs and planes in an exactly straight line abreast in the ME, wondering if it could be a turbulence/wake effect: but I still got one laggard in one test, two in another.  Perhaps get them all at exactly the same height as well?

 

I have a fairly new intel chip - it is not the processor. 

 

I hope you can get this solved: I planned on using the same method to get vaguely realistic looking formations for the current film project.

 

 Is it possible to link planes to a leader in formation without actually giving them a formation type?  In IIRC RoF I managed to construct flights moving together following a leader, so only the leader had a waypoint, but I was able to space out the aircraft how I wanted them. Just checked - I did not remember correctly.

Edited September 8, 2017 by unreasonable

[NETSCAPE]

Yeah it's odd because I and others use the "single plane method" and it usually works fine... kind of glad it didn't work for you so now I know it's not just me. 

 

If all the altitudes were the same it would look goofy though.

 

But tomorrow I will basically make the same thing, 9 planes in that formation with 9 way-points, EXCEPT not copy-paste anything. That could be the issue. I was just so irate earlier I wasn't thinking clearly. Actually I  need to just do the same single plane thing with say the ju88s or ju52s from scratch as well. Because if it happens across the board it could be an issue related to the last game update perhaps. 

[No601_Prangster]

Personally I never use more the four aircraft in each formation. So if I want nine bombers I'll use three groups of three. That way I never have problems with the AI.

[NETSCAPE]

You misunderstand the post or didn't look at the file. There isn't any group. Each plane acts individually. 

[unreasonable]

Try this: delete your Translator Mission Begins.  Add a trigger timer, set for 2 seconds. Target it at each of the waypoints.  Now add Translator Mission Begins, and target it at the trigger timer, (not the waypoints).

 

Doing that they do a little shuffle then settle into a straight(ish) line at the same speed, at least on the three tests starts I tried.

Edited September 8, 2017 by unreasonable

[Gambit21]

Yep - that's the proper way to set it up Netscape.

[Habu]

No need to link the timer to each waypoint, only the first waypoint.

[Gambit21]

Yep - I was assuming the first waypoint for each flight or individual aircraft.

 

I didn't bother to look at the group after I saw Unreasonable's post.

[NETSCAPE]

HELLLLLLLLL YEAH. I will fix er up and finally test the whole mission today. 

[NETSCAPE]

I deleted the mission begin, added new one, target linked to WPs... 3 tests, all failed. I did the same procedure a second time and  1 out of 3 tests it worked proper.

 

So the situation is not solved. Be sure to watch on 8x speed for a good 10 seconds, because at first it might seem like it's working, but it's not. 

 

If you guys have it working FLAWLESSLY, feel free to upload. If it works for me as well maybe I can group it and import it into the actual mission... then it's just a matter of object linking 9,000,000 things... but if it works it works. 

 

edit: I just tried it in my actual mission as well, the issue still persists. 

Edited September 9, 2017 by NETSCAPE

[NETSCAPE]

Ok I literally just placed all new planes (all at same altitude this time), near identical placement/formation, all new WPs and mission begin. And it still didn't work. I tried changing the plane type and that still didn't work.

 

This issue has to be caused by the last update. I am willing to bet the formations utilizing this single plane method don't work as intended in my stuka missions as well. I don't want to waste hours finding those out though. Because I can merely create a new scene in a minute or two and verify that it doesn't work. 

 

It's so simple. There is little room for user error. 

 

One of you guys should try and make a similar thing and see if it works for you or not, then maybe we make an official bug report?

[unreasonable]

Is the modified mission still in your initial post link? Because if it is, you have not put in the trigger timer between the mission start and the waypoints.

 

I have done a modification of your file making each plane and waypoint the same altitude - I had six starts. In one of them one plane fell rapidly out of line, other wise they act together. 

 

I am not offering this as a perfect solution BTW, but perhaps enough to make do - I need to do something like this too for my current project.

 

Speed test U.zip

 

 

[unreasonable]

When I play around a bit more I have convinced myself that it is the altitude stagger that causes most of the variance: the "all planes and WP exactly the same" scenario as per Speed test U works (almost) every time, but if you add in 5 m in altitude to each plane/WP along the line you get your original odd results. 

[SCG_Schneemann]

i chalk it up to fuel/trim/variations in piloting. I tell my guys "we have a straggler" and we either guard him as a "feature" of the mission, or let him fall behind. Engine problems...

[NETSCAPE]

Is the modified mission still in your initial post link? Because if it is, you have not put in the trigger timer between the mission start and the waypoints.

 

I have done a modification of your file making each plane and waypoint the same altitude - I had six starts. In one of them one plane fell rapidly out of line, other wise they act together. 

 

I am not offering this as a perfect solution BTW, but perhaps enough to make do - I need to do something like this too for my current project.

 

Speed test U.zip

 

 At this point not only have i tested timers before the WPs, but I've done everything from 100ms, 300ms, 500ms, 1, 2, 3, 5 seconds ect... oh and not to mention separating all the wp with those timers, ie target to one WP, wait 100ms, next WP, wait 100s, ect ect... just some random idea I tried along with countless others.

 

In my test with all fresh elements in a fresh test mission, like i said, i used all the same altitude to rule that out as the cause. The issue was still present as ever for me.

 

Seems like you're having better luck than me though. I take a look at your test in a minute and see what happens. Like I said man, I've used this method many times before and it has always worked..... but it's definitely not an isolated incident now, which scares me... I'd hate to see all my previous missions look "derpy" 

 

 

i chalk it up to fuel/trim/variations in piloting. I tell my guys "we have a straggler" and we either guard him as a "feature" of the mission, or let him fall behind. Engine problems...

 

ugh it looked so beautiful when they all flew together and bombed the enemy AF... also I have some pretty detailed attack scheme with migs. Not just the same old "attack area mcu". 1 group attacks bombers, 1 group attack you and the other escorts, and they each move onto different specific targets after a target is killed. For example the migs only engage the bombers on the outer most edge of the formation...makes sense. 

 

I might just have to settle for the actual formation command and use 3 Vs and just rework the whole bomber part...which is not going to be fun...

[NETSCAPE]

unreasonable, it looks like your version has about a 50% success rate. Tested 10 times. And no I have not reuploaded/edited my original file I have linked. If I uploaded every test I've done there would be literally like 50. 

[NETSCAPE]

So I was just thinking, why does this work in all my Stuka missions but not here? Then it dawned on me the difference is these planes all are enabled at the start at the wp's happen instantly, or with short time delay as some of you have tried, myself included... and in my stuka missions these types of formations are usually for bombers or ju52s for "eye candy" passing by ect, therefor they are disabled by default, and enabled later in the mission using the methods we all know by now.

 

So, in my ACTUAL MISSION, I made all the he 111s not enabled , Mission begin > 5 second delay > trigger activate all he 111s. I then made a 4s delay from that 5s delay to all the WPs. 

 

This just worked 6 times in a row. 

 

I call this a win. 

[unreasonable]

Well it  - as per your last post - is not working for me, at least not 100% of the time. Actually I get about the same failure rate as with my previous test with the single trigger timer - about 1 in 6 - oddly enough, usually including the first time I load the mission.

 

Possibly I have set it up slightly differently: would you attach the mission that seems to be working reliably for you please?

[NETSCAPE]

no can do. I want this next campaign to be official or else I'd upload the mission. 

 

So interestingly it does sometimes have the issue I have noticed, however the speed at which the derpy guys slow down is far more tolerable.

 

What I'm thinking, and going to try in a bit, is to put the first set of WPs very close to them, so that hopefully the second set will work proper. As it stands in my mission, when the group of second WPs gets hit, their speeds remain the same, no one decides to slow down.

 

And after a test flight of this mission. I forgot that medium priority WPs for bombers makes them weave back and forth. And if the priority is set on high they will fly nice and straight, but the gunners won't shoot. So in short, fancy looking formation is really cool at the start when they bomb the AF, then it kind of doesn't matter I guess since they all swerve around like all hell. They need to shoot, it looks awesome. 

[TP_Jacko]

I hace not had the chance to try your mission. Is it worth to stagger the activation of each bomber. I think if multiple timers are fired at exactly the same time there may be a lag and a message missed somehow. He111 are big aircraft.

 

Also in trying to follow the idea of yours I could not see a coment on why you make individual aircraft not a group.

Chers

[NETSCAPE]

Funny thing, I actually just reworked the he111s and made them into 3 groups of 3, with V formations command. It's horrific. a few of them end up 1km away from the rest of the pack due to weaving around. It's bad.

 

That is exactly why we use the single plane method. Because formations with bombers are just horrible. 

 

And ONTOP of that, 1 of the 3 flight leads went in slow mode.......so back to square one. It doesn't matter what I do. About to just throw in the rag and accept it as is. I can't alter what the AI does obviously so... 

 

I just want it to look as nice as possible. 

[NETSCAPE]

What we need from the Devs is the PRECISE distance, angle and altitude difference relative to the flight leader in each respective formation type and density. That way maybe placing the aircraft as close as possible to the formation might alleviate this issue some. 

 

Currently he111 formations are horrible to work with. It's as though they do some sort of equivalent to over-steer, in the sky... then they over correct, repeat, repeat, repeat, and in no time they have swerved around so much in the sky and lost speed, and at the distance of like 2km from flight lead they decide to gain altitude and fly straight... seems like a great plan mr. pilot... then flight lead is like "okie dokie boys drop yer bombs" and hans the drunk retard is like 5km behind flight lead, but still follows his orders and suddenly he drops his payload on some cows somewhere in Russia.

 

I'm trying all sorts of starting positions for the aircraft. Ironically so far one of the best, meaning quickest ways to get in formation, is by putting the 3 planes of a group in a straight line. This method still usually results in 2-3 aircraft (out of 9) getting WAY behind, not due to the original issue I posted about, obviously, but the issue of using an actual formation command and the problems I have just stated... And when this happens it makes the bombing so unpredictable. And knowing where the bombs hit actually matters because it triggers smoke/fire effects once certain blocks get killed among other things. And I just want it to be awesome when the player watches the replay. 

[Gambit21]

I have a formation of He-111's that bomb an airbase at the beginning of my first mission in the new campaign.

They seem to behave just fine. I'll take a look later and see how much they close up.

Mostly I've been concerned with the effect from the player perspective on the ground.

[NETSCAPE]

Ok I am back to solving this "simple" issue with this mission since the next mission I will be working on is another escort mission...

 

I have revisited my two versions of the mission, one using formation commands the other not. I must rule out using formation command with 111's based on issues already mentioned ^^^. 

 

BUT I realized something upon closer inspection today.

 

After I spot the obvious planes that are decreasing speed... I center the camera on them. I notice they are applying significant back pressure on their control stick. They somehow maintain level flight but are pulling up... This increases in intensity....until they start to deploy flaps... Watching long enough and the 111's that work proper drop their bombs and carry on. The 111's that derp out actually slowly deviate off course of their respective (remember, individual) waypoints. I don't know how I didn't notice this before!

 

This is a serious issue. I think it's time to make a technical/bug report. 

 

There is no way I am going to scrap escort missions. And I am assuming this issue is present in my stuka missions for 111's and maybe ju52s as well. 

 

This never used to happen. I'm sure one of the last few updates is suspect. I hope there is a hotfix. I think I'll continue work and assume that it will be resolved in the future. Because there sure as hell is nothing I can do on my end. 

Edited September 23, 2017 by NETSCAPE

[Gambit21]

There are issues with most AI after the FM update.

Just continue working as you said.,

Right now I'm reporting AI bugs only with aircraft that are supposedly working correctly.

[NETSCAPE]

I wrote up a little report on it.

 

righty-o... on to mission 12 we go. 

[F/JG300_Gruber]

Netscape, I have nothing for you but a profound admiration for your dedication and perseverance at working hard on the editor despite all the issues you are coming across.

 

Hats off !

 

[Gambit21]

It's pretty much what you have to do in order to get anything done.

 

With each of the 5 Ju-52 mission I discovered and reported (and had to work around) at least on major bug.

Currently AI is refusing to join up correctly and hold formation, and once finally in formation (can take many minutes/miles)  the formation then explodes whenever he leader turns.

Thus at the moment with my current campaign build, I have to build my timing around the leader and ignore the rest of the aircraft in the flight, knowing they'll be fixed later.

 

This between editor crashes.

[NETSCAPE]

Speaking of crashes I woke up today only to be greeted by full system lock-ups with no bsod/dump data.

 

Good times.

[Gambit21]

Speaking of crashes I woke up today only to be greeted by full system lock-ups with no bsod/dump data.

 

Good times.

 

Grrr...good times indeed.

Yep all planes are "derping" right now ...making timing/choreographing things a bit challenging.

Air start and then form up? Yikes. Taking off and forming up isn't much better.

[IckyATLAS]

I did a small test with 5 He 111 flying various formations at different waypoints.  There is one leader and the 4 other He 111 are target connected to the leader. The 5 Heinkels start in the air at 3000 meters. The default formation built in the editor is Echelon Right. So I put them more or less in an Echelon right configuration. I have then added 3 waypoints (set also at 3000 mt - very important, and  with 230 km speed) At each waypoints I have made a formation change, example V formation at the first waypoint, with the "Safe" formation parameter.  I was myself flying in a BF 109 150 meters above to see how the He111 behave. All this is pretty simple and can be done in 5 minutes (clock in hand).

 

Running this very simple test mission was without a hitch. The Heinkels started in air and flew the echelon right formation perfectly smooth up to their first waypoint. By using the default formation at start you avoid time lost by the planes to get into that default formation. At the beginning they do move slowly in position as the position of the planes on the map is not exactly the one they will have as the fly in air. And the position depends on the formation (Dense, Safe, Loose, the default is Safe). Then at the first waypoint very smoothly the five Heinkels moved in a V formation without a hitch with the leader in front and two in echelon on each side, and so on at each waypoint.

 

If you want to start a different formation at start then better put the planes in that formation and immediately after the planes appear in game have the formation command act on the leader with the required formation. 

 

The formations are limited to the ones of the system. There is no "Column" or "Line Abreast" or "Box" or "Vertically stacked" etc. type of formations. So if you design on the map a Box formation with a leader connected to all the other planes you will immediately loose the formation and all the planes will try to go in the default Echelon right formation, which will be a mess. 

 

In that case you have to consider each plane individually, and build your own formation and flight plan for each plane. I have not tried that yet. This will be the next test.

[NETSCAPE]

I reported the issue here: https://forum.il2sturmovik.com/topic/31422-bug-report-he-111-h-6-ai-only/

 

I reproduced it multiple times from scratch to check. 

 

Unless their was a hotfix, I assume the issue is still present. 

[IckyATLAS]

I designed a box formation made of 3 rows of 4 Heinkels, total 12 He 111 all flying at 3000 mt. One BF 109 was also flying cover on top of this box.

I set waypoints for each, so you get a box of 12 waypoints. For the position you have to consider that if the first row turns at the same time, you won't have a row after the turn. Your box will not remain a box but be deformed because the outermost planes won't turn as the innermost ones.

So you have to plan the turn so that you get a nice box at the end of the turn. This means adding some waypoints, playing with speeds and this takes some additional time. Anyway the whole boxed formation performed flawlessly keeping speed and position. So yes done this way you can create and fly any type of formation. 

[NETSCAPE]

I'll reopen my test and see if the speed bug is fixed later. 

 

The issue described in the report post is my concern, not turning. And if I use Ju 88s for example, they actually hold the same speed as I have commanded them to. And waypoint placement on turns is good enough to hold a formation, no speed changes are needed. I have a Ju 88 cover mission as well, and it works flawlessly. The bug seems to be only for the H-6. (I'll have to check the H-16 again though I can't remember)

 

With any of your tests make sure to have the planes move far enough to watch them for like 5+ minutes. (use time compression) and don't just fly over them. Set the camera on them to see if they do any of the things described in my bug report thread (ie slowing down, eventual FULL back pressure, then deploying flaps) 

[Gambit21]

It's all aircraft right now like I said - they don't like to make turns and hold formation...it's a mess.

[NETSCAPE]

Just checked the simple test on my bug report thread. Yep still the same issue. No turning involved. 12 planes. 12 waypoints. 

 

@ Gambit: Turning using the formation command an issue? Welp, one more reason to use individual waypoints I'd say! 

[Gambit21]

Yep...but I'm holding out for it being fixed.

At the moment I can't imagine all those waypoints, testi, adjust, test adjust, attack area/attack commands, rtb waypoints...test....test....test....test... etc.

I have more than enough of that already.

I have a handful of hours a day to work on this, so I have to be efficient.

 

It was fine before, it should be fixed soon.

[IckyATLAS]

Yep...but I'm holding out for it being fixed.

At the moment I can't imagine all those waypoints, testi, adjust, test adjust, attack area/attack commands, rtb waypoints...test....test....test....test... etc.

I have more than enough of that already.

I have a handful of hours a day to work on this, so I have to be efficient.

 

It was fine before, it should be fixed soon.

 I did it for a test, but I agree completely with you. It is a very large time consuming work and we should have some additional tools to handle that more efficiently. Maybe some tools for larger formation and different configurations.