Unreliable track files, please make them use able again.

[Maverick_NL]

Hi ED,

 

I know it is Low priority, but i would like to see that the replay function is getting updated, at the moment its not possible to record from the track files they are 80% of the time broken.

 

It would be nice to see a new replay function in DCS, the one we use now is really useless, especially when we want to record from cockpitview (example F-14) that track always breaks.

 

The server tracks are always good but don't give us the opportunity to record for what we want to record.

Also for bug tracking there are much 3rd parties and ED requesting a track file for the bugs we find. also if the track is not working, we cant show our problems of the bugs we are walking through.

 

It would be nice to see the replay function getting updated / fixed. as i stated i know it is low priority for ED. But a Must for the end user.

 

i know there a lot of threads about this, but please ED i hope you can look into it and try to fix / update the replay viewer..

[Nealius]

They've gotten particularly bad since 2.7. Modules that used to replay great in ideal, controlled conditions in 2.5.6 now break between the 18~25-minute mark, in a consistent manner indicating that the player's inputs either failed to replay after that timestamp, or failed to record after that timestamp; almost always resulting in a left or right descending turn into terrain, with no stick or surface control movement indicated between time of desync and crash.  It's like the .trk stops feeding the sim input data after a specific period of time. 

[Maverick_NL]

4 hours ago, Nealius said:

They've gotten particularly bad since 2.7. Modules that used to replay great in ideal, controlled conditions in 2.5.6 now break between the 18~25-minute mark, in a consistent manner indicating that the player's inputs either failed to replay after that timestamp, or failed to record after that timestamp; almost always resulting in a left or right descending turn into terrain, with no stick or surface control movement indicated between time of desync and crash.  It's like the .trk stops feeding the sim input data after a specific period of time. 

yeah it really need to have a rework. hope ED will make time for this. otherwise its impossible to report bugs / problems and make videos.

[Breakshot]

Tracks are the most unreliable thing in DCS.

 

Yet when someone needs to report a bug, the first question ED asks is to present "tracks".

 

Go figure!

[Jack1nthecrack]

Track?

 

[Flappie]

2 hours ago, Breakshot said:

Tracks are the most unreliable thing in DCS.

 

Yet when someone needs to report a bug, the first question ED asks is to present "tracks".

 

Go figure!

 

Short tracks are reliable.

Long tracks are not.

[RADARCH58]

Hi ED

 

Creating video content from replays of track files is one of the main attractions of DCS for me and many users. Please take note that as a DCS fan I am now very disappointed that most of my aircraft/flight track files are totally useless. It is no longer possible to get any useful footage from replays of track files.

 

This is particularly disappointing as I specifically bought the HIND with the purpose of creating video content with this module. This is simply no longer possible on DCS and I shall not be purchasing any more modules until ED resolves this issue. I know of other users who feel the same way. signed SPANKY - a very unhappy DCS user!

Even my 2 minute tracks with the HIND are totally useless! This applies to the MIRAGE as well!!!

[Jack1nthecrack]

6 hours ago, Flappie said:

 

Short tracks are reliable.

Long tracks are not.

This doesn't help lol, all tracks should be reliable. The system obviously needs some work if it only half works!

[Nealius]

20 minutes is short considering it often takes 20 minutes just to start up and taxi to the runway in a sim such as DCS...

Edited July 1, 2021 by Nealius

[Convexrook]

Somehow every other game that has replays work fine. 

What needs to be done is a capture instead of a track overlay on gameplay.

That short answer "Short tracks are reliable. Long tracks are not." could be a bit better tbh.

 

Something along the lines of " with new multi-threading and vulkan in the works this might pave the way on how track files are created. "

[Maverick_NL]

14 hours ago, Flappie said:

 

Short tracks are reliable.

Long tracks are not.

Dont agree about the short tracks.

Also short tracks are unreliable, we tried a few times to debrief our Aerobatics display in the harrier, and its directly out of sync after take off (There was no collision after take off or anything else).
 

Also when i made an instant action mission to record a short combat video, when the plane got hit the track is fooked.

So it will be really important to see this reworked or getting fixed. for video creators / Bug reporting. its useless at the moment.

Edited July 1, 2021 by Maverick_NL

[Barra1]

Its completely useless for analysing your carrier landings. You trap in the mission and then in the replay the carrier is in a different place and your plane lands in the water. Replays have been broken for as long as I've been playing DCS World (10 years I reckon).

[Maverick_NL]

please ED i know its low priority but please look into this.

[dorianR666]

On 6/30/2021 at 8:55 PM, Flappie said:

Short tracks are reliable.

Ive had a track fail within 2 minutes.

It didnt record my keyboard press-and-hold of rudder input on taxiway, resulting in me going full afterburner 90 degrees perpendicular to runway during replays. It seemingly forgot to insert that input into the trackfile.

 

There are software bugs in the trackfile functionality that simply are less probable to manifest in short trackfiles. But they fail too sometimes, rarely, but they do.

Edited July 17, 2021 by dorianR666

[oldcrusty]

On 6/30/2021 at 11:55 AM, Flappie said:

 

Short tracks are reliable.

Long tracks are not.

 

Start in the air and have a 1 v1 turning fight. It doesn't have to be a gun fight, even in BVR. once the missiles start flying, defensive maneuvers will throw the track off within seconds.

[Flappie]

The more AI units you have in your mission, the faster the track will be different from what you actually experience in the mission.

 

I use tracks everyday to fill bug reports and they work just great for me, as long as I keep them short (<10 minutes). ED crew and testers are using tracks for bug chasing (tracks are often effective for this task, as long as they are short), while players often use them to record videos (tracks are often very ineffective for this task, mostly because missions last way more than 10 minutes).

 

@dorianR666Your example is rare. You should report this as a bug with your track attached. This seems like an easier fix than the AI units doing their own thing.

 

@Gripes323I believe this happens because AI units "improvise" every time you play the track. That's why tracks as we know them are not effective to record gameplay videos, unless you film a "simple" CAS attack with no AI CAP.

Edited July 17, 2021 by Flappie

[oldcrusty]

1 hour ago, Flappie said:

The more AI units you have in your mission, the faster the track will be different from what you actually experience in the mission.

 

I use tracks everyday to fill bug reports and they work just great for me, as long as I keep them short (<10 minutes). ED crew and testers are using tracks for bug chasing (tracks are often effective for this task, as long as they are short), while players often use them to record videos (tracks are often very ineffective for this task, mostly because missions last way more than 10 minutes).

 

@dorianR666Your example is rare. You should report this as a bug with your track attached. This seems like an easier fix than the AI units doing their own thing.

 

@Gripes323I believe this happens because AI units "improvise" every time you play the track. That's why tracks as we know them are not effective to record gameplay videos, unless you film a "simple" CAS attack with no AI CAP.

 

IA improvising is probably the cause and that's what I was thinking too. What drives me nuts that 3 or 4 times I was able to record an event involving 8 v 8 AI + me, shooting a wall of Amraams with everyone turning and burning and the replay showed no problems. It was in one of the older builds but I couldn't figure out what made a difference.  It happened consistently, few times in a row when recording complex missions.  One thing I remember is that I used a specific area on the map over and over for weeks, without wiping out FXO and meta2 folders but... I don't think that had anything to do with it.  I'm only talking single play here, perhaps w/o AI in MP things are more settled?

[Madfish]

On 7/17/2021 at 9:40 PM, Flappie said:

The more AI units you have in your mission, the faster the track will be different from what you actually experience in the mission.

 

I use tracks everyday to fill bug reports and they work just great for me, as long as I keep them short (<10 minutes). ED crew and testers are using tracks for bug chasing (tracks are often effective for this task, as long as they are short), while players often use them to record videos (tracks are often very ineffective for this task, mostly because missions last way more than 10 minutes).

 

@dorianR666Your example is rare. You should report this as a bug with your track attached. This seems like an easier fix than the AI units doing their own thing.

 

@Gripes323I believe this happens because AI units "improvise" every time you play the track. That's why tracks as we know them are not effective to record gameplay videos, unless you film a "simple" CAS attack with no AI CAP.

 

As your signature is "A track or it didn't happen." and you seem to be an expert on the matter, two questions:

1. Is the short trackfiles <10 minutes a confirmed amount?
   • E.g. are tracks reliable 100% under 10 minutes, no matter if AI or not?
     • What if there's no AI? Still 10 minutes or longer? If so what amount is confirmed as reliable?
        
2. Can you confirm track files are historically compatible without issues across versions, including e.g. changes to missiles or FM, or how would one use a track to prove something happened?
   • If they are not compatible I would suggest adding that to your signature to avoid confusion e.g. "A track from no older than the most recent version"
   • Also what would you suggest to report bugs with if they are not reliable across versions when changes happen - as many bugs are not being looked at right away?

[Flappie]

I'm not a track expert, no. I'm a bug reporting track maker, if you may say so.

 

My sig means there's no better proof of a bug to me than a track, as long as the replay remains accurate. But it's not always easy to show a bug with a track, especially in MP.

 

1. I have no idea. I never tested this. If you want to, it might help identifying some of the track reliability issues. My bug reporting tracks and those of my fellow colleagues are working if they are short and not too old, that's all I know.

 

2. That is precisely the point: to me, tracks are more of a debug tool than a souvenir. You post a bug with a track, a tester replicates the issue, a dev replicates the issue then fixes the bug. The first thing a dev probably does after commiting a fix is using your track to ensure the bug is fixed. But commiting = new version. You see where I'm going? Tracks are often broken after a few versions because the facts (what happened when you record them) are based on an obsolete version. Example : "AI F4 missile range is too long", "Ok, we fixed it". Then your 1vs1 track doesn't show what had happened anymore.

 

I hope I'm clear enough because I prefer spending my time making tracks to hunt down bugs rather than talking about it.

Edited July 19, 2021 by Flappie

[Andurula]

in my experience the track is more likely to "fail" the busier the computer is.  A fast CPU flying high above the desert and your computer has no problem keeping an accurate record of your control inputs.  Flying low over a town with lots of ground objects to render and a less-than-speedy CPU is always going to fail rather quickly.  Additionally, the more AI to process the more load on the CPU thus more track recording deviance.

 

So I think the issue is the speed and load on the CPU rather than time per se.  Over time, accumulating mis-recorded control inputs will result in ever more deviant player aircraft behavior so time does appear to be a factor but it really isn't.

 

In my case, I have a less-than-fastest CPU and I always start cold and dark surrounded by lots of ground objects so my tracks always become hopelessly corrupted before I can get lined up properly with the runway.  Any flight or bug I might encounter after that point will never be recorded so asking me for "tracks" for any bug I notice is just a waste of everyone's time. 

 

While I think track files in their current state are a terrible way to "prove" bugs by the general DCS community, I also can't suggest a better way.  I will say it is annoying to be repeatedly told to submit track files for bug reports so I just don't report bugs.

 

 

 

 

[Maverick_NL]

Yeah same, i don't report bugs as the tracks are unreliable.
 

I don't think its an CPU issue / load issue as i have an reasonable CPU / GPU and the track still got of rails. it looks like when you hit an little bump when taxi the plane is going of rails. maybe some sort of detection zone.

Example: When i do a carrier landing with the f14, i fly a mission do a runway take off and landing on the carrier. what happens the track is already off rails when taxiing out to the runway and get stuck into the grass. and its not the f14 multiple planes i fly will end up off rails or in the grass.  the f14 is just an example.

Its really annoying, i think its worth that ED looks into it. as it is really important to report bugs/problems. i mean how would you ever resolve bugs, when we cant show it.
 

Maybe its time to rework the replay function instead with an overlay/ game capture funtion or something.

I hope you guys gonna look into this, at least report it to the Dev team.

Edited July 28, 2021 by Maverick_NL

[Flappie]

13 hours ago, Maverick_NL said:

Example: When i do a carrier landing with the f14, i fly a mission do a runway take off and landing on the carrier. what happens the track is already off rails when taxiing out to the runway and get stuck into the grass. and its not the f14 multiple planes i fly will end up off rails or in the grass.  the f14 is just an example.

 

Someone described a similar issue with Fw-190 aircraft when using left / right wheel brakes earlier this week. Can you reproduce the issue when you're NOT using left / right wheel brakes? (I mean, if you only use the "Wheel Brakes" command that controls both wheel brakes)

[Maverick_NL]

8 hours ago, Flappie said:

 

Someone described a similar issue with Fw-190 aircraft when using left / right wheel brakes earlier this week. Can you reproduce the issue when you're NOT using left / right wheel brakes? (I mean, if you only use the "Wheel Brakes" command that controls both wheel brakes)

 

erhm, if you need to taxi you need rudder and wheel brakes tho? how would i reproduce this if the tracks are getting bugged to?

But i never use separated wheel brakes. i only use the normal wheel brakes (Key W on keyboard, is binded on my stick). I don't think the wheel brakes is an issue.

So i find it really useless to reproduce it. let the devs look into it.

Edited July 29, 2021 by Maverick_NL

[Flappie]

1 hour ago, Maverick_NL said:

But i never use separated wheel brakes. i only use the normal wheel brakes (Key W on keyboard, is binded on my stick). I don't think the wheel brakes is an issue.

So i find it really useless to reproduce it. let the devs look into it.

 

I thought you were saying it happened specifically when using toe brakes. I don't have toe brakes and I cannot reproduce your issue.

[BIGNEWY]

Track replays, replay inputs only.

 

Many things can affect track replays, AI can make different choices, wake turbulence can have an effect, rudder / take off assist can also create problems. 

 

I work with track replays all day everyday, short tracks for bug reporting work well usually, longer tracks and multiplayer tracks, especially heavily scripted ones are less reliable.

 

If you want us to look at a specific replay problem we will need to see a track replay showing a problem, and as mentioned it may reproduce or it may not. 

 

We want to update the track replay system, but it will be a huge task and we are very busy currently with other projects, its something for the future for now. 

 

thanks