„watch track“ shows totally made up content

[Hiob]

That would not explain a total different set of actions by the AI.

5 minutes ago, jwflowersii said:

From what I can tell, the replay system is tied to every frame drawn.  It stores input based on frames per second, so if you have even the slightest stutter in frames, it'll miss an input and this is why replays are not accurate of players.  Server replay is accurate 100% of the time however.

 

[Fri13]

1 hour ago, Ironhand said:

Tacview basically records the mission flown. It doesn't draw data directly from the TRK file. You have to run the track and Tacview will show what happens in the replay. If the replay is off target, Tacview will be as well.

 

Interesting part is that Tacview can be wrong in first place too even when it records realtime gameplay.

 

Has happened to me.

 

In actual gameplay (realtime) I made a gun strafe at BTR. I shot a 25 round burst on target and all hit it (as gun is like sniper from 7000 ft).

 

In a Tacview replay I shot target 2 meters above it, missing it by 5-7 meters.

 

In a DCS track replay I shot 50 meters further from the target.

 

Tacview version from the replay was about 100 meters off from the target.

 

That is why I don't really care to provide track files for bug reports because they are usually invalid method to show the problem. Mostly proper method is to record the screen and show it from there that what happens.

 

This, unless it is something very specific that replay could do. But even then back it up with screen video....

 

 

[snipy]

Not sure if useful, or n of 1 example, but I have noticed if I fast forward a track, there is just no telling what I will get out of it.  But if I never alter the playback speed, it seems to play back what happened. 

 

That can be tedious, when what I want is at the end of the track, but it's an option if trying to use a track. 

 

 

[Exorcet]

In my experience the F-14 is the ultimate track destroyer. I don't know what it is about that module, but F-14 tracks can regularly break in seconds. Other modules tend to do better and can have no track issues for around half an hour, but as a whole DCS tracks have some issues.

 

ED tried to address this in a few patches ago but apparently didn't succeed so I assume that the problem are really deeply seeded and may not go away until the track system is redone from the ground up completely.

[Eugel]

For me, 100% of both A-10C II and Huey replays derail after about 10 minutes (since 2.7 - before that it only happened occasionally).

To be honest, I hadn´t found out how to fast forward in replays, so that´s not causing it for me.

 

Maybe it´s worse in VR because of all the head movement ...

[26-J39]

12 hours ago, jwflowersii said:

From what I can tell, the replay system is tied to every frame drawn.  It stores input based on frames per second, so if you have even the slightest stutter in frames, it'll miss an input and this is why replays are not accurate of players.  Server replay is accurate 100% of the time however.

I guess that's an option for SP people who need a reliable option.    Just start a private server and jump in as a sole player, easy enough.

[Hiob]

8 hours ago, Exorcet said:

In my experience the F-14 is the ultimate track destroyer. I don't know what it is about that module, but F-14 tracks can regularly break in seconds.

 

My recent experiences, which I've been reporting here were all with the F-14 - so that may be true.

The only other Track I did in recent times was a short flight with a F-16 for my 4K-Benchmark, which worked flawlessly (but was in no way complex).