Jump to content

Replay desync experiment - only the F-14 is affected (it's much better in DCS 2.9 but still needs a bit more work)


Katmandu

Recommended Posts

In continuation of this thread that was marked "[DCS ISSUE]" by the devs

 

See my experiment video below. TL:DR is that on replays of an identical simple mission (no AI, middle of the desert, take off and land at the same base - so very short time wise) the F-14B had 0 out of 5 working replays versus 4/4 for the F-16C, 4/4 for the AV-8B Harrier (including two STOLs), and 2/2 for the F/A-18C. The experiment was done over two sessions with no editing whatsoever. 

 

While long and complex missions may corrupt replays for ALL modules, this experiment shows that the extent of the F-14's problems with replays is beyond typical. I hope that this will focus the minds on finding the issues that are particular to the F-14 in this regard. Salut!

 

TIMELINE:
F-14B: 
0:00 1st Takeoff & Landing
3:51 1st Replay 
5:10 2nd Takeoff & Landing
8:59 2nd Replay 
12:14 3rd Takeoff & Landing 
15:56 3rd Replay 
19:36 4th Takeoff & Landing 
23:47 4th Replay 
----------------------------------------
F-16C
27:08 1st Takeoff & Landing
31:07 1st Replay 
34:48 2nd Takeoff & Landing
38:24 2nd replay 
41:58 3rd Takeoff & Landing 
45:27 3rd Replay 
48:45 4th Takeoff & Landing 
52:20 4th Replay 
---------------------------------------------
Harrier
54:51 1st Takeoff & Landing 
58:22 1st Replay
1:00:43 2nd Takeoff & Landing
1:04:38 2nd Replay 
1:06:43 3rd Takeoff & Landing (STOL) 
1:10:49 3rd Replay 
1:13:04 4th Takeoff & Landing (STOL) 
1:16:26 4th Replay 
-------------------------------------------
F/A-18C
1:18:37 1st Takeoff & Landing 
1:22:17 1st Replay 
1:24:32 2nd Takeoff & Landing 
1:27:53 2nd Replay 
------------------------------------------
F-14B
1:30:02 5th Takeoff & Landing 
1:34:08 5th Replay 
1:35:06 6th Replay


Edited by Katmandu
  • Thanks 1
Link to comment
Share on other sites

Part 2 of the experiment, now with multicrew L-39C - as a follow-up to IronMike's comment

Quote

The problem is Jester multicrew, and that tracks in DCS record every one of your inputs and then try to recreate from that. It doesn't account for 2 cockpits though, Jester, etc...

Multicrew L-39C had 4/4 synced replays, and, with another added quick flight here, the F-14 has the total of 0/6 synced replays on the same identical mission.

 

 

TIMELINE

L-39C (two seater)
0:00 1st Takeoff and Landing
05:56 1st Replay
08:39 2nd Takeoff and Landing
13:25 2nd Replay
15:46 3rd Takeoff and Landing
19:30 3rd Replay
21:47 4th Takeoff and Landing
25:36 4th Replay
----------------------------
F-14
29:21 6th Takeoff and Landing
33:02 6th Replay 

 

Link to comment
Share on other sites

3 hours ago, Katmandu said:

Thanks for the reply! I hope you will find the culprit. The F-14 is made for cinematic replays and to me personally there is no bigger bug/issue/problem with the module 🙂

 

 

If you want consistent replays for cinematics, I recommend using server tracks since they just play back physical data instead of the deterministic method used on single player tracks. Ideally the replay system should have an option of which type you want to use. Ask ED for that feature, I can't imagine it would take a lot of time for ED to add since it already exists for servers.


Edited by fat creason
  • Like 3
  • Thanks 2

Systems Engineer & FM Modeler

Heatblur Simulations

Link to comment
Share on other sites

Wow, sounds intriguing! Will be the first thing I test next time there is a free for all (yes, this issue is THE reason why I am not an owner of the F-14 just yet :)). From what I have now googled, apparently one can run a server and a client on the same machine in different instances of DCS . 

 

Link to comment
Share on other sites

19 minutes ago, Katmandu said:

Wow, sounds intriguing! Will be the first thing I test next time there is a free for all (yes, this issue is THE reason why I am not an owner of the F-14 just yet :)). From what I have now googled, apparently one can run a server and a client on the same machine in different instances of DCS . 

 

 

IMO that is a crazy reason to not own the Tomcat, but I'm biased 🙂

Systems Engineer & FM Modeler

Heatblur Simulations

Link to comment
Share on other sites

On 2/16/2021 at 11:24 PM, fat creason said:

 

If you want consistent replays for cinematics, I recommend using server tracks since they just play back physical data instead of the deterministic method used on single player tracks. Ideally the replay system should have an option of which type you want to use. Ask ED for that feature, I can't imagine it would take a lot of time for ED to add since it already exists for servers.

 

 

The F-14 still has some server track issues as well, such as jerky/incorrect control surface animations (often one horizontal stab up and the other down when they should both be up), nozzle position desync, no NWS animation, no tailhook animation, no bounce on cat launch, nose stays up after trapping and doesn't come down until the jet has come to a full stop, RIO model frozen in awkward positions on final approach, etc. The Hornet and other modules do not experience these on server tracks. Something's just weird with the F-14, specifically. 


Edited by Nealius
Link to comment
Share on other sites

17 hours ago, Nealius said:

 

The F-14 still has some server track issues as well, such as jerky/incorrect control surface animations (often one horizontal stab up and the other down when they should both be up), nozzle position desync, no NWS animation, no tailhook animation, no bounce on cat launch, nose stays up after trapping and doesn't come down until the jet has come to a full stop, RIO model frozen in awkward positions on final approach, etc. The Hornet and other modules do not experience these on server tracks. Something's just weird with the F-14, specifically. 

 

 

I have no idea on that. We really don't have any control over replays and all that stuff is just linked to standard animation arguments. There's nothing special we do to make replays "work" per se, that's all handled by the DCS engine.


Edited by fat creason

Systems Engineer & FM Modeler

Heatblur Simulations

Link to comment
Share on other sites

On 2/22/2021 at 8:02 AM, fat creason said:

 

I have no idea on that. We really don't have any control over replays and all that stuff is just linked to standard animation arguments. There's nothing special we do to make replays "work" per se, that's all handled by the DCS engine.

 

 

Did you guys do anything special with the Viggen, or was that all on ED's side? It was broken for a while in replay tracks then miraculously starting replaying perfectly. I wonder what it was that solved the problem, and if it can be extrapolated to the Tomcat.

Link to comment
Share on other sites

  • 3 months later...
On 2/16/2021 at 2:24 PM, fat creason said:

 

If you want consistent replays for cinematics, I recommend using server tracks since they just play back physical data instead of the deterministic method used on single player tracks. Ideally the replay system should have an option of which type you want to use. Ask ED for that feature, I can't imagine it would take a lot of time for ED to add since it already exists for servers.

 

I'm trying out the server replays with the F-14 right now and once again they seem to desync at 100% rate, even for a short take off and land missions.I'm creating a server, flying, quitting, then loading the server replay track from DCS>Tracks>Multiplayer folder. The tracks are automatically named "server-20210611-193418.trk" etc. Other modules' replays work fine again, but the F-14 desyncs very quickly. 

Am I doing something wrong?

Link to comment
Share on other sites

Got it to work in the end 🙂 The server replays do not work if the server and client are sharing the same DCS.exe, so need to install dedicated server and run that in parallel to the client. As can be seen from the videos, the workaround is only partial as significant amount of detail and nuance is missing from the aircraft physics - the landings look off, the gear suspensions are rigid when taxying, the wing vibrations, the gear retract animation and the ailerons/flaperons are juddery, the steering wheel animation does not work at all.

But, the F-14 does have replays working at last! In my tests, 4/4 replays have now synced.

 

 


Edited by Katmandu
  • Like 1
Link to comment
Share on other sites

  • Katmandu changed the title to Replay desync experiment - only the F-14 is affected - SOLVED in DCS 2.9! Huraay:)

It brings me much joy to report that as of version 2.9, the F-14 has fully working replays! Unbelievable but true - HUGE thanks to all concerned! Heatblur, you've got a new customer, this issues was a showstopper for me, but I will certainly join the club now as this gorgeous jet now gorgeous on the outside too, its beauty is no longer "cabin deep" 🙂 

  • Like 1
  • Thanks 1
Link to comment
Share on other sites

14 hours ago, Katmandu said:

It brings me much joy to report that as of version 2.9, the F-14 has fully working replays! Unbelievable but true - HUGE thanks to all concerned! Heatblur, you've got a new customer, this issues was a showstopper for me, but I will certainly join the club now as this gorgeous jet now gorgeous on the outside too, its beauty is no longer "cabin deep" 🙂 

Glad to hear that! Thank you.

  • Like 1
Link to comment
Share on other sites

  • Katmandu changed the title to Replay desync experiment - only the F-14 is affected - it's better in DCS 2.9! Huraay:)

Argh, high hopes crashed somewhat 😞 Getting nothing but desyncs in short missions guns only 1v1 against F-5 AI. I attach couple of tracks- I won in both of them but not according to these:

https://app.box.com/s/rykn2grcpy9sdo184620cb02knn2pm5e

https://app.box.com/s/enwgxxax23idh7yv6ccvef3fwzd8anot

Definitely a very-very good progress, but no cigar just yet. Hopefully this new 2.9 replay system can get better still.

Link to comment
Share on other sites

16 hours ago, Nealius said:

AI decision-making is still "live" in replay tracks. It isn't recorded.

Yes, but the same AI decision making syncs correctly in replays with player planes other than the F-14. 

What is truly weird, is that despite combat being desynced, the F-14 does NOT accumulate positioning errors! In this fight, I chased and shot down AI Su-27 and then landed my F-14. The replay gets the dogfight completely wrong, but the landing is spot on. So yes, the F-14 makes only the AI planes to desync and the situation is looking VERY promising.

Track:

https://app.box.com/s/3d9kcuup6ias9ldbs1q7ps6u95hgz9vp

Video:

 

  • Like 1
Link to comment
Share on other sites

  • Katmandu changed the title to Replay desync experiment - only the F-14 is affected (it's much better in DCS 2.9 but still needs a bit more work)
On 10/21/2023 at 7:03 PM, Katmandu said:

Yes, but the same AI decision making syncs correctly in replays with player planes other than the F-14. 

What is truly weird, is that despite combat being desynced, the F-14 does NOT accumulate positioning errors! In this fight, I chased and shot down AI Su-27 and then landed my F-14. The replay gets the dogfight completely wrong, but the landing is spot on. So yes, the F-14 makes only the AI planes to desync and the situation is looking VERY promising.

Track:

https://app.box.com/s/3d9kcuup6ias9ldbs1q7ps6u95hgz9vp

Video:

 

 

Noted. I'll let the team know for further investigation 🙂

Thank you for your report!

  • Like 1
  • Thanks 1
Link to comment
Share on other sites

  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...