New since 82901: Visual state interpolation fully enabled - Feedback thread

[sickboy 13]

For the new MP bugs introduced by the interpolation changes, i've created this ticket: http://dev-heaven.net/issues/23555

Looking for more experiences and actual repro missions.

Just would be really sad if these bugs are overlooked in the coming patch, and no-one wants to install the patch on their server because of it ... :) (Deja vu anyone?)

So this isn't about actually fixing/improving the warping as seen in the official 1.59 patch (http://dev-heaven.net/issues/1915),

but instead to address the issues caused by the interpolation changes in the current beta patches (probably slated for 1.60 patch).

Edited August 18, 2011 by Sickboy

[sickboy 13]

Some more votes and peoples experiences could be helpful http://dev-heaven.net/issues/23555

[Black Russian 10]

Since Bug #1915 is only marked as 40% done I suppose they already planning to work on it. But it is probably not so easy.

Anyway another bug report doesn't hurt.

[MavericK96 0]

Voted up, this is extremely important IMO. Should be a primary focus. :cool:

EDIT: It's really hard to create repro missions because it generally happens when there is a bit of lag on the server in MP, not so much in SP.

Edited August 21, 2011 by MavericK96

[Black Russian 10]

Btw is there any other game with this many [ai] players which doesn't have this issue?

I suppose you probably can only fix this issue by rapidly increasing the polling time for every moving object you see. If the traffic isn't killing standard broadband connections great, but if not ...

Even if the behavior before might have been better they were still warping.

[maddogx 13]

I wonder how much work it would be for BIS to release a full patch without the VSI (visual state interpolation) functionality? Or even better, enable it only in SP for now?

The reason I ask is that the current problems this functionality has in MP seem to me like something that will still need some major development time to sort out, not to mention lots of testing and tweaking by us beta testers; but meanwhile all the other great fixes and major AI improvements that the betas have brought are being "withheld" from the main patch stream until this complex feature is properly implemented. Not to forget that the BAF+PMC V2 signatures issue in MP has been present since 1.59, which was released almost 5 months ago and this problem has only been fixed in the beta patch stream.

Therefore I feel the community might be better served if patch 1.60 was released some time soon, either without VSI or with it disabled in MP. Then we can concentrate on getting this functionality properly implemented, tested and tweaked with more beta patches.

My 0,02€. :)

[gossamersolid 155]

Disabled in MP would probably be the easiest route and that's what I'd like to see as well (unless Suma has a breakthrough with interpolation in MP).

[domokun 515]

I strongly agree as the last official patch is now almost 6 months old and since then BI have done a fantastic job of fixing countless bugs. While I'd give my right arm to have interpolation in MP, I think that its a very ambitious feature which probably requires a lot more development and testing. So until that great day which is the realease of interpolation in MP, how about an stop-gap release ?

[MavericK96 0]

I wonder how much work it would be for BIS to release a full patch without the VSI (visual state interpolation) functionality? Or even better, enable it only in SP for now?

The reason I ask is that the current problems this functionality has in MP seem to me like something that will still need some major development time to sort out, not to mention lots of testing and tweaking by us beta testers; but meanwhile all the other great fixes and major AI improvements that the betas have brought are being "withheld" from the main patch stream until this complex feature is properly implemented. Not to forget that the BAF+PMC V2 signatures issue in MP has been present since 1.59, which was released almost 5 months ago and this problem has only been fixed in the beta patch stream.

Therefore I feel the community might be better served if patch 1.60 was released some time soon, either without VSI or with it disabled in MP. Then we can concentrate on getting this functionality properly implemented, tested and tweaked with more beta patches.

My 0,02€. :)

Agreed 100%. There are still a couple bugs introduced in the latest beta to sort out, but after that I'd like to see a final patch without MP VSI (unless that feature is REALLY close to being done, but I feel like it would require a lot more beta testing).

[alef 0]

For the new MP bugs introduced by the interpolation changes, i've created this ticket: http://dev-heaven.net/issues/23555

Looking for more experiences and actual repro missions.

The video shows shifts also at near distance, and on loopback network interface, that means, on the same PC at the same time.

In this data I've recorded, thanks to perFrame hanling in @CBA, of a single walk and stop. Recording data is easy, difficult is read it.

What I saw is that some animations update from remote resets the position of unit back to when the same animation was triggered before on the local. Now, most videos shows AI walking and jumping ahead, while here is jumping "back".

I hope someone else can see other patterns that explain this strange behaviour.

[MavericK96 0]

To me it just looks like the unit position updates just need to occur faster. I imagine there would be a bandwidth/CPU usage penalty for this, but I'm not sure if interpolation is going to work when units change direction dramatically (as the AI seem to like to do often). In any case you are going to get some sort of jumping from one direction to another.

[.kju 3244]

The basic question of mine is why it worked a lot better in OFP.

And became worse since arma 1 and more and more with future engine versions.

(my guess is still bad/flawed optimizations for ofp elite)

[sickboy 13]

The basic question of mine is why it worked a lot better in OFP.

And became worse since arma 1 and more and more with future engine versions.

(my guess is still bad/flawed optimizations for ofp elite)

Little regard for MP?

At least that impression exists.

[maddogx 13]

In his blog article, Suma talked about reducing the simulation frequency to improve performance, and I guess that's exactly what they've been doing since OFP. As the engine has been lacking visual interpolation until now, the visual warping problem has been becoming more and more pronounced every time the sim-step time increases with a new engine iteration.

Anyway, since their interpolation tech is completely new, It's not surprising that there are still kinks to work out in certain situations - MP just being one very large and common situation with specific quirks :). It's not going to be easy to get the interpolation right in MP, since there are a lot of variables to consider and a lot of testing will need to be done. Hence my suggestion.

[sickboy 13]

That doesn't really negate the point that it seems that MP support is low on their TODO list :)

As kju points out, ever since A1 the situation has worsened, and it was already bad at distance in OFP.

And now with this interpolation (that supposedly shouldn't affect MP at all according to Suma), the situation has worsened yet again.

Kinda 1 step forward, 2 steps back idea :P

Just hoping this will not be integrated as is into upcoming official patch, or MP has yet again become worse :)

And trying to make BIS aware of it is kinda important too, especially referring MP issues.

Edited August 23, 2011 by Sickboy

[maddogx 13]

That doesn't really negate the point that it seems that MP support is low on their TODO list :)

It wasn't supposed to negate your point; in fact I agree with it. It's sad but probably true. :)

As kju points out, ever since A1 the situation has worsened, and it was already bad at distance in OFP.

And now with this interpolation (that supposedly shouldn't affect MP at all according to Suma), the situation has worsened yet again.

That's true, but at least Suma has said that he will be looking to make it work in MP aswell. It's not like the functionality is considered complete (see the DevHeaven ticket - "40% done"), so I don't see much ground for complaining about it not fully working, at least not just yet.

Kinda 1 step forward, 2 steps back idea :P

Just hoping this will not be integrated as is into upcoming official patch, or MP has yet again become worse :)

This would indeed be very bad, and it's one of the things I believe they have to avoid at costs. That was also part of the reasoning behind my suggestion: release a full patch without the buggy interpolation in MP, then take the time to get it working properly for the next patch. :)

Edited August 23, 2011 by MadDogX

[alef 0]

It's not going to be easy to get the interpolation right in MP, since there are a lot of variables to consider and a lot of testing will need to be done.

For the developer SP debugging is easy compared to MP. In MP you will have 2 or more programs to follow with more debuggers, but I hope modern IDE should allow following more than one instance at the same time, and probably even remotely.

I think this feature would help people develop missions and testing in MP better.

[.kju 3244]

It wasn't meant to be negative - I am merely wondering about the causes.

The point of MDX might be one key aspect of the degradation.

That said one has to differ between SP and MP too - my point was more about MP.

However lets not confuse us - the current situation is that it does not work even in SP.

Suma also confirmed that they have not touched MP at all so far.

In any case my vote also goes to move the interpolation to 1.61 and get 1.60 out of

the door ASAP without it.

[Rough Knight 9]

I am not sure it is as easy as just punching out the patch. My fear is that these interpolation changes have to be sorted before the patch could be issued because some issues with animations etc didn't exist before these changes where made. Also many other ai related bugs have been resolved in-between these fixes further complicating matters.

I am not very savvey like many of you guys when it comes to programing, I hope I am wrong.

[.kju 3244]

Of course only BI can judge if it can be done with a reasonable effort.

What we are saying is that the current situation would be bad to become a full patch

and at the same time it would be bad to postpone 1.60 considerably only because of this.

[.kju 3244]

Suma

I understand this issue is important for you, but in the current form I am afraid it is not much workable, and I definitely feel lost when reading it. There is no clear and simple repro (and perhaps it is impossible to create one), and analyzing logs or theories is not something I am willing to do at this stage.

I am currently working on improving the situation in MP, and I think it has no sense to give any MP feedback until the version is a bit more stable and completed.

The only exception would be if someone can create a really simple and short repro with only a few units, which would clearly show some particular problem - a report like this would be probably worth checking.

http://dev-heaven.net/issues/23555#note-16

+

Suma

Note: this time there is a server-side fix, which should partially reduce unit lagging in MP (sometimes the same position was broadcasted twice erroneously). The fix by itself is not very significant, but it should reduce the artifacts caused by the interpolation.

http://forums.bistudio.com/showpost.php?p=2014129&postcount=1

---

So please test the latest beta patch and provide simple demo missions.

[alef 0]

So please test the latest beta patch and provide simple demo missions.

Done.

[.kju 3244]

It seems that people and Suma think it is a MP problem - it is NOT.

It is also true for pure SP play.

[suma 8]

It seems that people and Suma think it is a MP problem - it is NOT.

It is also true for pure SP play.

If you have some situation where you see the problems in SP, by all means, provide a simple an consise repro showing this. I have not seen any SP problems for some time and I consider the SP interpolation mostly completed.

[.kju 3244]

Alright.

I wasn't able to repro it for SP yet, but here is a simple and good demo for MP:

http://dev-heaven.net/issues/23555#note-19