Rigs-of-rods: Tracked vehicles not working correctly.
It appears as if tracked vehicles that use rotators to move slidenodes around the rails, are slightly broken. From what I can see it looks like the submeshes attached to the rotator part are no longer smoothly moving the slidenodes along, causing the tracks to jerk in unrealistic manners.
Vehicles tested with:
http://www.rigsofrods.com/threads/115690-Hitachi-EX1200-5?p=1347026&viewfull=1#post1347026
http://www.rigsofrods.com/threads/112719-volvo-EC140C
And a private tracked vehicle I'm working on.
DarthCainRor
All 14 comments
Can you try disabling Intra-Collision from RoRConfig->Advanced.
Max98
on 12 Aug 2015
That causes them to not work at all, as it needs to collide to move.
DarthCainRor
on 12 Aug 2015
Are you sure that you disabled intra and not inter?
Max98
on 12 Aug 2015
Quite certain. The way tracks are designed they need intra vehicle collisions. It's how they work. Submeshes are attached to the rotator, which sit between slidenodes and push the nodes along the tracks.
DarthCainRor
on 13 Aug 2015
No further comments? :-/ I mean they're definitely not working correctly, and there is no error related to them in the log files. >.<
DarthCainRor
on 17 Aug 2015
the error seems to be at the looping part of the slide rails, some of the nodes jump at the intersection causing shaking
fidoj
on 29 Aug 2015
any word on this? been working on a few things with tracks, and it'd be nice if they actually worked :-/
DarthCainRor
on 1 Jan 2016
I'm still not certain tracked vehicles are operating 100% the same as in .38. I would definitely say they are definitely better now than they were in .4.5, and are semi usuable, but they're still no where near as smooth moving as they used to be.
DarthCainRor
on 13 Jun 2016
@DarthCainRor I completely understand your point of view, but fixing this isn't a matter of reverting some particular changes or looking into some specific area of simulation. From a developer's standpoint, the only way to restore 0.38's behavior is to roll back all changes since that version and never touch the code again. Or, we may tune the simulation to get closer to what it was before, but to do that, we need a lot more detailed description of the problem. Please do some more testing and describe what exactly is disturbing you.
RoR's simulation code is a house of madness. There are calculations that we know are broken (normalization), but correcting them breaks some vehicles. The collision system is "optimized" for performance by actually skipping collision tests. And until quite recently, everything was subtly dependent on the framerate. However, modders of 0.38 era somehow tuned their vehicles to work well under these broken conditions. Devs today have 2 options: either carefully observe what the broken simulation does and maintain compatibility (= supper annoying work, we all went thru "why did I even touch it" experience) or try to straighten up the logic a bit and hope nothing breaks noticeably. @ulteq and @mikadou did some really big improvements to the simulation, both regarding performance and sanity, but as I explained, it probably had side effect.
Conclusion: we're doing what we can to keep vehicles working right, but we can't do it by not touching the code - we need to make changes and continuously test if we didn't break anything. Perhaps our test should include tracked vehicles more - what should we use? What should we concentrate at? We need more input from you.
only-a-ptr
on 13 Jun 2016
I don't believe I ever stated to revert the code and go back or to not even touch the code at all. I was just adding it's still doing the same thing as when I reported this bug, only it's not as bad as it used to be. Tracked vehicles are semi usable, but they still suffer from the jolting and shuttering I previously reported, because the rotator still seems to be skipping some of the contacter nodes along the slide rails. I only didn't notice this sooner because I don't really "play" RoR, except when I'm taking pictures, or working on one of my mods, of which I recently tried working on my tracked loader/dozer again. The two mods I originally linked are still viable options to use
DarthCainRor
on 13 Jun 2016
I only didn't notice this sooner because I don't really "play" RoR
Welcome to the club, I think this is something many developers have in common. We (me at least) are noobs at the game which is why we need pointers by experienced players at what to look at.
Hiradur
on 13 Jun 2016
We (me at least) are noobs at the game which is why we need pointers by experienced players at what to look at.
My case exactly. My initial idea about RoR was "hey, this could be fun to drive around, if it just worked a little better and had some meaningful tools". So I started coding.
I don't believe I ever stated to revert the code and go back or to not even touch the code at all. I was just adding
True, I was assuming too much. Anyway, thanks for the extra clarification.
only-a-ptr
on 14 Jun 2016
the rotator still seems to be skipping some of the contacter nodes along the slide rails
It will be very hard to find the reason for it without knowing which commit introduced the bug.
ulteq
on 10 Sep 2016
Track Test N/b using rotators: http://www.mediafire.com/file/9d3gcii9mgv69rc/trackNBtest.truck
Track test made by lifter using wheels: http://www.mediafire.com/file/9syw6j1w1wglu77/tracktestv15.zip
DarthCainRor
on 9 Oct 2016
Related issues
WillyB1337
路
4Comments
CuriousMike56
路
6Comments
ghost
路
7Comments
Bob-Z
路
11Comments
tritonas00
路
10Comments
Most helpful comment
@DarthCainRor I completely understand your point of view, but fixing this isn't a matter of reverting some particular changes or looking into some specific area of simulation. From a developer's standpoint, the only way to restore 0.38's behavior is to roll back all changes since that version and never touch the code again. Or, we may tune the simulation to get closer to what it was before, but to do that, we need a lot more detailed description of the problem. Please do some more testing and describe what exactly is disturbing you.
RoR's simulation code is a house of madness. There are calculations that we know are broken (normalization), but correcting them breaks some vehicles. The collision system is "optimized" for performance by actually skipping collision tests. And until quite recently, everything was subtly dependent on the framerate. However, modders of 0.38 era somehow tuned their vehicles to work well under these broken conditions. Devs today have 2 options: either carefully observe what the broken simulation does and maintain compatibility (= supper annoying work, we all went thru "why did I even touch it" experience) or try to straighten up the logic a bit and hope nothing breaks noticeably. @ulteq and @mikadou did some really big improvements to the simulation, both regarding performance and sanity, but as I explained, it probably had side effect.
Conclusion: we're doing what we can to keep vehicles working right, but we can't do it by not touching the code - we need to make changes and continuously test if we didn't break anything. Perhaps our test should include tracked vehicles more - what should we use? What should we concentrate at? We need more input from you.