1. Trouble with the game?
    Try the troubleshooter!

    Dismiss Notice
  2. Issues with the game?
    Check the Known Issues list before reporting!

    Dismiss Notice
  3. Before reporting issues or bugs, please check the up-to-date Bug Reporting Thread for the current version.
    0.30 Bug Reporting thread
    Solutions and more information may already be available.

new groundmodel =/= old groundmodel

Discussion in 'Troubleshooting: Bugs, Questions and Support' started by torsion, Jun 11, 2016.

  1. torsion

    torsion
    Expand Collapse

    Joined:
    May 31, 2015
    Messages:
    1,600
    Are the new groundmodels still a work in progress, or are they considered stable for now? The new groundmodel seems to provide significantly less traction than the old one.

    A good example is madden45's "Get Up!" scenario. I updated the scenario to fix the incorrect truck spawning, but you can't even make it off of the top of the slope. Comparing either madden45's video or my video with actual gameplay at this point it's clear that there is much less traction available.

    While the grass groundmodel is where you get stuck in Get Up!, the changes also affect the dirt/gravel groundmodel as well as I found while attempting to update my Shea's Bluff scenarios for v0.5.6.1.
     
  2. randomshortguy

    randomshortguy
    Expand Collapse

    Joined:
    Aug 9, 2013
    Messages:
    1,562
    I don't exactly know if the groundmodels have changed, but they certainly need work. Currently, grass and gravel and dirt all basically act like slippery asphalt. It's really obvious how much they need depth and more traction if you try to do any sort of rally racing ingame.
     
    • Like Like x 2
  3. Dummiesman

    Dummiesman
    Expand Collapse

    Joined:
    Sep 17, 2013
    Messages:
    4,685
    I'm noticing the exact opposite, at least on Utah. Dirt is so powerfully grippy that it makes rallying a pain there.
     
  4. torsion

    torsion
    Expand Collapse

    Joined:
    May 31, 2015
    Messages:
    1,600
    There is no doubt that the groundmodels have changed.

    Could you give a specific example of where you've run into undesirable behavior?

    That's interesting. I've driven around offroad in Utah for at least couple of hours now, mostly in pickup trucks, and I did not notice the tires being too grippy. I've driven in Utah (less seriously) using some official Miramar configurations, the official Rally Sunburst, and the Demon 200BX. The only one that seemed crazy grippy was the Demon 200BX, but cars with sticky tires getting too much traction offroad is a known issue (always happened with drag cars too).
    • Could you briefly describe a route or post/PM some waypoints?
    • If you are interested in seeing the other side of the coin it's very easy. Grab the 'fixed' scenario from this post and compare it with this older video - specifically note what happens as you hit checkpoint #3 around the 1:10 mark!
    I suppose the difference could have more to do with tires... I haven't looked at the files in depth. I didn't see tires in the changelog though and I do see groundmodel. ;-)
     
  5. torsion

    torsion
    Expand Collapse

    Joined:
    May 31, 2015
    Messages:
    1,600
    I'm pretty certain that I've answered my own question. I've reviewed the release notes again for the last couple of releases and it's now clear that it's still in flux. v0.5.3 said "Please do not change / work with this yet. This format will change again drastically in the next updates. We are working towards per level/vehicle/scenario collision/ground model changes" and no more recent update has said anything to contradict this. v0.5.4 mentioned some corrections to that system. Nothing since then, so clearly this is slated for more changes!
     
  6. torsion

    torsion
    Expand Collapse

    Joined:
    May 31, 2015
    Messages:
    1,600
    FWIW I've taken a look comparing the v0.5.1 groundmodel values with the current (v0.5.6) groundmodel values and here's what I came up with...
    • "gravel" has changes which range from tweaks to significantly different (aliases ["rockydirt", "dirt_loose", "dirt_loose_dusty", "dirt_rocky", "dirt_rocky_large", "dirt_sandy"])
    • "grass" has no changes at all to the values (aliases ["grass2", "forest"])
    • "dirt_dusty" had a mistake in the v0.5.1 files and was never set correctly - those settings were applied to RockyDirt (apparently).
    • "asphalt" has changes from tweaks to significant (aliases ["groundmodel_asphalt1", "rock_cliff", "grid", "concrete", "concrete2", "rocks_large"])
    ... that didn't really get me very far.
     
  7. torsion

    torsion
    Expand Collapse

    Joined:
    May 31, 2015
    Messages:
    1,600
    The v0.6.0.0 update restores the old groundmodel behavior!
     
  8. torsion

    torsion
    Expand Collapse

    Joined:
    May 31, 2015
    Messages:
    1,600
    After more testing I've learned that I was wrong. My testing has been in v0.6.1 and while the current groundmodel behavior is trending somewhat towards the old groundmodel, it is definitely not the same as older versions. I believe that the difference may be [less] lateral traction but I'm not certain.

    @synsol - mod support question: do we have any idea what will happen in the near future with the groundmodel? My bug report in this thread was never acknowledged. The devglog/changelog did not acknowledge the TWO sets of obvious changes that have happened recently. I just don't know whether to try and update my mod or whether to wait until groundmodel changes stabilize for a while.

    The obvious answer is to wait, but with no communication I could wait forever. I would feel foolish if I waited for 6 months and the current groundmodel behavior remained stable for that entire time.
     
  9. Goosah

    Goosah
    Expand Collapse
    Global Moderator
    BeamNG Team

    Joined:
    Aug 30, 2013
    Messages:
    790
    I reviewed the revision history for the groundmodels file to look for anything relevant to your observations:

    Feb 09 0.5.2 released

    Mar 5 the groundmodels were converted into a .json file.

    Mar 11 0.5.3 released

    Apr 14 some typos were discovered and fixed, which means that the engine had not been properly reading the files since the conversion to .json. After the fix, groundmodels should be behaving as they did before conversion to .json

    Apr 19 the dirt groundmodel had its fluid properties reduced in strength a bit (only affects ground with depth) in order to make the depth map in utah a bit softer.

    Apr 19 0.5.4 released

    Apr 29 0.5.5 released

    Jun 3 some aliases were removed from some ground models since they had been added as duplicates, making the actual groundmodel choice indeterminate. I think this was an old issue persisting from before conversion to .json. However no aliases were removed entirely, just the duplicate removed leaving the alias assigned to the one most appropriate ground model.

    Jun 4 0.5.6.1 released

    July 22 0.6 released

    So in conclusion, depending on what groundmodel alias's the maps were using, some materials in the map could have shifted to a different groundmodel, but hopefully not one too different (like going from dirt to gravel, etc). As well, if a map was using the "dirt" groundmodel with a depth map, things could have changed a bit.

    As for the future... This is a tricky question. In the short term (like this month) I have been wanting to calibrate the dirt-like ground models a bit, I feel they are too grippy still (cars pulling 0.8g +). So I would expect some changes there, and its hard to say whether that will come from a change in fluid friction/depth behavior or from the basic friction values :/ Which means depending on the map design, it could have different effects.

    In the longer term (beyond this month) there are bigger things we want to do, like some way to achieve a tire specific behavior, so that that drag tires don't get more grip on every groundmodel, for example. As well, the friction model could change significantly in our quest to improve the tire simulation. So we may see more vehicle to vehicle variation in performance on a given groundmodel. Since the end result is a tire X groundmodel, the answer starts to get complicated :p Making a scenario or map that sits right on the ragged edge of traction for a given groundmodel would seem risky, because I think as developers we have to reserve the right to continue tweaking performance.
     
    #9 Goosah, Aug 10, 2016
    Last edited: Aug 10, 2016
    • Like Like x 3
  10. torsion

    torsion
    Expand Collapse

    Joined:
    May 31, 2015
    Messages:
    1,600
    @Goosah thank you very much for taking the time to produce such an in-depth response! As far as the timeline goes, while I appreciate it... I do have a clear, easily reproducible example of the grass behavior on "Cliff" changing from v0.5.6.1 to v0.6.x so I suspect that some (important?) information may be missing from your timeline. (which shows no change there)

    Will you give me just a little more of your time and test a brief scenario in both the v0.6.x and in v0.5.6.1? No need to complete it - simply drive up through checkpoint 4 in both versions of the game. I recommend doing that in v0.6.x first where the scenario works correctly and then again in v0.5.6.1. Doing that will clearly demonstrate different groundmodel behavior. Please use the zip in this post for the scenario which is called "Get Up". I have personally (just now) tested in v0.5.6.1 as well as in v0.6.1.0. Just now I have also diff'ed the "Cliff" folder contents between the current version and v0.5.6.1 - the change doesn't appear in the level contents. So I think it must be the groundmodel...
    EDIT: feel free to try the above for kicks, but as Darren9 points out below what I saw was probably actually changes to the Pickup.

    With that out of the way for the moment, thank you especially for the information on the future. When I initially produced my mini-level/scenario pack I did not expect it to stay viable for a long time. I was surprised when it did and I grew comfortable with the idea of sustaining it. I'm afraid that it probably won't be feasible to do that for a while. :-( It's all in the name of progress though and I look forward to all of the work you mentioned! For now I may focus on some scenarios like what Scepheo has done recently, as well as stuff with a smaller scope.
     
    #10 torsion, Aug 11, 2016
    Last edited: Aug 11, 2016
  11. torsion

    torsion
    Expand Collapse

    Joined:
    May 31, 2015
    Messages:
    1,600
    Argh, that's a good point @Darren9! The "Offroad" tires haven't changed a bit but off course supercharger simulation was added in v0.6.0.0! The engine in "Offroad.pc" pickup's engine was replaced with a different one between v0.5.6.1 and v0.6.0.0. I suspect that this is exactly the reason for what I've seen.

    EDIT: and Goosah's post does add some food for thought. Maybe I should play with depth a little bit on my map and see where it gets me.. obviously the groundmodel will be a moving target for a while so I don't want to go crazy, but if I can tweak the map a bit to keep it fairly playable I'll be thrilled. ;-)
     
    #10 torsion, Aug 11, 2016
    Last edited: Aug 11, 2016
  1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.
    Dismiss Notice