1. Do not share user accounts! Any account that is shared by another person will be blocked and closed. This means: we will close not only the account that is shared, but also the main account of the user who uses another person's account. We have the ability to detect account sharing, so please do not try to cheat the system. This action will take place on 04/18/2023. Read all forum rules.
    Dismiss Notice
  2. For downloading SimTools plugins you need a Download Package. Get it with virtual coins that you receive for forum activity or Buy Download Package - We have a zero Spam tolerance so read our forum rules first.

    Buy Now a Download Plan!
  3. Do not try to cheat our system and do not post an unnecessary amount of useless posts only to earn credits here. We have a zero spam tolerance policy and this will cause a ban of your user account. Otherwise we wish you a pleasant stay here! Read the forum rules
  4. We have a few rules which you need to read and accept before posting anything here! Following these rules will keep the forum clean and your stay pleasant. Do not follow these rules can lead to permanent exclusion from this website: Read the forum rules.
    Are you a company? Read our company rules

Playing with Filters....

Discussion in 'DIY Motion Simulator Building Q&A / FAQ' started by Archie, Mar 6, 2015.

  1. Archie

    Archie Eternal tinkerer

    Joined:
    Dec 31, 2014
    Messages:
    1,081
    Location:
    Wollongong, NSW, AU
    Balance:
    3,796Coins
    Ratings:
    +1,379 / 4 / -0
    My Motion Simulator:
    2DOF, DC motor, JRK
    Hi Everyone,

    I've setup all my games and the motion is nice. I'm happy with most of the settings now.
    Of course, we always tinker, and I've started playing with the Axis filters...

    I'm interested in the smoothing and washout filters, but I need to understand what they are doing, using my own SimTools settings as an example.

    I find it's really adding to the experience, especially on a full frame sim where (and this blew my mind) I get the sensation of "hard" or "soft" suspension from the front mounts. For example, when bombing down Nords in the Formula A in pCars, the sensation of "springy bounce" at my feet is quite amazing.

    Here is what the doco by @eaorobbie says:

    Smoothing – Will smooth the value, if a value seems jerky or not flowing this can be increased to add a smoothing effect by creating a mean value of the values coming in.

    Washout – Is handy when we require a value to wash back to 0. i.e. like for a yaw axis that doesn’t actually turn a full 360 degrees. So to give the feeling to the rider that he/she is still actually turning.

    If I have my SWAY set to 60% and smoothing is set to 30%, what is the actual effect of this?

    If I have my ROLL set to 60% and Washout is set to 30%, what is the actual effect of this?

    When I have 30% smoothing set on Sway axis at 60%, the cars kind of "settles" back to level after a sharp corner, instead of just 'flicking' back to centre / level.

    When I have 30% washout on Roll axis at 60%, the front slowly settles back to level after hitting a bump.

    Last night I tried these exact settings in pCars and I really like the effect it gives, but I still don't understand the "science" behind it, so I'm just entering random number and hoping for the best.

    Any elaboration of these settings very much appreciated.

    Thanks,

    Archie.
    • Useful Useful x 2
  2. noorbeast

    noorbeast VR Tassie Devil Staff Member Moderator Race Director

    Joined:
    Jul 13, 2014
    Messages:
    21,051
    Occupation:
    Innovative tech specialist for NGOs
    Location:
    St Helens, Tasmania, Australia
    Balance:
    148,004Coins
    Ratings:
    +10,884 / 54 / -2
    My Motion Simulator:
    3DOF, DC motor, JRK
    @eaorobbie pretty much nailed the explanation but perhaps it would be clearer if you think of flight sim rather than racing, as the examples are more apparent.

    Smoothing allows for the more flowing feel of flight than the bumps of racing, as it averages the motion.

    Washout is used to mitigate the real physical limitation of a motion sim by exploiting flaws in our sensory systems. So, a 2DOF sim can't execute a barrel roll or loop, as that is outside its physical capacity. But say in a loop you can use washout so you initially feel the plane pitch up, but as the loop continues over the top the sim slowly returns to level below your perception of feeling it do so, then makes you feel the pitch down as the plane starts to head back down toward the ground. Your perception is that you are doing a full loop with washout letting you exploit your perceptual weakness to make it fell real, when in fact your sim is bound to 2DOF.
    • Informative Informative x 2
  3. Archie

    Archie Eternal tinkerer

    Joined:
    Dec 31, 2014
    Messages:
    1,081
    Location:
    Wollongong, NSW, AU
    Balance:
    3,796Coins
    Ratings:
    +1,379 / 4 / -0
    My Motion Simulator:
    2DOF, DC motor, JRK
    Thanks @noorbeast - I guess my confusion is around the "math"....

    So when you have two axis at 60% with 30% smoothing, the actual output would be 42%, (30% of 60)
    Or, does it ensure that the final 18% of the movement is "smoother" at the extremes.
  4. noorbeast

    noorbeast VR Tassie Devil Staff Member Moderator Race Director

    Joined:
    Jul 13, 2014
    Messages:
    21,051
    Occupation:
    Innovative tech specialist for NGOs
    Location:
    St Helens, Tasmania, Australia
    Balance:
    148,004Coins
    Ratings:
    +10,884 / 54 / -2
    My Motion Simulator:
    3DOF, DC motor, JRK
    I think the actual filter documentation is pretty limited: http://www.xsimulator.net/simtools-game-engine/#Filters

    SimTools mixes effects with some adding and some contracting depending on the axis. If I understand it correctly increasing the filter % will slow the effect for an axis but SimTools will mix the outcome. Think pitch down being somewhat cancelled by surge up but pitch up being added to by surge up.

    Filtering as a means to emulate what we expect by minimising the limitations of our motion rigs and by exploiting our perceptual weaknesses is in my view as much art as it is science, and not just simple math, as can been seen when I started asking questions about my own assumptions as to how SimTools actually implements filters: http://www.xsimulator.net/community/threads/dx-racer-compact-simulator.5866/page-8#post-67999

    So far @BlazinH's explanation regarding what SimTools can and can not do re filtering is the most comprehensive I know of, at least in the area I was interested in, with @value1 kindly adding extra values so I could achieve a more realistic flight effect for use with the Rift: http://www.xsimulator.net/community/threads/dx-racer-compact-simulator.5866/page-9#post-68099
    • Like Like x 1
    Last edited: Mar 6, 2015
  5. yobuddy

    yobuddy Well-Known Member Staff Member Moderator SimAxe Beta Tester SimTools Developer Gold Contributor

    Joined:
    Feb 9, 2007
    Messages:
    5,282
    Occupation:
    Computer Technician
    Location:
    Portland, Oregon - USA
    Balance:
    49,207Coins
    Ratings:
    +5,093 / 18 / -0
    @Archie

    I can help with this.
    Adding a filter to a DOF for an Axis, does not change the original output % you set for this axis.
    For instance, if you add sway to Axis1 at 100%, it will always be 100%.
    The filter % is just how much the filter is being applied to this DOF (in this case sway at 100%).

    hope this helps.
    yobuddy
    • Like Like x 1
    • Agree Agree x 1
    • Useful Useful x 1
  6. Archie

    Archie Eternal tinkerer

    Joined:
    Dec 31, 2014
    Messages:
    1,081
    Location:
    Wollongong, NSW, AU
    Balance:
    3,796Coins
    Ratings:
    +1,379 / 4 / -0
    My Motion Simulator:
    2DOF, DC motor, JRK
    Thanks @yobuddy - So when a 30% smoothing filter is being applied to a 100% sway, what is that filter actually doing? That's what I am trying to understand. Sorry if I'm not explaining myself properly :)
  7. Pit

    Pit - - - - - - - - - - - - - - - - Gold Contributor

    Joined:
    Oct 2, 2013
    Messages:
    3,012
    Location:
    Switzerland
    Balance:
    30,439Coins
    Ratings:
    +3,089 / 31 / -0
    My Motion Simulator:
    DC motor, Arduino, 6DOF
    @Archie: Sometimes for better understanding it can help to try it out by experimenting with the values as x times asking for theory. ;)
  8. noorbeast

    noorbeast VR Tassie Devil Staff Member Moderator Race Director

    Joined:
    Jul 13, 2014
    Messages:
    21,051
    Occupation:
    Innovative tech specialist for NGOs
    Location:
    St Helens, Tasmania, Australia
    Balance:
    148,004Coins
    Ratings:
    +10,884 / 54 / -2
    My Motion Simulator:
    3DOF, DC motor, JRK
    I think experimentation is fine, and necessary, but with filters it is not always obvious how they are implemented . For example, originally I wrongly assumed 100% washout would be the full effect of washout being applied. In other words I expected that 100% would result in the fasted washout return when in fact it is the slowest, as it is applied as a % effect to that axis, which is what @Archie seems to be asking with his example. It is not easy to discern that from the documentation and when you try and have it be subtle, which it is meant to be, it is also not easy to pick up in experimentation when you are expecting something different, albeit arising from incorrect assumptions.

    As far as I have been able to determine there is not a lot of info re the actual use of filters. I suspect that is perhaps because they are not widely used in sim racing, likely because in the main I expect sim racing profiles do not try to implement a full 3D style simulation of say a rollover, which would bring into play the real limitation of our rigs. Of course I am sure there would be exceptions and I have seen a couple of examples in presets available for download, but no explanation of their use. I suspect flight profiles are likely to make more use of filters, but I have to be honest and say I have not found much info there either. I would actually welcome someone like @yobuddy giving a good overview of filters with practical use examples, and that being made part of the official documentation.
    • Agree Agree x 1
    • Useful Useful x 1
    Last edited: Mar 6, 2015
  9. bsft

    bsft

    Balance:
    Coins
    Ratings:
    +0 / 0 / -0
    I used to use boundary to take the big spikes out of the motion, but that was more to stop errors I had with JRKS. Time spent with motor settings and PID overcame this.
    The boundary helped soften overall big hits a bit, I think I ran between 5 and 30 depending on game.
    I dont use it anymore.
  10. prodigy

    prodigy Burning revs

    Joined:
    Oct 27, 2013
    Messages:
    459
    Location:
    Croatia
    Balance:
    6,698Coins
    Ratings:
    +399 / 4 / -0
    My Motion Simulator:
    2DOF, 3DOF, AC motor, SCN5, JRK
    @Archie

    The picture says 1000 words :)

    I believe the smoothing filter works like this:

    smooth_filter.jpg

    It doesn't affect the Axis range like @yobuddy explained, but will affect / smooth out the spikes and this would also affect the range of the effect itself (sway, surge etc) - making you think that Axis range was changed - but actually it wasn't, it was the range of effect that was changed.

    If you take a look at the big spike in the center of the first image (noisy data), you would see that it reaches a point around 70% of the axis, but on the second image with smoothing applied - the same spike is ranged at point somewhat below 60% of axis because of the smoothing, making a difference of about 10% in the range. The second big spike has even bigger range impact because of smoothing. So it's not your fault thinking Axis was changed, you were 'tricked' into it..

    And if you take a look, smoothing will also make you lose motion details and make your motion smoother but that could also feel slower in movement / response if you apply too much of it. Just imagine turning your car into a boat on waves ;)
    • Informative Informative x 4
    • Like Like x 3
    • Useful Useful x 3
    Last edited: Mar 7, 2015
  11. Blame73

    Blame73 Well-Known Member

    Joined:
    Nov 6, 2014
    Messages:
    1,213
    Location:
    Italy
    Balance:
    8,281Coins
    Ratings:
    +1,115 / 2 / -0
    My Motion Simulator:
    2DOF, DC motor, Arduino, Motion platform
    How much smoothing filter was applied in that example?
  12. prodigy

    prodigy Burning revs

    Joined:
    Oct 27, 2013
    Messages:
    459
    Location:
    Croatia
    Balance:
    6,698Coins
    Ratings:
    +399 / 4 / -0
    My Motion Simulator:
    2DOF, 3DOF, AC motor, SCN5, JRK
    @Blame73

    Actually, the above graphs are not representing actual force from Simtools, I've found those graphs on google from some other test but edited them a bit with some info. Actual motion graphs would be a little different, with positive and negative values etc.

    But the point was only to show how can smoothing affect the motion line. From the test pic above, I would say 15-20% smoothing was added..
    • Like Like x 2
  13. Blame73

    Blame73 Well-Known Member

    Joined:
    Nov 6, 2014
    Messages:
    1,213
    Location:
    Italy
    Balance:
    8,281Coins
    Ratings:
    +1,115 / 2 / -0
    My Motion Simulator:
    2DOF, DC motor, Arduino, Motion platform
    I see, it was very clear and informative!
  14. zigzag49

    zigzag49 Active Member

    Joined:
    May 27, 2014
    Messages:
    251
    Balance:
    712Coins
    Ratings:
    +220 / 3 / -0
    My Motion Simulator:
    2DOF, Arduino
    hi
    I also used the smoothing filter to heave Indeed, it was 3%, and has felt the change with the same filter too, I found the car too soft fly off the blow.
  15. Archie

    Archie Eternal tinkerer

    Joined:
    Dec 31, 2014
    Messages:
    1,081
    Location:
    Wollongong, NSW, AU
    Balance:
    3,796Coins
    Ratings:
    +1,379 / 4 / -0
    My Motion Simulator:
    2DOF, DC motor, JRK
    Thanks guys - Sorry been very sick this weekend so only just catching up.

    As you pointed out @noorbeast , just trying it does not always work in this case as some understanding is always needed, otherwise you are just fumbling in the dark and not understanding the output.

    All I know, is that at 60% sway and 30% smoothing on pCars I get really nice movement on the Sway axis.
    What this does for me is this:
    Lets say you go around a left then right bend in succession, in a real car you would still be swaying out of the left bend and you go into the right bend, and with the smoothing applied on the Sim it gives you a sense of what the backend is doing. If I skid and the backend loses it, it actually feels (tricks my mind) into feeling like the backend is coming around.
    Again, maybe because I am on a foot sim and my legs are giving me feedback as well, it's "feels" pretty awesome.

    I kind of wish I had not found them now as I keep tinkering, but it seems a nice ratio is 2:1, so if you have 60%sway, 30% smoothing is good. YMMV.
    • Like Like x 1
  16. Krzysztof Dej

    Krzysztof Dej Active Member

    Joined:
    Jun 3, 2020
    Messages:
    310
    Location:
    Poland
    Balance:
    1,090Coins
    Ratings:
    +73 / 0 / -0
    My Motion Simulator:
    2DOF
    I have a problem with the Surge Axis alignment in Richard burns rally rbr game. I set this in dirt rally and assetto without any problems, but in rbr I need to find a compromise, I have to set the washout filter. By a well-set axis, I mean such a setting that when I add gas on 1 gear, it gets a kick and the strongest deflection occurs, then 2nd gear is the same, only lighter and so on to 5.6 gears. If it adds gas, it deflects if it inhibits the return. When it comes to rbr, when I hit 1, it is in this state, if it brakes, it also comes back nicely. If I wait to cut off and punch 2, 3, and on, do the same. The problem arises when I do not wait to be cut off, I only stick in earlier. Although the car has turbo and there should be an impact, it does not occur or it occurs randomly with a delay. I caught the values of max min. I tested different values. As it writes in every other where I set it up. Does anyone have any advice on how to set this up? I have the impression that the problem is that after tilting the axis there is no room for tilting again? Maybe someone has experience in rbr. Maybe the profile was made for a different physics and the present ngp has a problem with it? Please help. I love this game
  17. noorbeast

    noorbeast VR Tassie Devil Staff Member Moderator Race Director

    Joined:
    Jul 13, 2014
    Messages:
    21,051
    Occupation:
    Innovative tech specialist for NGOs
    Location:
    St Helens, Tasmania, Australia
    Balance:
    148,004Coins
    Ratings:
    +10,884 / 54 / -2
    My Motion Simulator:
    3DOF, DC motor, JRK
    Please post pictures of your Axis Assignments and Tuning Center.
  18. Krzysztof Dej

    Krzysztof Dej Active Member

    Joined:
    Jun 3, 2020
    Messages:
    310
    Location:
    Poland
    Balance:
    1,090Coins
    Ratings:
    +73 / 0 / -0
    My Motion Simulator:
    2DOF
    When I capture max min it's about 20 and I feel ok (when I press throttle it's working fast and strong on 1st gear, 2nd slower and less quicker and... 3rd also less. When I save settings and press stop capture ist ridiculous becouse it's like 20 % those forces!! I don't know why. Those settings is without any filters like on screens settings. When I'm lowering the max min settings to 10, 5 it's starting working again but it's not good for me becouse I want on gseat surge force for the car about 300 hp to kick me on back not like that. I try with different car from 200 hp to 400 and it's feel the same on 1st 2nd gear its OK. On throttle from low speed feel fine but the other gears and constantly increasing speed from 3rd gear there is signal from put the gear but when I'm reaching max rpm for me I'm finding quite OK when I'm apply washout about 80 %. Output testing feel much stronger on game engine. I'm putting all 100 % axis assignments and not feel this at all like 100

    Attached Files: