by laurivs » Wed 4. Mar 2015, 14:56
Hello everyone,
There is still a problem with roll acceleration value as it spikes when ever you pass the point of bank angle -179 to 180 and to +179. This would occur every time when you fly a complete roll. Or even worse if you fly inverted and jiggle your bank angle a few degrees right and left and then you'll continuously spike..
I used to use a spike filter + mean value filter (+ smoothing filter maybe), but this adds insane amounts of lag, which is horrible, especially if you're using a head mounted display such as oculus rift.
I have sent a private message to Sirnoname, asking for a fix to the lockon plug. All we need is a very simple code that keeps the roll acceleration value constant as long as you keep the roll input applied.
The following rant is about fighter simulation (or anything else you want to fly inverted) and a 2DOF rig.
I have to comment on people using bank and pitch; I hope that no one is using only them..! Why on earth would you only use them? They have nothing to do with g-forces as such, unless of course if you mix them with roll acceleration and pitch/vertical acceleration values. If you're using a 2DOF rig, using roll acceleration value gives you the most accurate feeling for fighter simulation IMHO. If you build a physical motion simulator, why not actually find out how a professional 6dof rig performs and then try to mimic everything you can even with a 2DOF rig. There is no sense in making a good physical product and the completely mess the software side. It's like people who buy a 60 000€ sound system and still use an old 32" CRT TV, or vice versa..
When you perform a roll in real life, you do feel the initial roll ( bank angle changing) and then you feel positive Gs on your arse and back and everything else as you pull back on the stick to keep the airplane level while turning. What you don't feel is falling to the side. Why? Because you're pulling Gs! If you were to use the bank angle only, this would not accurately create the real feeling!
However, if you maintain constant heading and keep a bank angle of say 30 degrees to the left, this is when you will feel "leaning" or "falling" a bit off your chair. I use slipball value to recreate this(not bank!). It gives you exactly the same feeling as in real life. This also works when on ground and taxiing (2DOF), giving you an impression of linear g-forces when oscillation the plane to the left and right with rudder. If you crash your plane in such fashion that the left main wheel collapses and you are now banked to the left and sitting on the tarmac, not moving, you will notice that your 2DOF seat is also tilted to the left giving you an accurate feeling of having the aircraft tilted on its side.
How about using pitch? Well, when you are in a constant cruise climb, not accelerating, not banking, not changing your pitch: if your pitch is a constant +10 degrees ANU (aircraft nose up), the only way you're going to recreate the proper G-force with DCS and x-sim that I know of, is to use the pitch value. It's not perfect but it works OK.
When doing loops, if you use only pitch and not any other acceleration values considering the vertical axis, you will get funky results. When you pass the highest point of the loop and are flying inverted, if you keep the loop all around positive G's, a 2DOF seat should be kept in about the same position throughout the loop. This will not happen when using the pitch value only, as the pitch value tells you that you're going nose down, and thus wants to move your seat forwards. In real life you would in fact feel quite the opposite! You would be kept very firmly in your seat as long as you have positive G's of course. In real life our inner ear feels that you're rotating, but your 2DOF seat shouldn't do this, as this is not the primary feeling that should be recreated! You are limited to simulating g-forces with what you have and using the pitch value only, gives you the opposite results that you want! However, if you mix the pitch value with "Accl units"(in x-sim converter), you can have the accl units override the pitch in such cases and you will get more correct results!
So, all in all, you can get around many of the problems associated with x-sim, dcs, and a 2DOF rig, but the roll. acceleration bug is very annoying. You can get past it using math plugins such as spike filter, mean value filter, and smoothing filter but these will considerably increase lag in motion output, and a bit of unnecessary extra stress on cpu(not that much really..)
Attachment: my current setup on the x-sim side. (using 2DOF rig with wiper motors + rudder+stick simple ffb motors etc)
-L
- Attachments
-

laurivs x-sim setup01.jpg (455.07 KiB) Viewed 3737 times