Posts by Jet-Pack (IPACS)

    Terrific! Can't wait to see them!

    Have you ever made models for FSX/P3D? Apparently the "worlkoad" for the developer is much less than AFS2. There are such great models for FSX/P3D that some kind of FSX/P3D to ASF2 converter would be great. I would be willing to take that on (using converter), but just digging into the AFS2 wiki tutorial scared me off. And I have 30+ years of computer programming!

    I've spoken to several developers, it may be quicker initially but you spend much longer to get it to fly correctly and aerofly offers you much more freedom and power over what you are simulating. You can essentially program anything in the tmd file.

    Please don't compare the two world's here, one is made specifically simulating just airplanes, the other (aerofly) is a physics engine that does not even know if it's a car, a rocket an airplane or a helicopter. There are hard coded plane elements in the aerofly physics engine, the fact that it flies comes from simulated aerodynamical part of which you can have as many as you like or none at all.

    I think I already put that in the wiki...

    InertiaLength is the size of a box that is representative for the inertia. If there is a lot of mass concentrated in the center and only light structure around that, then the inertia lengths are smaller than the bounding box. If it's just an empty shell then it should be bigger.

    Take the bounding box for now but it's numerically more stable with bigger boxes

    Lengths are relative, so from r0 half a length in each direction, r0 is center of mass

    When you depress the center button for a bit longer it will always go to level off and wings level.

    If you press it short it centers one thing after another (keep tapping it): first bank hold and flight path angle hold, then wings level and fpa hold, then wings level and level off. I found that that was more useful during an actual manual flight than always leveling off, e.g. climbing turn or descending turn, where you just want to roll out on present heading.

    Hi Alex,

    the wheel can only rotate around its normal rotation axis, any steering requires an extra rigidbody.

    For your tailwheel you can probably copy the pitts or f4u tail gear construction and then delete the steering input and set the K and D properties of that joint to 0.

    K and D are spring and damping constants, K in N/m or Nm/rad and D in N/(m/s) or Nm/(rad/s).

    Increasing K makes the connection stiffer (oscillating frequency goes up) and leads to less gear contraction at a static load, making D higher dampens any movements but does not affect stationary deflections.

    If you make K and D too high it gets nummerically unstable, essentially the update rate of the physics engine falls below the frequency of the oscillation and things blow up and explode. Either parts fly away or you are stuck in an endless repositioning loop, in both cases check the leverage like tail gear joint to tail gear center of mass (R0 of rigidbody should be near the rotation axis) and if that doesn't hep then reduce K and D, e.g. setting them to K = 10 and D = 0.01 or something like that or straight to zero.

    Collision hull K and D are like spring and damping constants when the mesh goes below the terrain. Higher damping means less of a bounce when the part hits the ground.

    Weight and balance is currently not implemented as such, but we have prepared "passengers" and "fuel" masses in some of the airplanes. If you want to do that then use the empty weight of the fuselage + engine + oil + unusable fuel. And then add a fuel mass where the tank is actually located in 3D and set its mass to either full or some intermediate value. None zero though.

    hydraulic_cylinder is a contraction device, it pulls the the two bodies together or pushes them apart. It defines the maximum retraction but not the actual pivot and rotation axis of the landing gear (just like IRL)

    That happens when the uv map for that triangle has no area on the texture, the entire triangle gets mapped to a line.

    Or in other words you are projecting the texture not onto the surface but along the surface and it never hits the triangle.... texture projection is 90° to the surface normal...

    Change the uv coordinate for one of the vertices and the issue will go away.

    <[tmvector2d][TargetScale] [ 896 768]>

    This should be whatever your texture size is, e.g. 1024 1024 or 2048 2048

    Internally we compute TargetSize / TargetScale to give us UV coordinates from 0 to 1 again. You could also set the target scale to 1 1 and then enter the uv coordinates directly in the target position and target size values but no designer likes to do that, they work in pixels not fractions of total width and height.

    I'm always working on the hot tmd and reload within the sim (we have a short cut for this, check your control settings). I think there still is a safety check before you convert, it will give an error that the tmd and tmc files in your intermediate are older than the ones in the aircraft folder.

    R0 is relative to aircraft origin, so basically the same as your global 3d model origin.

    Hmm, not sure I follow. I just copy pasted from the F4u and changed the pivot, axis and radius.

    Roger, that is for the graphics but did you also change it for the [wheel] objects in the physics? Cause the collision happens in the physics, the graphics only needs the pivots to properly display the deformations.

    Sorry I edited my post just as you replied. Yeah the rough values for the wheel sizes are in.

    for your wheel graphics what position id and q are you using? if you attached the wheel physics to the fuselage you should do that in the graphics as well and if you still have them on the gear than use the gear or lower gear for the graphics, too.

    Is there a way to turn off the 2d display to the monitor completely? So that all available processing is utilized in the headset and not wasted on the monitor.

    From my understanding it does not cost any performance to just mirror the same image onto the 2D screen as well. It costs performance to render the 2d view in full screen, no split, though because then the entire scene needs to be rendered again. But just cloning the frame buffer and send it to the vr headset and monitor shouldn't make a difference compared to just sending it to the vr heatset.

    Jürgen, ich glaube das haben wir schon oft genug von dir gehört, jetzt gedulde dich doch bitte ein wenig. Wir managen mehrer Produkte auf verschiedenen Platformen und können nicht alles gleichzeitig machen. Immer wieder auf derselben Aussage herumpochen führt nicht dazu dass es schneller geht, eher nur dass es das ganze mehr verzögert, weil wir dieselben Fragen mehrmals beantworten.

    :huh::huh: That sounds incredibly useful! How in the world are you supposed to do it without a software like that?

    It's not as easy to visualize but still possible to get a mental image from the coordinates. We didn't have that tool for years and all of the Aerofly FS 2 aircraft except the Q400 and R22 were initially developed without it.