Alterscape

How hard they are to set up depends entirely on what quality you want and how well-made the machine is. Also, how handy are you? If you've never turned a bolt before, the challenge level probably looks much harder than if you've got some time in an electronics lab or metal/wood shop. I've got personal experience with the Lulzbot AO-101; I also know folks with Makerbots of various stripes and have a friend building a Rigidbot. None of it has been turn-key. I print stuff for prototype purposes at work, and the AO-101 has treated us fine. Suspect we could get somewhat better quality out of things if we put more into calibrating the extruder, leveling the bed/etc, though.

To be clear -- the Rift DK2 is definitely the best price/performance ratio in VR today by a longshot. Two years ago, if you wanted this good an experience, you were easily spending $20k+. My thought is not that it's not good, but that there are going to be better things coming down the pipe shortly, both from Oculus and other vendors. If you're a consumer and you can only afford to buy one HMD in the next five years, I suspect you aren't going to want the one you buy to be a DK2. I have not got my face in Crescent Beach (sorry, Crystal Cove was the pre-DK2 prototype, I misspoke earlier) but I'm told it looks _good_, and I can only assume whatever comes after will look better. I bought two more 24" monitors recently and I run nVidia surround at 3600x1920, with OpenTrack. For simulation I think it's a better option than any currently on-the-consumer-market HMDs. I'm hopeful about the future, though..

My perspective, as someone who develops VR apps: don't buy a DK2 as an end user -- wait for the commercial product. DK2 is ok, but the resolution is still too low to be useful for sims. Crystal Cove suggests the resolution is going to get another bump before commercial release, and also better head-tracking at all angles. Also, because of the way the DK2 rendering pipeline works, you need to reliably hit 75fps to avoid judder. I seriously doubt you're going to hit 75fps off a 560 in DCS. :(

Define "work?" -- I think you could bind them, but I don't think it'd be usable. The antenna elevation controls expect a rotary (ie, you can turn the knob to some position and leave it alone). The SpaceMouse axes return to center, like a joystick handle. So if you wanted to hold your antenna at a certain elevation, you'd have to keep a hand on the 'puck.' Usability nightmare -- just buy an X55 or X52. [edit]In case the OP is reading this, have you tried double-clicking on the intersection of the SpaceMouse column and the axis command row (eg: "pitch" or "throttle" or whatever -- the labels down the left side of the control screen) you want to bind and then twisting that axis on the spacemouse after the assign-axis window pops up? It just occured to me that since you're new you've maybe never bound any axis in DCS ever. Apologies if you already know this..[/edit]

I've got a SpaceMouse Pro with the latest 3dconnexion drivers (for CAD) and have to make sure to unmap it as axes in DCS so I don't get all manner of weird inputs from it. Not sure why it's not showing up for you..

Bravo zulu, Leatherneck!

Finished reading page one, flipped to page two about ready to describe more or less what Hempstead described, although less sophisticated. Read Hempstead's post, learned some things that should have been obvious about protocol design. Thanks for the thoughtful posts, Hempstead!

Thanks for the update, Cobra. She's looking great.

Thanks for the update, Cobra. Good luck with the remaining milestones until release. I'm too bloody busy with my own work to be mashing F5 in here, but I am looking forward to getting my hands on the finished product when it's done!

The main problem I see here is that this hardware seems to provide recognition of a limited set of hand gestures, not positional tracking of the hand or fingers. Based on my understanding, the Myo can tell if you're pointing, making a fist, have your hand flat, etc, but not where in 3d space your hand is. Won't map very clearly to virtual cockpit controls, I think.

I'm going to "me too" Raistlen's points. I fly DCS with some of the same people, and I agree that it would be useful to be able to check controls (and ideally rebind) while in the 'pit in MP. If nothing else, it would help with the "oh my, haven't flown this in weeks, now where did I put my key for <whatever>?"

Just got my 3 monitor setup, and neither the "single wide viewport" or the "3 viewports" settings work in a way that I'm satisfied with. I'd love to use something like a Directx9 interceptor shim to grab the calls to configure the viewing frustum and projection matrices for perspective-correct off-axis projections (given my monitor positions and freetrack eye position). Has anyone tried this before? Just want to avoid spending a ton of energy going down a blind alley if there's in-engine geometry culling or etc that would render this approach horridly broken.

V4Friend, is there any plan to allow the user to select which webcam to use for input when using the Point Tracker? I understand this feature is useless for FaceAPI, because the freebie face API uses only the first webcam enumerated. I'd like to try Point Tracker, but FaceTrackNoIR opens my laptop's webcam first, always. :( I know I could disable the webcam, but I use it for Skype and would prefer to avoid the Device Manager dance..

I can't remember the name of the airport I was landing at off the top of my head -- B-something? Waypoint 59 or 58 I believe, from the divert menu (I ended up there because I got shot up something fierce by a MANPAD). Does that fit?

Lights is and has been set to 2; still no joy.

In DCSW (A-10C module; have BS2 but haven't tried night-fighting in it), the engine does not render any of the runway lights! In a MP mission I flew yesterday, I realized that the other two pilots could see the runway lights (glideslope, marker lights, blue taxiway lights, etc) and I see no lights. Is this a known graphics setting thing? Would be really nice to have the option of not doing ILS every single night landing, even with CAVU conditions.

I'm in the very early stages of designing and building a custom force-feedback stick for sim flight -- something floor mounted, with a fairly long extension between the stick grip and base/pivots. As such, I'd have to replace the motors and motor drivers with something capable of providing enough torque with the stick extension, and probably build new motor mounts, and possibly new geartrains. I have access to a machine shop and some 3d printing gear, so I'm fairly confident I can take care of the mechanical part. My question: I have a "spare" FFB2 stick that I was thinking about gutting to drive my frankenstick -- connecting my own H-bridge to the motor outputs, etc, and connecting my pots to its pitch/roll inputs. The FFB2 seems to be extremely well-regarded by the sim community, however, and as the saying goes, "they just don't make 'em like that anymore." It seems a shame to wreck a perfectly good FFB2 if there are other options that would work just as well for my purposes. So, my question is, does anyone know if the problems with the Logitech FFB sticks (or Saitek, for that matter) were primarily in the hardware, or the firmware/circuit board? If I'm going to be DIYing my own gimbals/motor mounts/pot mounts/etc, poor quality in those parts on the original stick doesn't really matter. Thoughts?

Just to add a datapoint: I fly with Mike frequently, and we were flying a mission when he encountered the problems he's reporting. My rig is a Macbook Pro (ya rly) with a bunch of external hardware connected. Core i7 2.6 ghz; 8gb PC8500 RAM; GeForce 330M 512mb VRAM; 1920x1200 display All my graphics settings are at LOW; Cockpit Displays at 512; Disable Aero enabled; Clutter/Bushes at 0m; Trees at 1500m; Preload Radius at 150000m; all checkboxes disabled except "Disable Aero" and "Fullscreen. I am getting acceptable framerates with both the Shark and the Hawg and I did not notice a major hit between 1.1.1.1 and 1.2.0. Nobody's going to confuse my rig for a gaming machine, but I can fly. Not sure what's up on Mike's end, but his specs aren't that much lower than mine, so I'm not sure what's going on there. I don't think it's just the lower-grade GPU, though...

I hope some forum necromancy is OK here. PeterP mentioned in another thread that TriggerHappy had a parametric CAD (SolidWorks, I think) model of the KA-50 cyclic grip: Does anyone still reading this thread have a copy of the model? I can't find it earlier in the thread, and a lot of TriggerHappy's attachments/links seem to be stale. I'm interested in modifying it to be easily 3d-printable -- I have access to a Dimension 3d printer through my grad program -- and making my own stick grip, probably with the guts from a Sidewinder FFB2 or similar stick.

Do any good 3-views of the cyclic control (stick grip) exist? I keep tinkering with the idea of building up an Autodesk Inventor model and printing it on the 3d printer I have access to, but the lack of good drawings/reference photos is daunting.

After fiddling some more, I came up with the correct bindings. The trick was to assign an axis, and then open the axis in "Axis Tune" to check if it was correct. Turns out SLIDER_2 was the axis name I was looking for, after a bit of trial and error.

I work with Computer Vision as a graduate researcher. I'm more an end-user of CV tools than a developer of novel techniques, but my impression is: Computer vision is painfully slow. TrackIR and other marker-based systems make it faster by giving the computer a much simpler task ("find the 3 or 4 brightest objects in this scene, ignoring everything else, and then recover their orientation given known relationships between them" vs "Find a thing that looks like a face in this video image, and figure out what orientation it is in based on subtle cues that vary from person to person"). I am skeptical that Kinect-based approaches can match the speed/accuracy of marker-based solutions in the near future, although the depth buffer should allow better speed/accuracy than pure CV solutions like the PS3 GT5 setup, or FaceTrackNoIR.

Thanks for the info! This is all true but as I understand it, not the fix to my problem. My complete axis assignment is as follows: X axis: 1 Y axis: 2 Microstick X: 7 Microstick Y: 8 Throttle: 3 Antenna Knob: 4 Range Knob: 5 Rudder: 6. The question is "Inside the sim's config applet, what is the name of axis 8?" I know that 7 is apparently JOY_RX, and 4 is JOY_RY (which is odd but hey)..

I just purchased a used TM Cougar, and I'm trying to set it up for DCS A-10. I cannot just assign axes by moving the relevant axis, because the "ANT ELEV" and "MAN RNG" pots are currently inoperative and jumping all over the place -- I have replacement parts on order from Thrustmaster, but it'll be a week or so before they arrive and I have time to do the surgery to swap them out. I'd like to manually bind the other axes, particularly the microstick, in the mean time. I have the microstick bound as follows in CCP: Microstick X: Windows Axis 7; Microstick Y: Windows Axis 8. In the A-10C controls setup screen, the axis identifiers don't match up with the Windows axis names. JOY_RX is ministick X, but JOY_RZ is the ANT ELEV knob (bound to Windows axis #4 in CCP). Any idea which axis is ministick Y (Windows Axis 8)?

Haggart, The machine I have access to is made by Dimension. It's not a mill -- it builds up an object by laying down melted plastic out of a "print head." The cool thing about it is that the objects come out of the printer pretty much structurally solid! We also have a small 3-axis mill, but it's wood-only in the current setup, so not so exciting. You're right that a project box would actually work almost as well -- I might just make version 0 out of interlocking laser-cut acrylic. Thanks for the links!