[Daboule]

Hi gents,

A few years ago, I've experimented some things with MS Flight Simulator (two programs called Wideview and FSUIPC).

FSUIPC is an utility that exposes the variables used by Flight Simulator (speed, GPS Position, meteo, engine power etc...) and which is enable to modify those variables on the fly.
By this way it is possible to build real instruments like radios, GPS or any other fly instruments and to interact with the simulator. The instrument reads/write directly on the Flight Simulator by using FSUIPC. For example you set the radio frequency on your real radio and send this value to FSUIPC by the serial or parallel port to write it on the radio in the Flight Sim. You can also imagine sending the GPS position of your plane in the MSFlightSim world to a real GPS and display it on a real GPS, there is no limit...

Another cool thing is WidewieW, this program has two parts : a server and a client. The server exposes all the variables of FlightSim through FSUIPC and the client part read this values through the network and TCP/IP sockets, then duplicate the variables on another PC. The idea is to have a Master PC running FlightSimulator and as many as you want slave PCs using exactly the same variables values, except the angle of the view... you can by this way build a 3D cockpit like this : http://www.wideview.it/wideview.htm. As each screen is connected to a different PC you don't have performance issues because each screen displays an image calculated by its own PC with its own graphical card and its own processor...

The funny part now: I've already built a cockpit for FlightSim few years ago, and I'm on the way to build the same thing for Freespace(OK actually I'm only planning it). I have some small and dirty programming skills (as dirty as my English) and I wonder how hard it would be to write a program identical to FSUIPC for Freespace...

Thank you

Daboule

[phreak]

Well this is getting into the commercial sim realm here.  Anyway I don't think anyone here has the hardware (or the money to get the hardware) to do that.  not to mention the engine would need to totally be rewritten.

[Nuke]

well if were talking full motion simulation, i dont think it would be hard to take a ships physics vars every frame, then use shared memory space to export the data to some other program with some kind of serial/parallel interface to some kind of industrial motor controller. you could probibly get some kinda sensor feedback through either the serial/parallel or joystick interface. essentially find out how you want the rig to be oriented. read the sensors on the rig to figure out how its currently oriented and turn on/off the motors accordingly.

thinking about it i bet my lego rcx has the programming capacity for such a thing (just not the baud rate). just need to hook it to a relay operated power supply to some industrial gear motors. from recent robotics programming endeavors ive found out the rcx has 10 bit resolution adcs, which i wired to some 360 degree pots to create makeshift servos. would really be the same thing. its just a matter of hardware.

for the yaw axis id just use some sort of ring shaped rail system as the base. like those camera track systems they use for movies or just go down to a fabricator and have em bend some steel tubing for you. to make some sort of base structure you can lay out on the floor. next part would roll on that track via some gear motor rotated rollers. it would just need to be a platform of sorts. the yaw ais would probibly use some gearmotor driven jackscrews to elevate or drop the nose of the cockpit module, while the tail end is attached by some sort of uv joint. the roll axes would run from bearings forward and reverse pivots, probibly being chain driven by a gear motor. the cockpit would be a box with a seat, some, controlers and a monitor. either a big monitor and a track ir or you could probibly use a tripplehead setup, the game just sees a big wide screen monitor, then tweak fov.

[Reprobator]

if you're speacking about a triplehead software, it just work with directx at the moment 
I don't know if the matrox triple head hardware do work with scp too.

[Daboule]

Well this is getting into the commercial sim realm here.  Anyway I don't think anyone here has the hardware (or the money to get the hardware) to do that.  not to mention the engine would need to totally be rewritten.

Hi Phreak,

My MS Flight Sim 2002 simulator, cost me around 600 $... I done all by myself (cut the wood, build the instruments etc...).
I didn't buy the 3 PCs and the 3 19" screen I used for this simulator, because a friend gave them to me.

It's just to explain than with bit of chance it can be not so expensive (of course 600$ can be considered as an expensive entertainment but on the other side 600$ for a complete simulator is not a big quantity of money)... it just requires work for the hardware (wood is quit easy to work with when you're used to, but the first time you try to cut a piece of MDF it is not really simple...), but at the end, you're really proud to have done it by yourself.

It took me 8 months to build all the simulator...

Daboule

[Dilmah G]

Well this is getting into the commercial sim realm here.  Anyway I don't think anyone here has the hardware (or the money to get the hardware) to do that.  not to mention the engine would need to totally be rewritten.

Hi Phreak,

My MS Flight Sim 2002 simulator, cost me around 600 $... I done all by myself (cut the wood, build the instruments etc...).
I didn't buy the 3 PCs and the 3 19" screen I used for this simulator, because a friend gave them to me.

It's just to explain than with bit of chance it can be not so expensive (of course 600$ can be considered as an expensive entertainment but on the other side 600$ for a complete simulator is not a big quantity of money)... it just requires work for the hardware (wood is quit easy to work with when you're used to, but the first time you try to cut a piece of MDF it is not really simple...), but at the end, you're really proud to have done it by yourself.

It took me 8 months to build all the simulator...

Daboule

I thought he was referring to the actual game engine? 

[Daboule]

well if were talking full motion simulation, i dont think it would be hard to take a ships physics vars every frame, then use shared memory space to export the data to some other program with some kind of serial/parallel interface to some kind of industrial motor controller. you could probibly get some kinda sensor feedback through either the serial/parallel or joystick interface. essentially find out how you want the rig to be oriented. read the sensors on the rig to figure out how its currently oriented and turn on/off the motors accordingly.



thinking about it i bet my lego rcx has the programming capacity for such a thing (just not the baud rate). just need to hook it to a relay operated power supply to some industrial gear motors. from recent robotics programming endeavors ive found out the rcx has 10 bit resolution adcs, which i wired to some 360 degree pots to create makeshift servos. would really be the same thing. its just a matter of hardware.

for the yaw axis id just use some sort of ring shaped rail system as the base. like those camera track systems they use for movies or just go down to a fabricator and have em bend some steel tubing for you. to make some sort of base structure you can lay out on the floor. next part would roll on that track via some gear motor rotated rollers. it would just need to be a platform of sorts. the yaw ais would probibly use some gearmotor driven jackscrews to elevate or drop the nose of the cockpit module, while the tail end is attached by some sort of uv joint. the roll axes would run from bearings forward and reverse pivots, probibly being chain driven by a gear motor. the cockpit would be a box with a seat, some, controlers and a monitor. either a big monitor and a track ir or you could probibly use a tripplehead setup, the game just sees a big wide screen monitor, then tweak fov.

Hello Nuke,

In fact I don't want to make the simulator move.
I only want to build something similar to this simulator

I can have 4 x 42" Full HD plasma TVs for only 150 $ (because they have a problem with the audio circuit).
I'm already seeing me shooting shivans sitted on this seat...

Of course I can use a big video card with multiple video out, but this simulator use one PC per screen and another for the cockpit instruments.
Each PC display the same scene with an offset of 20 degrees. Each PC is running its own MSFlightSim but only one is exposing is variables, the other are reading this variable and uses it to display the same scene.
The PC dedicated to the instrument reads the variables (speedometer, altimeter, oil pressure etc) on the master PC and send them to an interface card connected via RS232. A microcrontroller on the interface card can then drive the small step by step motors that are used to move the gauges, light on the leds etc...
Another interface card uses a keyboard controller to send standard keykoard key codes to the simulator, so when you push a button on the cockip you simply send the same key code you would have sent if you have pressed a key on you keyboard (such interfac card can be found on ebay for 24$).
The big advantage is : you can add as many screens or instrument pannels as you want, without impacting the general performances...

If I can read the shields values for exemple I can display the gauges of the HUD on a small graphical application running on a small tactile screen and inserted in the control pannel of my cockpit. If I can write the values directly on Freespace, the tactile screen could also be used to change the shield value directly by touching tactile zones... I touch the gauges on the screen to tranfert enegy to the shield, my application send the corredponding keyboard key code to Freespace and the values are changed on Freespace, then my application read the new values from freespace and display the gauges with the new values on my small screen...

Exciting game experience isn't it?

But to do so, I need a way to read and write all the values from an external application. I try to figure out how long it could take to write a code that would be able to expose this values to other applications via TCP sockets...

That's all

[chief1983]

I don't think it would require a complete rewrite of the FS engine to support this stuff, but it would still be an extensive amount of work probably, to do it right.

[Nuke]

not a complete rewrite, you could probibly rewrite (and by rewrite i mean brutally hack) the network code to allow some sort of network rendering. sorta like multiplayer except youre running one player and a bunch of observers. i wouldnt call it a complete rewrite, but its something nobody wants to do cause they dont have the hardware to test it, few can use it, and it would probibly bork multiplayer or reverse compatability something.

it would be something like
convert all single player missions to multi missions
write a camera script for all the observers to lock it to whatever ship the player is viewing from including an offset vector and a matrix
then hack the engine to fix any bugs this might cause

to do it with one computer id
get a guadro/geforce sli array, 2-4 of them.
span the monitors in a single display
fix rtt in scripting (get it back to how it was on the build i use for my youtube vids, and fix that resolution bug we had)
have the engine black out all the pixels on all but one of those screens. (sorta like the bars on the death screen)
set up some cameras in scripting, lock em to the players position and give them a proper matrix
rtt those cameras to textures equal to one screen resoluton
draw thoe textues to blacked out areas

[chief1983]

Again, to do it right probably wouldn't involve breaking anything, and I don't think the lack of hardware is as big of an issue, since you'd only have to test with a program designed to feed certain inputs, so you could use a keyboard if you wanted to for testing the external program.  But there's an issue of who would benefit from all this effort, as cool as it is, there's even fewer eccentric people who would do this with FSO than there are with a real flight simulator.  It would still be awesome to have someday though.  It can be broken up into separate projects at least.  One part is the multi-pc rendering, and the other is the additional input/output device stuff (like writing to a LED display).  One person wouldn't necessarily have to do everything, and even any one part of it would be useful.  With a good set of matrox cards, and the aspect ratio bug now fixed, you could do some interesting stuff.  3 pcs, and you could have 6 monitors, and each pc still only thinks it's running the game on one monitor (go Matrox ).

[Nuke]

for $300+ bucks those matrox units better be damn good. and their drivers if they have any (and they shouldnt, since its in essence a screen demultiplexer) should work on whatever operating system i want to use. id rather just use sli or quadro cards to do the monitor stacking. ive found so many flight sim gadgets and utilities to have isues with their drivers and compatability with exotic operating systems (like xp64 or vista ).

[Bob-san]

for $300+ bucks those matrox units better be damn good. and their drivers if they have any (and they shouldnt, since its in essence a screen demultiplexer) should work on whatever operating system i want to use. id rather just use sli or quadro cards to do the monitor stacking. ive found so many flight sim gadgets and utilities to have isues with their drivers and compatability with exotic operating systems (like xp64 or vista ).

Meh. For $300 (for a DualHead2Go or TripleHead2Go), I'd rather buy a new motherboard and a new pair of graphics cards.

[killerkitten]

Hiya guys,

I used to play freespace, then falcon 4, so I hope I can offer something helpful here. As far as representing the various switches and bits that you'd need to build a nice pit, there are a range of hardware designs out there to allow interfacing a computer to an array of switches and transparently assigning those switches to keystrokes. The products are aimed at flight simmers, especially ms fs, but they'd be equally applicable here. The main issue is that most of these aren't cheap things to buy. I'd offer some links, but I'm a bit drunk and lazy to be googling up items I've not looked at in a while, at the mo.

As far as the multiple screen display goes, I'm not sure this is anything other than having a graphics driver to divide the workload and rendering between a number of cards and getting fs to detect and support that driver. That's basically how the matrox devices work..I assume.. having possessed a g200 in dualhead mode, years back. I've got a set of stereoscopic shutter glasses hidden away somewhere, too, that hooked into DX that perhaps present an alternate way of setting things up.

[Bob-san]

Question with the FS2 setup: each monitor is driven by an independent box? Because I'd be interested in doing something like that, though I'd prefer not to spend $300 for a pair or two of mATX desktops. I've got the hardware to do almost anything--my Corsair TX750 can run about 10 E5200s with IGP, but only ~3 with a decent graphics card (E5200+nForce 7100 mATX+4GB DDR2 800+HD4830). Other than that, would there be any way to configure that using gigabit LAN instead of RS232?

[voidSkipper]

Sorry to gravedig, but I was thinking on this myself recently.

All this speaking of networking multiple computers together and rewriting singleplayer to multiplayer... is it not possible, given enough video card grunt, to output what you would see from the left/right/up/back hatswitch views simultaniously?

Obviously there would be a lot on screen and it would be very taxing, but is it at all possible to do, ignoring whether or not any rigs could handle it?

[Ibn Saeed]

Well this is getting into the commercial sim realm here.  Anyway I don't think anyone here has the hardware (or the money to get the hardware) to do that.  not to mention the engine would need to totally be rewritten.

Hi Phreak,

My MS Flight Sim 2002 simulator, cost me around 600 $... I done all by myself (cut the wood, build the instruments etc...).
I didn't buy the 3 PCs and the 3 19" screen I used for this simulator, because a friend gave them to me.

It's just to explain than with bit of chance it can be not so expensive (of course 600$ can be considered as an expensive entertainment but on the other side 600$ for a complete simulator is not a big quantity of money)... it just requires work for the hardware (wood is quit easy to work with when you're used to, but the first time you try to cut a piece of MDF it is not really simple...), but at the end, you're really proud to have done it by yourself.

It took me 8 months to build all the simulator...

Daboule

Daboule thats impressive. I think its a very good idea, if you have the resources and the money to see it through.

Can you please post some snapshots of your simulator that you built using wood.

Thanks

[chief1983]

Yeah pics would be awesome.

Adding multiple monitor support to work on either a local monitor or a remote one would need some real abstracting of the output code, I think.  The game itself would have to be told which views to render on which monitors.  I'm guessing common layouts would looks something like:

Code: [Select]

2 monitors:
 ___
|   |
|___|   <- Up {fov} rad from straight ahead
 ___
|   |
|___|   <- Straight ahead, 0 rad with {fov} field of vision



Code: [Select]

3 monitors:
                                           Straight ahead
                                       ___      ___      ___
                                      |   |    |   |    |   |
Left {fov} rad from straight ahead -> |___|    |___|    |___| <- Right {fov} rad from straight ahead



Code: [Select]

4 monitors:
                                                ___
                                               |   |
                                               |___|   <- Up {fov} rad from straight ahead
                                       ___      ___      ___
                                      |   |    |   |    |   |
Left {fov} rad from straight ahead -> |___|    |___|    |___| <- Right {fov} rad from straight ahead

                                           Straight ahead

It really wouldn't have anything to do with the hat movements as those go more than just 1x{fov} past your straight ahead view anyway.  Panning would still be done normally with the hat switch, it would just pan for all 4 monitors.

Actually the AR of the monitors would have to be taken into account as well or you'd end up with gaps/overlap depending on how the fov is cut out of the possible viewing area.  But you get the idea.

The way the engine is set up, having it render more than one 'view' should be possible I believe.  Big problem right now is lack of experienced graphics programming ability though.  If we could attract an OpenGL guru who hates his life...

[Ibn Saeed]

If we could attract an OpenGL guru who hates his life...

Classic !!! LOL

[Reprobator]

Hi,
i just ordered a matrox triplehead2 go, i should receive it within 2 days.
I  was wondering wich value i should set in the launcher for my triplescreen setup (3* 1280*1024)
default is 0.75 for one screen, so i guess for 3 screen i should set it up something like 1.5 right?

[chief1983]

It depends on if it uses that angle for the vertical or horizontal.  And .75 is high anyway.  You'll probably just have to experiment.  I can say that you'll probably have some really goofy looking guages.