21 replies to this topic

[Golden Gun]

The added Joystick support has been great! But not to sound ungrateful, One big fix was over looked.

To bring joysticks up to the same speed and accuracy as mice, the input needs to be changed. instead of constant movement of the the torso and arms when using a joystick, the input should be changed to a grid reference type response.

Constant movement is when you push the joystick half way to the right, everything will continually move to the right at a medium pace until it hits its threshold. The further you push the joystick away from the center, the faster it moves.

In contrast, grid type movement would set the joystick's center as 0,0 and all movement would equate to an x,y value and the torso and arms would move to that value. Since -255 to 255 are the standard extreme limits for most joysticks, each x value would equate to a certain left/right torso twist and arm movement and the y value would determine how high or low you were looking and aiming.

Using this scheme would mean that moving the joystick halfway to the right would now cause the torso to twist half way to the right and stop. How fast you twist would now be determined by how quickly you move the joystick from one position to the next. This would not just allow better control over the weapons and speed of moving them (we'd now have full use of torso twist speeds), but adjustments wouldn't require all the joystick "wagging".

I'm not a programmer and have no clue if this is even possible. I also realize the arms' ability to move further then the mech's torso twist and up/down looking would have to be factored in.

These issues aside, I really think making this fix will give Joystick players the edge they need to take on mice players.

Thanks for reading!

Golden Gun

[Inconspicuous]

Good idea.

[MCXL]

The problem is if you have 256 positions on a joystick (pretty standard 0-255 X 0-255 Y) and 180 degrees of side to side sweep (90 degrees to each side, which is pretty low) then each notch on the joystick is 0.703125 degrees. That is a LOT at range.

The sensitivity would be insane, at 300m from a target one degree is 5.23333333333 M, so that same sweep on the 180 degree mech would be 3.71 ish per tic on the joystick. Now try hitting something at 600m where moving your joystick less than one mm makes your crosshair move 7.42 meters across.

I guess this would be a way to make long range mechs useless for joystick players. No more ERPPC for you!

I understand the ideas root, but really the whole point of a joystick is so that you can do minute changes, while maintaining a smooth, linear flow.

If you really want to try it you can use the program Xpadder to convert your joystick input to a grid allocated mouse input (used mostly for controlling the mouse on a joystick only computer), then run the mech in non arm lock mode and you might see the problems inherit in the idea.

-MCXL

Edited by MCXL, 12 May 2013 - 12:23 AM.

[Golden Gun]

When I do my calibration check in windows and click on the "Show Raw Info" gadget, the range on a joystick is -255 to 255 for both x and y, not just 0 to 255.

I'll give Xpadder a try and take some notes to see how well it works and post them here.It may be a week or so since I have a week long training starting Tuesday.

Thanks for the lead!

[MCXL]

Some nicer/newer joysticks have dual range, as you describe, but that only doubles the fidelity. That's still 3.7 m of sweep per joy tick at 600m range.

If you can be precise enough to hit things with a setup that finicky more power too you, but I don't see it happening.

Edited by MCXL, 26 May 2013 - 11:44 PM.

[Inconspicuous]

MCXL, on 12 May 2013 - 12:17 AM, said:

The problem is if you have 256 positions on a joystick (pretty standard 0-255 X 0-255 Y) and 180 degrees of side to side sweep (90 degrees to each side, which is pretty low) then each notch on the joystick is 0.703125 degrees. That is a LOT at range.

Actually the stick I use has (for x) -1 to -255 (left deflection) and 1 to 255 (right deflection) so each notch on the joystick is closer to 0.35 which should be pretty good at most ranges. I mean MW4 can do it, IL-2:1946 can do it, Old racing games can do it. Why is it that PGI can not?

[MCXL]

Yea, it worked OK at best in MW4, and the ranges and fidelity of this game are more dynamic. If you are saying that an air sim had this I don't know how to react to that.

Racing it makes perfect sense to have it be that way, because the game doesn't see a difference from the stick to the wheel. It's the same thing to it, and since you aren't aiming at a dot on the horizon the fidelity of those increments is much less important. On top of that, by affecting the angle of the wheels a static amount you affect the cars turning rate, which is a consistent factor based on the angle of the front wheels to the rear and the speed of the vehicle (minus slip adjustments) EG even though I turned my wheel and let it sit there I am turning around and around.

Again, I tell you, try it with X Padder, you can map the input that way, that is as a mouse that returns to center every time.

[Inconspicuous]

X-Padder does a terrible job with mouse emulation - the diagonals don't work very well...

It worked well in MW4 you could be competitive with a joystick.

And to get into the specifics I have access to my joystick driver source code and the ranges reported to windoze are quite larger that in your example.

Quote

// Joystick Axes
if (xaxis > 16129) { xaxis = 16129; }
if (xaxis < -16129) { xaxis = -16129; }

-16129 to 16129... The fidelity is there, other games can do it fine... We really don't need to make up reasons that it does not work when it works fine in other games using the same (or similar) hardware...

Edited by Inconspicuous, 27 May 2013 - 01:01 AM.

[Loc Nar]

Just found this thread, and for you guys that haven't yet read my writeup on controls, you, and MWO, you really should. Knowledge is power, but so long as people use tired misconceptions as the basis of their arguments, well... lets just say it's not super-effective at convincing anyone that has bothered to do their homework.

http://mwomercs.com/...ls-demystified/

Arguments about digital resolution and bit rates of controllers has little if anything to do with why a regular joystick seems is so disadvantaged against a mouse or other suitable controller for MWO, although is an interesting topic on its own...

[MCXL]

Just a quick search produced this PDF which is a sheet with some various joysticks and the resolutions associated. 256 scale/256x2 is VERY COMMON.

http://puu.sh/32KLB.pdf

This is all determined by the bit depth the on board control chip has. Since old joysticks were actually MIDI based they use a 0-255 scale. Often a manufacturer will use an extra axis to double the range of a controller, but not that often.

Sure, nice joysticks might have the resolution for this type of control, but there is also the issue of the sensitivity, Even if the resolution is infinitely high, the joystick still only will have approximately 25-35 degrees of angle to it, which means that your motions get massively up scaled on a lateral basis.

[Loc Nar]

Quote

Sure, nice joysticks might have the resolution for this type of control, but there is also the issue of the sensitivity, Even if the resolution is infinitely high, the joystick still only will have approximately 25-35 degrees of angle to it, which means that your motions get massively up scaled on a lateral basis

Hmm, I think you mistake my intentions, and perhaps skipped the inportant parts of the article. I'm not at all advocating joystick use. I'm advocating proper use of the word joystick since it's used too loosely and in the process bypasses the real issue that prevents joysticks from working and these conversations from having meaning. Control-order types is the necessarry starting point... then the road is paved to ponder the devices best used achieve them. Leaving control order out of the discussion and getting on to why stick resolution is the problem is like concerning yourself about the gas mileage a car gets (and engaging in discussions on the merits of good gas mileage) when you need to cross a body of water... it doesn't matter cause it's not a boat.

To clear up any ideas about the zero-order stick I built just for mech piloting and to prevent limitations that don't apply from being assigned to it; it's a mutated Cougar (0-65535 on x/y...) but built around a highly specialized (not high tech, just purpose built) zenith/azimuth gimbal, which is then running as a mouse emulator in TARGET. The gimbal is butter smooth and highly damped. The stick's y-axis moves +/-22.5 and allows ~1:1 tracking in that axis, and although I can technically use <170deg of the range of the x-axis (no stops, Hall sensors good to 180 but only linear for ~170), my wrist prefers something closer to +/-60deg for full range of motion and spend most time between +/-45, but this equates to 1:1 to near 1:1 on most mechs.

It has on-the-fly sensitivity change for those snipy moments, and in normal mode has full range of motion with minimal amplification if any (this changes from mech to mech...). The stick always maintains the same physical orientation as the mech onscreen and stays in position if not being moved. Again butter smooth and has no detentes or centering springs fighting the movements, no deadzones and is using absolute outputs. It works really well in MWO, but is also truly an anomaly. It's a joystick yes, but not your garden variety first-order spring-loaded jobby and has more in common with a mouse than a normal joystick, other than it's shape which offers quite ergonomic access to more buttons than are needed to operate every in-game function of MWO, and owing to TARGET running it as an emulator has so far been imune to the many times MWO stick support has been wiped out/made useless from patch to patch, not to mention my kbm is active the whole time too.

As far as analog turning and throttle go though, resolutions of 0-255 are <100% as effective as full 16bit since those are not precision inputs and in MWO are actually well implemented first-order inputs. The need for precision decreases as control orders increase since as control orders increase, pretty much in proportion to the lag between inputs and reactions... For aiming the reticules in MWO howver which are designed around zero-order inputs, if one were using one of those 8 bit joysticks, then the specific argument you are making for resolution indeed comes into play, and makes another good case point against using joystick this way, particularly a cheap old one. It's a non-stater though since it's a first-order controller being used in a zero-order application, so resolution is the least of it's worries and even the best resolution stick can't compete on even footing with even a cheap mouse.

I love using my mech joystick with MWO because it's a dedicated zero-order controller designed from the ground up to be exactly what it is so it does it quite well, but the rest of my joysticks are essentially a nightmare to drive a mech with since they are still first-order controllers, including the much revered Warthog with its 16bit resolution since that means little in the face of a zero-order application.

[Golden Gun]

Hey all, I tried the Xpadder prog. and it was too jittery for any real testing but that has already been stated .

The reason I know my joystick can do what I stated from the beginning is that it did it for Steel Battalion!

Yes, I'm one of those crazy pilots...

All kidding aside, the same stick covered the same range, if not more, of motion with very fluid and exacting precision in both Steel Battalion games so I not only know it is possible, I know it is possible with THIS JOYSTICK!

So getting back to the original question, how do we do it? Can we change the DPI within the driver? Is there a better software layer/wraper we could use than Xpadder?

What do they (Piranha) need to do to do it right? Is there code/script for Cryengine that can implement this? I know others have posted that they were able to do it in other games, but I've read that one of the bigger challenges with Cryengine is that it wasn't built with joystick functionality. I wonder if this is why we aren't hearing any more about the super cool joystick that was shown off a bit ago...

[Phobik]

found this in another thread to help tweak joystick settings. still have not found a setting that plays as well as a mouse but it did help some.

Inside the Mechwarrior Online Folder edit the user.cfg file with:

cl_joystick_gain = 6
cl_joystick_sensitivity = 0.85
cl_joystick_throttle_range = 0
cl_joystick_invert_throttle = 1
cl_joystick_invert_pitch = 1
cl_joystick_invert_yaw = 0
cl_joystick_invert_turn = 0
cl_joystick_deadzone = 0.1

cl_joystick_gain - The higher this value is set to, the faster a full throw of the joystick is going to move the reticule.

cl_joystick_sensitivity - The smaller this value is set to, the more sensitive your joystick will become. E.g. a slight throw of the joystick will trigger the movement. i.e. If you set this number to something big like 5, you'll smash the crud outta your joystick trying to get anything to move (which is funny in itself).

cl_joystick_throttle_range - This is a flag to tell your Joystick to have a middle STOP zone when the throttle is centered. Pushing forward makes you go forward, pulling back makes you go backward and putting the throttle in the middle will make you stop. I HIGHLY suggest you don't mess with this value. No I'm not going to help you with anything if you've messed with this value.

cl_joystick_invert_throttle - This will make sure that when you push forward on your throttle, you're going to accelerate.

cl_joystick_invert_pitch - When this is set to 1, you will look up when you pull BACK on the joystick. You will look down if you press FORWARD on the joystick. This is how airplanes work and how you should leave it unless you're some sort of demon spawn that does it backwards. If you are demon spawn, then set the value to 0 if you want to look the wrong direction when pressing forward or back on the joystick.

cl_joystick_deadzone - This is the amount of non-movement wiggle room when you're joystick is centered. The smaller you make this, the less wiggle room there will be.

[Loc Nar]

Quote

What do they (Piranha) need to do to do it right? Is there code/script for Cryengine that can implement this? I know others have posted that they were able to do it in other games, but I've read that one of the bigger challenges with Cryengine is that it wasn't built with joystick functionality. I wonder if this is why we aren't hearing any more about the super cool joystick that was shown off a bit ago...

Simply put, reticule aim in MWO is a zero-order application and will never be able to be manipulated as effectively/efficiently with anything but a zero-order controller. This has nothing to do with PGI 'not getting it' or not understanding how joysticks work, or being big meanies trying to spoil your fun, although I'm not sure why they don't explain this and educate their users instead of leaving users like me to try to clear up the myths and misunderstandings.

There are no limitations on cryengine of which you suggest, and in fact analog throttle and analog turning are beautifully functioning first-order inputs; and as such first-order devices like throttles/pedals etc work quite well with them, better than kb in fact. A lot better. A dedicated throttle and pedals are first-order controllers, as is a normal spring-loaded joystick. While these are the goto devices then controls are actually coded as first-order inputs (directional velocity) such as throttle/turning or flight simulators or previous mech titles even, they can not compete on even footing in a zero-order environment like reticule aiming in MWO or browsing the internet, which is the same exact coding and it's positional based. The mechanical strengths that make them excel at first-order are all hindrances as soon as you slave them into zero-order conditions. I have demonstrated that a zero-order controller is not limited to a mouse, so one needs to call it like it is -zero-order control.

I wish you all the luck with your SB controller, but unfortunately even if you had full functionality to use it to the absolute best it was able to muster, it will still be far short of any and every zero-order controller. Even the great Thrustmaster Warthog is just as much of a fish out of water in a zero-order application. A $500 HOTAS, and it's more of a liability than an asset, even with the utter control and flexibility TARGET allows. It's too bad that so many people are stuck so far behind simple hardware compatibility issues that they rarely get to actually experience what those controls they are working so hard to implement are really like. They are doomed before they start due to being the complete wrong device for the task. While first-order applications such as flight sims or previous mech titles that are coded as such can be adapted to zero-order controllers such as a mouse, the same does not work in reverse. Once down to zero-order, only a zero-order device will be completely effective, and all else represent different levels of compromise.

The exact myths that continue to plague these discussions are exactly what I was addressing when I wrote Controls Demystified(?), which I highly recommend reading to better grasp the issues. I'm not trying to derail this thread, but without understanding what is actually going on in game or hardware-side you will be no closer to coming up with a solution or figuring out how to maximize a sub-standard setup. I'd say I qualify as 'one of those crazy pilots' too. Consider the lengths I was willing to go to make a zero-order joystick just so I can play MWO with a stick and not have my hands tied behind my back as a result. It has also forced my hand to learn a lot more about controls and how they work both game-side and hardware-wise than I intended, and I almost want to thank PGI for that. I fully advocate alternate controls, but the journey should really begin by gaining understanding of controls and how they relate to games. YMMV...

[Golden Gun]

Tankz for the TUT on inputs LOC Nar. I think I have a much better understanding of what you are talking about now and, hopefully, can use the same terms so we can be on the same page.

I agree about the inputs strengths and weaknesses you describe within certain game types, but the Zero-order type input can be replaced by a First-order if done right.

I am going to go back to Steel Battalion as it is the best example I know of. In the game you are completely sealed into the VT (mech) and every thing is relayed to you via cameras. Also, you control the main camera, weapons and VT direction separately each with it's own joystick.

The direction, or steering, joystick is a second-order control as it only has a lateral axis.

The viewing joystick is a first-order control as it has spring return and the ability to control the rate of motion by pushing more or less in a certain direction.

The weapons joystick is the enigma. It has two axis but no spring return. In Steel Battalion you aim with it by shifting the joystick towards the target until the cross-hairs are set.

In many ways this is just like using a mouse without having to take it off the mouse pad. If you moved either the mouse or the joystick say up and to the right about half way to the corner, the weapons would dip down (I play inverted ) half way of it's range and to the right half of its range and stay there. Both Joystick and mouse do not move back to the center and if they did, so would the targeting system. Also, the area covered in Steel Battalion by the weapons is roughly the same as in MWO.

Also, how fast you move from one position to another determines the speed of travel. Right now this is the same for mice.

Going back to the pros and cons of zero and first order control types. I think there is more room of first-order controllers in MWO than something like a FPS which is made for zero-order / mouse controls. First off, MWO viewing controls are in a box, meaning you can only look so far left and right, up and down until you hit your limit. In FPS games you can continually look left and right until you get dizzy. I concede you can't do that up and down since most games have a button to do back flips and forwards rolls.

What I'm getting at is that a joystick, responding the same way as in Steel Battalion would work in MWO because they both are "boxed" in. Therefore you COULD make it so that the top right extreme of the joystick is the same for the targeting in MWO. I know that each mech has its own limits, but that can be mapped as well.

When doing the calibration of my controller, it does only give a range of 0-255 for both axis. This is why I was wondering if there was a way to up the DPI in the driver so there would be a higher resolution. If so, the Steel Battalion sticks, or any other joystick, could have the finesse to work like a zero-order controller.

I hope this makes sense, I'm typing this up at work and have been side tracked tons!

[Loc Nar]

I'm impressed that you read it and got some understanding out of it. Not impressed because you lack any capacity or anything like that; impressed you took the initiative even after my borderline hostile/whiny response (not intentional) and understood all or most of it and are applying it towards coming up with a solution to your problem you're trying to solve. This. This is why I wrote the article.

Quote

but the Zero-order type input can be replaced by a First-order if done right.

While true it *can be replaced with first-order device, a first order device can not compete on even footing as a zero-order one in a zero-order application. There are many steps that can be taken to minimize the gap between them, but there will always remain a gap. A zero-order application relies on direct manipulation of position, and a first-order controller uses directional velocity to 'drive' the cursor/reticule to the new position. It will never be 100% as effective, but again steps *can be taken to reduce the problematic aspects within the finite limits of what is possible.

Quote

The weapons joystick is the enigma. It has two axis but no spring return. In Steel Battalion you aim with it by shifting the joystick towards the target until the cross-hairs are set.

Interesting information, and I will now begin trying to solve the enigma. By your description is sounds like a zero-order affair, but may be coded first-order and in practice achieves something like the classic 'aim to fly' system used for adapting zero-order controllers to first-order stuff like flight mechanics. Take all this with some salt, cause I'm shooting from the hip and need to do some actual research on this after hearing this intriguing bits. Hmm...

Quote

Going back to the pros and cons of zero and first order control types. I think there is more room of first-order controllers in MWO than something like a FPS which is made for zero-order / mouse controls. First off, MWO viewing controls are in a box, meaning you can only look so far left and right, up and down until you hit your limit. In FPS games you can continually look left and right until you get dizzy

Yes for the reasons in my above explanation, since there is nothing stopping it from working, but any first-order controller will always take a back seat in a zero-order application (mind you in light of the above, the jury is out for the moment on the SB controller until more information is known). MWO is coded exactly like any FPS or browsing. The fact it is limited to a small window (emulating a mouse in a browser with my MWO settings yields ~25mm x 14mm of cursor movement on a 23" 1080p screen BTW) changes nothing of what order it is or how it can be adapted to. In fact, a first-order controller is pretty much the only type of joystick that adapts to that environment, since one is relegated to using relative inputs because you need the ability for infinite travel in the axes. Hawken is a perfect example. My fancy mech stick is worthless in that game because they tied movement steering to reticule aim, giving the the need for infinite travel in that axis. Made me a sad sad panda, cause the only way to control the mech if I want to be able to aim and walk in a 120deg arc I *have to use relative inputs aaaaand there goes my zero-order controller. I have been thinking about revisiting it and keeping my zero-order on my stick but using the pedals to generate mouse x-axis relative commands, allowing me to steer with my feet and aim with my hand like in MWO, but I'm not sure if Hawken will allow this in it's control bindings.

Quote

When doing the calibration of my controller, it does only give a range of 0-255 for both axis. This is why I was wondering if there was a way to up the DPI in the driver so there would be a higher resolution. If so, the Steel Battalion sticks, or any other joystick, could have the finesse to work like a zero-order controller.

Unfortunately if this is correct, it may kinda be a deal breaker for the reasons MCXL has covered, even it it *is a zero order device. As to there being a driver or some way to change that, I don't think there is since it's generally determined by the actual chipset inside the controller itself. There is a possibility that it is artificially dumbed down to 0-255 to make it console compatible (being an x-box game/device...) and that the sensors and chip are capable of higher resolution. Another interesting thing to look into. If you have any more details you can add about SB or it's controller, I'd be happy to hear more cause it sounds like there's more to it than I am aware of. Also if you know any good places to start looking up more, I'll check them out. Until now I've just lightly combed Hack-n-Fly/Von Pilsner's efforts in getting it to function. Sort of.

[Golden Gun]

Once again, sorry for the long delay.

The Steel Battalion controller does not have an official driver, but uses Vjoy as a portal to 64-bit Windows systems. I think it is Vjoy that uses a set 0-255 limit for the driver. This is why I think, like you mentioned, that the controller is dumbed down and not allowed to use its full potential.

For Steel Battalion specific information, I'd say LineOfContact.NET (BTW, LOC is the abreviation. I thought you'd find that interesting) and steelbattalion.org have the most comprehesive information about the game and controller.

[Loc Nar]

Heh... no worries on being a lagger, it's practically my middle name! I've been looking around on that site already since these revelations about the SB controller having a zero-order joystick built into it, trying to confirm or deny this. While I wasn't able to yet turn anything up on LOC (which I do get a kick out of...), another interesting and fruitful conversation with another peripheral junkie confirms at least as much as it being an actual zero-order joystick. This being the case, the SB controller is definitely a ripe candidate for hacking the hardware to a new board since it's already mechanically a contender. In actuality that's the easier part since it's soldering and reading, with little to no shop tools involved and is quite doable. More doable than finding some magic driver that increases resolution, since it's highly unlikely the controller was capable of higher, and besides even if that did work the advantages of hacking new guts to it outweigh the advantages of even a best case scenario using the stock SB board.

The Teensy++, or Leo Bodnar boards *could be used, but this is an utterly perfect example for a Thrustmaster T16000M to give up its guts for a worthy cause. At $40 it's cheaper than the raw boards but comes with an MLX90333, the same x/y high res dual axis (single chip) sensor from the Warthog and more importantly than it's high res it's a TARGET capable board which means you can grab it by the balls the way you need to as the final step in order to make this work. TARGET allows absolute inputs and is not even hard to set up dual sensitivity and such in addition to many other useful features. Take a look at this thread that shows what's inside the T16000M, cause I think it would be an easy conversion since you would only need to attach a magnet to the bottom of the SB gimbal in proper relation to the T16000M's chip stationary below it. In addition to the easy sensor and 2 other pots that can be used for whatever else, all 16 buttons on the stick and base have discrete (independent) wires on them, making rewiring it to the SB switches a trivial matter so long as they're not integrated to the SB board.

Do you have an SB controller handy that you are willing to open up to take some pics? All I ever find are lame ones with resistors and diodes but the boards are irrelevant for this project. We need close ups of the mechanical aspects of the gimbal and it's friction joints, and some shots of it's buttons in relation to the boards. What I envision is completely re-gutting the right controller, and leaving the rest alone since it seems to be able to be used in it's current form, just not for aiming the torso, although it's unclear if the SB controller will run without being hooked to it's guts. There are workarounds for all of this, and my interest in piqued. The t16000M has the brains, and the SB stick has the brawn, and the integration of the two is likely very feasible.

[Golden Gun]

Yes, I have 4 controllers but you can get some really good pics of the innards of the controller here.

http://www.bigmech.c.../hacksmods.html

He's done several Steel Battalion controller mods but it is the first one you want to look at. It is how to add a rumble mod to the sticks and shows excilent close-ups or all three blocks opened up.

Side note: I have 4 sets because I was originally going to make a LOC Cell of fully imersive pods. Fully imersive=interactive surroundings to include Reactive lighting, 5.1 surround sound, visual responses to certain switches and buttons and force feedback in the seat. I have some rough diagrams and lists of parts up on da web but the LOC.NET forum has been down for a while.

[Loc Nar]

Confirmed. Totally doable to gut a T16000M to replace all it's guts, and to make it extra easy the T16000M handle could be left in tact and swapped to this gimbal, which would easily accept the magnet and sensor. Here's the money shots, and good catch on that site as it shows me everything I need to see. BRB, going out for an SB controller

Edited by Loc Nar, 04 June 2013 - 05:12 PM.