Support for Nascar 2003 and F1C

[dradecki]

I posted this in the development forum, but it probably belongs here:

I see support is coming for EA Nascar 2004, but that really doesn't help alot of us. Since EA's games for the most part is somewhat lacking. F1C is an exception. I did an experiment that added another joystick which I assigned to the look left and right to the X+ and X- axises. These worked perfectly for both F1C and Papy's NR 2003. I suspect this will also work for GPL. Why not hook into DirectInput and add the ability to assign the head moment to joystick axises instead of native support or mouse moment support. What issues are there with this? With this capability you don't have to wait for game companies to support your product. Hopefully you will take this into consideration.

If you need help with this please let me know.

Thank You,
David Radecki
dradecki@charter.net

[Quboid]

That's an extremely clever idea - many games have axes control for looking left and right (although not up/down) and this should be doable. I think that this would basically be a third mode, along with mouse and enhanced. It should be no problem to tie this in with enhanced mode's profiles so we can have nice deadzones and such.

[ August 12, 2003, 06:03 PM: Message edited by: Quboid ]

[dradecki]

I just noticed today that someone else already posted a similar idea, or should I say request in the Feature Request Forum. So this idea I guess isn't original, but I hope NaturalPoint will give serious consideration.

[Jim]

Hello:

We are curious about this idea and I am not opposed to it by any means.

Because I am not a programmer, just an engineer, please tell me what the precision of a DirectInput axis is, what is the value range? I am worried that it isn't precise enough.

[dradecki]

You know that's a good question. But axis value precision is a relative term. It's probably not going to be as exact as say cx/cy values are for mouse cursor movement, but should be more then sufficient for an application like this. For example in the axis experiment I tried with a second joystick I was able to get very fine movement with both the throttle control of my MS sidewinder joystick, and also with the X+/X- axis of the joystick, so I believe the precision is there.

It's been awhile since I programmed a WDM driver (I've been doing Linux and VxWorks drivers over the last couple of years), but I have been looking into the Mouse Filter driver in Windows and this looks very do-able. I would just prefer it was somebody else that does it, since I am time limited right now. But I am continuing to investigate.

[Quboid]

When testing my joystick with the Racer controller testing program (uses DX8), each axes gives values between -1000 and 1000. However, I couldn't get it to give properly smooth input and it only came in in steps of 4 - at best. These means rather than have 2001 point, it may only have 501 - surely still enough?

The steps of 4 seems strange - why do it in steps of 4 rather than just divide all the results by 4? It was be no means always 4 - depending on device and axes being tested, it was somtimes over a hundred step - I once got 3 but I couldn't recreate that. Possibly this is a feature of the testing program and DX actually does use -250 to 250. Perhaps this is a limitation of my contollers, which are both wireless so might make efforts to reduce bandwidth, so maybe the whole 2001 is available.

On the other hand, perhaps the Racer config program's returned values are completely irrevelvent - but Direct X 8 still must have a range of least 501 for the config program to pick them up, right?

What does the TrackIR need? By the looks of the config program it is 100x100 so DX has plenty enough.

[dradecki]

I now believe that the precision is based on the joystick driver being used. The DirectInput entries that are in the Registry (used for calibration, and deadzone) are DWORD's which means they are 32-bit values, which are easily capable of exceeding +/-1000 (many, many times over). This leads me to believe that values recorded in the calibration programs that are part of games simply interrupt whatever the driver returns.

So I believe a filter driver placed on top of the trackir driver could translate the cx/cy values to axis movement with the same granularity.

David Radecki

[ August 13, 2003, 01:57 PM: Message edited by: dradecki ]

[Jim]

Hello:

We currently use +/- 16K values, which is enough to be totally smooth. If you divide a circle, 360 degrees, by the total resolution then you get the actual granularity per degree. For reference, a degree is about equal to the width of your thumb at arms lenght, so that is a large unit of measure. One of us should just reference the DirectInput spec for joystick axis!

[dradecki]

Hi Jim,

I did look at the spec, and all it states is the value can be in DWORD's (32-bit). It's very open ended and is left to the device driver writer to determine the granularity, which of course equates to whatever the hardware is capable of. If your hardware is capable of up to 15-bit accuracy (+/- 16K) I believe the returned joystick axis values shouldn't be an issue. The main issue I see at this point is in user-space. Specifically if the engineers who wrote the main application use good coding practices and use the 32-bit capability of the DirectInput API or instead did things like use 8-bit values for reading the joystick axises. While this isn't likely, I have seen stranger things over the years.

David Radecki

[frentzen2000]

regarding F1 C there was a statement from
Naturalpoint that they are in contact with ISI in order to implement trackir enhanced mode.
What is here the latest status? Is this project now canceled?!?
Thanks for reply and best regards
HHF