Quote from Cue-Ball :

Hopefully a widescreen projector will be a good compromise. I'm just waiting for the day I can afford TWO widescreen projectors running in multi-mon mode.

I've used a widescreen projector quite a bit and it's ok. I run LFS at 960x600 and 80 degrees FOV. Still cuts off a lot of mirrors. Have to go to 98 degrees on the BF1 to see the mirrors. Main problem is that even with anti-aliasing, 960x600 is pretty grainy once you get close enough to give you proper immersion, since the projector has distinct boundaries between its LCD pixels.

Waiting to see what the th3go will do. I love LFS at 1280x1024 on my LCD and getting that x3 would be amazing, even if that means loosing DVI.

Then there's this thing: http://www.elumens.com/products/visionstation.html Saw that last year at E3. Needs a special lens and then software that renders the 3D distorted to match the lens. But you get great immersion

Last edited by sdether, Wed 26 Apr 2006, 14:36.

Quote from al heeley :

Erm, no, it learns how to overtake any car. It doesn't need to learn a different procedure to overtake each different car, if you get my meaning!

Ok, i didn't know if it needed to feel out the dimensions for the different types, since when i did some runs, the BF01 would clip the rear wheels of some cars, as if it didn't know where the car actually ended.

So each AI learns separately from the next and learning 1 car doesn't mean anything to learning the next car, right? And knowing how to overtake one type of car does not mean the AI knows how to overtake the next type of car. Add tracks and multiply out the combinations and you'll never be able to train everything

I hope that at least setups aren't encoded as well, because then the combinations are infinite.

Maybe we need a database of trained AIs, so everyone can concentrate on a car and a track and train 1 AI, then post it to the DB and we can download and combine the right types of AIs for any particular races. Sort of like setupfield for AI. Anyone build something like that yet?

Sure online is superior, but having good AI might improve newbies, since right now the only way to learn how to race with others is to race online which causes endless strive

Quote from joeynuggetz :

I wanna do this, how to I ensure the AI doesn't run out of gas?

Right now you can't unless you just set the race short enough. Until the AI learns to pit, they'll never have more than a tank full

I just set up a race of 4 FOX, 4 F08 and 4 BF1 on BL1 and let them run for an hour to train the AI. Seems like they're all getting better. The BF01s could not figure out how to pass without clipping the FOX's rear tires, but within 20 laps they were got it down and only crashed when trying to overtake an F08 trying to overtake a FOX. It's a lot of fun to watch too, lots of overtaking and the doppler when a BF1 blows by a FOX on the long straight is fantastic. Something to burn CPU cycles while not using the computer for anything else.

Got about 50 minutes in before the BF1s ran out of gas. At that point the FOXs were an average of 6 laps down. The others finished. I was surprised to see that the F08s (~50% left) were quite a bit more fuel efficient than the FOXs (~25%)

Can't wait for the AI to learn about pitting.

Quote from JogDive :

Do you need the source code how to bind a gaming device with direct input and use it ? i will give you my source code of JBindX and you can extend the keybind function to control more than one key. this app was only made for push to talk in comm apps

Unfortunately i'm doing the opposite. I can easily consume any directinput device. However what would be nice if I could write an App that shows up to Windows as a Human Interface Device inside of the Game Controllers control panel and pretends to be a joystick, so i can proxy a real joystick through my app and then have other apps consume my proxy joystick, allowing me to un/re-map controls on the fly.

The SendMessage approach does the trick for now, but it means i can't use any axis for my UI.

Quote from Kegetys :

You can do it by sending WM_KEYDOWN and WM_KEYUP messages to the LFS window.

Well, that's what I get for assuming that the native function uses the p/invoke behind the scenes and writing off that solution without trying. Switched my code to using SendMessage with WM_KEYDOWN and WM_KEYUP and that works like a charm. Thanks again!

Quote from bal00 :

Or you could check out PPJoy (http://www.geocities.com/deonvdw/Docs/PPJoyIntro.htm). It's a virtual joystick driver than can take inputs from other apps.

I looked at that briefly yesterday. Since it was talking about emulating a gameport device, i didn't know if it would be a valid device for DirectInput, so i put that approach on the back-burner

Quote from Kegetys :

You can do it by sending WM_KEYDOWN and WM_KEYUP messages to the LFS window.

Thanks. I'll give that a try. I figured SendKeys was doing that behind the scenes, but I'll try WM_KEYDOWN & WM_KEYUP explicitly through P/Invoke.

Quote from Scawen :

I wouldn't want programs to just exit saying they don't know the protocol, or something like that, for such a tiny change as this.

Not a problem. Just wanted to make sure it was supposed to be this way. Thanks!

I've been trying something for the past two days and I've exhausted every means I can think of, so maybe someone here has come across this as well.

Basically I want to selectively let my paddle shifters control up/down shift or control my insim app. Both LFS and my app capture the DirectInput device, but I have no way to tell LFS to not listen to the shifters when I don't want it to. My solution was to bind upshift/downshift to keys, let my app exclusively handle the paddle shifters and send key presses to LFS when I want them to mean upshift/downshift.

I figured, i'd use SCH packets to send the keypresses. But apparently no such luck.. While I can send keys for things like changing view, the key presses for controls don't seem to get processed. Don't know if it's a bug, or just not an intended use for SCH... So I tried the win32 API for sendkeys to the active app. Again nothing.

Googling around I thought i found the answer: LFS is using directinput, so it doesn't listen to external keypresses. Using some code I found at gamedev, i created a program that would inject key presses into DirectInput. Worked fine with the sample windows directinput app. But when I tried it with LFS, it never acquired the device to inject.

Anyone have any suggestions? Any pointers to code for creating a virtual device driver so i can pretend to be my own DirectInput device?

cheers,
sdether

Just tried the T patch. InSimVer is still 2. Is it supposed to be? I know the changes are not affecting packet size, so should InSimVer only increment when packets change in size?

cheers,
sdether

I really like the TC on the BF1. Part because it saves my ass and part because the stutter it creates in the engine is a sound i've heard so many times during F1 races and makes it more real for me.

Tried the FZ50 with TC and didn't like it because it didn't work with the way I drive that car. With the FZ50, i usually come up on a hard turn, turn the wheel in low gear and make the back step out and throttle steer it through. But with the TC it wants to help me and stops the slide widening my turning radius.

This patch does hateful things to my productivity. Just spent another hour trying out different cars. The LX6 and RA are an absolute pleasure now. So tossable and happy to be throttle-steered. What fun.

I just love the sound of an engine at 15k RPM plus. I think the BF1 sound is another testament to LFSs sound modelling instead of using samples. Sure, it isn't quite as visceral as the real thing, but it behaves properly. I found myself just doing up and down shift practice to hear that high pitch symphony so familiar from watching F1.

If there is one thing that I miss in the sound model is that it currently doesn't seem to include the transmission. When I did my engine swap, I removed all the sound deadening between my rear seats and the engine and now on top of the engine sound, that dips and rises with gear changes i have the constantly rising whine of transmission. It's the closests i come to high pitch with my engine redline of only 7k rpm.

While on the subject of sound, while the turbo whine is in there and very reminiscient of my MR2 Turbo days, one thing I noticed with my sound deadening removed is that even on naturally aspirated cars you can hear the sucking sound of air rushing through the intake when you put the pedal down. It's not just a Turbo noise and would be nice to find in a future version of LFS.

Quote from JTbo :

Grip has gone towards arcade, imo.

It has more grip but it still slides so easily, tires don't bite as in reallife, now it is practically impossible to spin with XR GT and such car in reallife is surely spinnable and when it starts to spin at corner entry it is nothing like in lfs, it is much more brutal and faster episode.

I have to disagree. My daily driver is a 2300lb car with 210bhp and it's got a 36/64 weight distribution. So at best it should be more tail happy than the XR GT. However, when i try a hard turn in second, the traction usually chokes the engine that I plow more than understeer. The only way I can easily make the rear step out is to do start in 1st and slip the clutch into 2nd to stay at the upper range of the RPMs. (Or come down a straight too fast and unbalance the car slowing too fast going into a sweeper -- that felt nasty)

Anyway, the XR GT is similar to my car in power/weight but better balanced. In pre-S it would step out at low speeds in 1st and second without any effort. In S it now has a tendency to plow a bit if you do the turn starting in the low-to-middle range of RPMs in 2nd.

Tried my standards test setup of FXR on Aston Cadet Reverse. On the first try, i beat my usual average by a second and within a couple of laps i was 3 seconds up. Main difference was being able to downshift more agressively (without the the engine braking being stronger than traction and causing a spin) and coming out of turns more agressively. IMHO, a lot more like slicks should grip. My only experience with slicks is in a 914, but i can tell you it that that traction is more than capable at overcoming any engine braking that the transplanted flat-6 could muster. But even my Falken Azenis give me more traction at autocross than the FXR used to provide in the slow turns.

I like the patch.

Should Server Serial be incremented to 3? I was revving my lib for the S patch and hit the dedicated server, which reported a serial of 2.

cheers,
sdether

Quote from filur :

Different indenting is enough to make me loathe a piece of code. :zombie:

The indentation wars are the worst. When i'm using emacs, I can hit some keys and have the whole thing re-indented. But i have to refrain myself from the temptation, because it really messes with revision control, as it sees every line as changed which will come back to bite you when you try to dig for when a change entered the system.

In Visual Studio, our team has just agreed to all suffer to the MS auto-formatting. Sure it's not exactly the way any of us would do it, but at least its a consistent compromise. Although I still use 2 spaces for indent when everyone else uses 4

Quote from filur :

Regular expressions are by far my favorite tool.

Same here. It's the reason i got into perl in first place. I admit that regex is a pain in almost every other language, but it just part of the landscape in perl.

Most of my programming now revolves around building object models, but when someone hands me a text file and needs the data to be converted into something else, or have something extracted from it, I always whip up a perl script to do the regex and write the data out in a format that's easier to parse for my other code.

Last edited by sdether, Sat 15 Apr 2006, 17:04.

Quote from Greboth :

Laguna seca? thats got some serious elevations and drops. Can support fast cars. Although it hasnt got that "home town" track feel.

Ah, Laguna Seca. There's a track that's got everything. The corkscrew is the ultimate "oh shit" elevation change in any track I've seen. While it's a bit technical, it can definitely support fast cars. I was there 2 years ago for the Monterey Historics, when Ferrari was the marque of distinction and they had Micheal's 2003 car there trying to break the track record. Didn't quite do it.. a couple of hundredth off, but seeing a car that fast maneuver the track was hair raising even from the stands.

What is your programming languages of choice?
C#

What programming languages do you know?
C#/C/C++, perl, php, java, lisp

What do you use the most as your text editor for programming?
Emacs if not C#, Visual Studio, if C#

What IDE do you use, if any, and if you don't do you have problems with them?
Visual Studio w/ C#, no IDE otherwise

What is your compiler of choice, for your primary programming language?
Default .NET compiler, or mono if on non-MS platforms

What libarys do you recomend for your primary programming language?
.NET has more libraries in its core than i'm likely to use, but i use log4net & NUnit for most of my projects as well. Really just dev support not as a feature of the final product.

I've got to second the request for larger autocross area. I do most of my autocross at Qualcomm Stadium in San Diego, CA and even their smallest lot is larger than either of the autoparks (it helps that the large one is an odd shape). I tried to set up the autox for my next autox in LFS and ended up creating it at 1/2 scale, and used the MRT on it instead of the street cars.

http://maps.google.com/maps?f= ... 0.007631,0.009881&t=h

In the satellite image you can even see the groove from a recent autocross.

The west lot it about 700m x 250m. The two east lots are about 250m x 400m each.

Quote from Dygear :

Now the program runs, but when ever a user sends a message, the program crashes when trying to right to textBox2 thanks to the multithreaded problem. Do you know of a quick way of turning off, or just plain getting around this issue?

Your problem is that the event gets fired on a thread different than the thread in control of the UI and only the UI thread can change UI elements. Here's details on the issue:

http://www.codeproject.com/csharp/begininvoke.asp

Quote from Dygear :

I really don't think that Traction Control is all that bad, but that's just me. ABS is really quite unnessacery, locking tires can be worked away easily with a good setup. But think of the mousers. Its only right to have TC from them atlest. Speaking of Outsim, where can I read up on that? I would love to get information such as revs and output it to a GUI program. Also speed, throttle and break would be awsome.

Outsim is briefly described in the main readme, but it doesn't really provide what you are looking for. It's basically physics data feed to be used by motion simulator rigs, such as the Force Dynamics 301. There's been talk before about creating a new output format that feeds data for feeding physical gauges, but I don't know if that's going anywhere or where it would stand in relation to all the other things that are planned.