Racer v0.8.31 released!

[Ruud]

Just before the weekend... v0.8.31

Get it at http://www.racer.nl/download/racer0.8.31.zip (75Mb)

Changes:
- Added racer.ini dev.note_ini_unused to check for unread car.ini parameters. See also http://www.racer.nl/tutorial/development.htm
- Color grade shader improved.
- The wind could get stuck in growing infinitely, making particles shoot away after a while
- Differential reworked, since it had a bug (it accepted ratios <1.0).
- Clutch wasn't restored after shifting when autoclutch was 0 in racer.ini and the car.
- Flare colors could dip below 0 (weird colors) when using non-klux flare color values
- Flare default color is now '40 40 40' (was '1 1 1'), see also http://www.racer.nl/tutorial/car_lights.htm
- TrackEd would save splines where lines.paint_end would always be the number of splines, even if you specified less
- TrackEd would overwrite the gfx.sky values in special.ini with generated values (when saving)
- TrackEd still used Newton 2.26, so loading any track would recreate trackcollision.bin. Upgraded to Newton 2.29.
- Flare billboarding now screen-aligned by default; see http://www.racer.nl/tutorial/car_lights.htm#flares
- A multiplayer client pressing Shift-R would get stuck waiting for a race start. Now modified to let Shift-R behave as Shift-F.
- Smoke/grass particles no longer spawn when a wheel is off the ground.
- Traction control gives a warning if a ratio is <=1.0 ; that gives serious jitters.

 

[Alex Forbin]

I got interested in surface friction, and googed up some info. I found this page http://www.physlink.com/education/askexperts/ae139.cfm with a list of materials and friction coefficients.
What I found more interesting was this statement

This statement would imply that Racer uses a very simplistic friction model?

Actually this is what we have right now, the friction coefficient for the road surface is normally one unless specified otherwise in the special.ini the tires pacejka determines the friction coefficient for the tire.

Alex Forbin

 

[boomer541]

KS95, try different codec's, mpeg-4 works best on my computer. The audio is broken and needs fixing, so you won't get any until Ruud does his thing!

 

[DavidI]

@Alex I suppose we do, we only have rubber to another material frictions. Well, at least you should, unless your continually rolling your car.

 

[KS95]

KS95, try different codec's, mpeg-4 works best on my computer. The audio is broken and needs fixing, so you won't get any until Ruud does his thing!

Cheers, will do! I'll just have to use some good music for now

 

[Ian Strom]

Or just use a different program to do the recording.

 

[KS95]

You loose fps recording.

 

[camsinny]

Ruud, can you elaborate on steer_noise? I.e. What's the function? It seems that it has no correlation to the speed at which you're travelling which I think's wrong (although I may just be using an incorrect value)

 

[Mr Whippy]

steer_noise is a fudge for things like gravel traps and rough ground... it just sends 'noise' via FF rather than relying on costly perlin road noise (physical implications on contact patches), or polygonal surface lumpiness.

I don't think it's important to see it as something we tune to be 'right'... it's just there for an effect only, to fool us.

Perlin noise (the proper road_noise) is probably better to be used now CPU's are so much faster!

Dave

 

[camsinny]

Well it'd be good for if you go off-road. Where you don't need nicely modelled noise but just a fudge - the problem is that the effect is the same at 1km/h as it is 100km/h

 

[Mr Whippy]

But for off-road, you do really want nicely modelled noise, rather than FF noise.

Those not using an FF wheel will be at a huge advantage just flying across surfaces that have the FF users struggling to keep hold of the wheel

As in off-road off-roading, or just slide off the road accidentally?

It's easier to just use road_noise. It's fast and effective... ie, it makes stiff cars really bounce around, while softer cars can take some grass and get back on track without a big penalty... so it's realistic too.

ff_noise imo, is just not worth worrying about. The last thing we want is to have it tinkered with and break something else just in time for v0.9, when ultimately it may as well actually be removed as people never set it right anyway (ie, nice feeling settings on someones machine suddenly feel like armageddon on someone elses when they touch the grass, shaking your house down one evening )


Best ignored. Use road_noise (imo) for an easier life!

Dave

 

[Alex Forbin]

I agree, steer noise is useless to me and I just don't use it on any of my tracks. It is no different than having your cellphone set to vibrate and setting it off when you run off the road. Road noise on the other hand is actually tied to the track, the perlin noise means that if you hit a bump at a certain spot in the corner, it will be there the next time you come round. You can see the effect it has on your wheels and you will notice that it directly responds to your speed, no fudge here.
The fact that steer noise is easily avoided is why I haven't made a big deal of it.

Alex Forbin

 

[camsinny]

Ok, I'll just use that. I was trying not to use too much CPU but I guess it isn't being stressed that much ATM anyway.

EDIT###
Also, is there really any point in having standard_skinned_f.cg? It has no effect on the actual processing of skinned meshes - just another shader to upkeep with no need.

 

[Ruud]

Actually this is what we have right now, the friction coefficient for the road surface is normally one unless specified otherwise in the special.ini the tires pacejka determines the friction coefficient for the tire.

Alex Forbin

Most physics engines use friction per material. Then you make up something as 2 materials brush; sometimes you hear min(u1,u2) or perhaps u1*u2. The rubber on concrete is 1.0 is quite a gross approximation; tires go well above 1.0 (1.8 on nice sport tires, dragster tires upto 4.0) even though some people in the past didn't believe going above 1.0 is possible.

 

[Ruud]

Ruud, can you elaborate on steer_noise? I.e. What's the function? It seems that it has no correlation to the speed at which you're travelling which I think's wrong (although I may just be using an incorrect value)

Steer noise is a bit hard to feel with consumer wheels. We use it on the high-end side a bit more to generate a bit of subtle noise in the wheel, to fake a bit of environment noise in the wheel. It's an added effect on top of the 'realistic' path using the wheel movements themselves. I still feel the steering wheels here feel a bit too slick though.

 

[DavidI]

Hope you had a good break Ruud? Thanks for the info, that was what I was after. What materials have a one to one friction coeffient? or what is the reference materials the physics is based on for 1 to 1?

 

[camsinny]

Steer noise is a bit hard to feel with consumer wheels. We use it on the high-end side a bit more to generate a bit of subtle noise in the wheel, to fake a bit of environment noise in the wheel. It's an added effect on top of the 'realistic' path using the wheel movements themselves. I still feel the steering wheels here feel a bit too slick though.

Brilliant, thanks. Seems I'll probably ditch that and use the proper perlin noise anyway.

 

[Alex Forbin]

Most physics engines use friction per material. Then you make up something as 2 materials brush; sometimes you hear min(u1,u2) or perhaps u1*u2. The rubber on concrete is 1.0 is quite a gross approximation; tires go well above 1.0 (1.8 on nice sport tires, dragster tires upto 4.0) even though some people in the past didn't believe going above 1.0 is possible.

I wonder how difficult it will be to find accurate data, I haven't seen much other than what was posted and it didn't pertain to tires specifically.

Alex Forbin