Racer v0.8.28 released!

[Ruud]

Since X-mas is still so far off. ;-)

Get it at http://www.racer.nl/download/racer0.8.28.zip (83Mb)

This one does seem close to v0.9, except perhaps for some replay enhancements.

Changes (some already posted in the v0.8.27 thread):
- Minimap was painted even if no splines were present
- SMD chase cam shadow focus is now on the car, not near the camera itself
- 3rd shadow split distance reduced to 500m for detail. Last split is now faded out.
- Large car CG offset could get some 3D objects culled incorrectly.
- 'Select car' screen gave car.ini errors which were incorrect (due to fast loading of only the first part of car.ini)
- Force feedback was passed to steering wheel even if steering was not done through that wheel
- resolution.render_aspect didn't work; fixed
- Shadowmapping tweaked again for even less Z correction
- Pacejka Player tweaking lambda values wasn't correct for MF5.2 curves
- Backfire 'likeliness' removed until we get better turbo behavior (>v0.9)
- 'grip 0.5' appeared to also reduce engine torque. Fixed (doubly stored data in the code, as if grip=1).
- Only the first 2 splits are blurred
- Autoclutch RPM now slams the clutch harder around the optimal RPM. Works to get at the optimal torque rpm
when driving off, although it is a bit harsh. Seems ok when I look at g-force graphs though.

 

[krsman3d]

That's because I'm blind. The car shader used standard_v instead of dyn_standard_v.

 

[Ruud]

Something weird is going on with the shadows on that box. Hmm, either it is the setup (racer config, track or car) or there is something fishy in the shadowmapping shaders...

It seems to be the track normals. When modifying standard_f.cg and at the end add:

outColor.rgb=IN.normal;

just after the normal outColor.rgb=lerp(...), you can check ^2 (autoexposure needs to be on). When I check the normals on the boxes (my card shows black but in fact in 'RGB at mouse' you can see negative numbers), both boxes have the same normals, while one of them should have rotated normals! So I get pure (-1,0,0) for one side, then the same for the other cube, while they are oriented 45 degrees with respect to eachother. So something is wrong with your track normals.

My guess is some kind of instancing, no X-form reset or something.

 

[Mr Whippy]

Tsk, these track creators eh Ruud ;-)

On that note, you left the SVN folder in the garage1 track folder, which is a mighty 15meg of stuff we don't need to download



Just thinking out loud about things again here:

What is the current state of thought for skies in Racer. On a night the clouds stay there and it looks a bit odd. Without clouds the sky is pitch black, which also looks a bit odd.
Also, the cloud maps don't look so hot at dusk/dawn.

Is it best to just see v0.9 as having nice skies at dusk/dawn without clouds, or generally nice skies through the main daytime? Or were you wanting to achieve more with the dynamic TOD stuff and skies for v0.9?

I'm easy about this.

Also just on clarification, what is the difference between just dropping sun.diffuse to 0 vs setting sunny to 0?

Right now we can automate the diffuse curve with TOD, but not do any automation with 'sunny', so I guess that is a feature to develop v0.9 onwards?


Lastly, I'm a total shader noob all considered, so is there much chance of that gradient over the tree thing been added for v0.9? Just let ambient fill the tree as per usual, but diffuse fill the tree from 100% to 0% over the gradient?! Or maybe 100% to 25%, not sure (test to see?!)

I had a further play with X trees and I'm fairly happy with them if we can do that I think. Happy enough to use them for my tracks anyway, and then in the future if better options are available, then we can always upgrade them (big task, but I guess that step might be years away anyway and could coincide with a bit of an update perhaps )
Just thinking, Carlswood must be about 9 years old now, and it's not looking bad all said and done!

Dave

 

[DavidI]

Tsk, these track creators eh Ruud ;-)

Without us you've got nowhere to run and nowhere to hide...

 

[DavidI]

I thought it was all good, until I noticed some strange shadows, trees, fences, then the pits building.right up close its ok though. I've only just finished fixing normals on pits, they are 'normal'

Also that autoclutch is now weird. The lambo almost has a cvt now.
'I don't like it' (the aussie's will get that)

 

[luthobu]

Don't want to be a vagina, but that track looks like this. Car used is the Alpina B6S with added CG shadarz.

Whoa.. I really don't feel like doing the shaders all over again.. is it only this track or does it apply to the others as well?

 

[Ruud]

I thought it was all good, until I noticed some strange shadows, trees, fences, then the pits building.right up close its ok though. I've only just finished fixing normals on pits, they are 'normal'

Also that autoclutch is now weird. The lambo almost has a cvt now.
'I don't like it' (the aussie's will get that)

That effect is just CSM. I got rid of it by modifying constants.cg a little (increasing some Z offsets):
const float smCor[SM_MAX_SPLITS]= { 0.00004,0.00017,0.00027,0.0012 };

The pit building btw floats a bit above the grass, giving funny shadows perhaps on the right side of that big yellow wall.

 

[FiftyOrange]

Whoa.. I really don't feel like doing the shaders all over again.. is it only this track or does it apply to the others as well?

It's everything. Racer looks terrible for me now, the lighting is f**ked beyond belief, no matter the combination of car, track, or CG/No_CG - the only thing that works half decently is the default car/track. Everything else has either masses of darkness, dark cars, stupidly thick fog, transparent body shells, contrast so stark it's almost black and white, and other problems. Using no_cg and turning shaders off completely is one of the best options to see things clearly.

CG car on CG track, I'm guessing lack of skydome, but that car also looks crap on Carlswood.

 

[Ruud]

Also that autoclutch is now weird. The lambo almost has a cvt now.
'I don't like it' (the aussie's will get that)

What don't you like about it specifically? The older method also found an rpm at which to accelerate (depending on how much throttle you gave it; the less throttle, the lower the rpm). It would stick to that rpm, holding the clutch partially up, giving a balance of car acceleration forces and engine torque.
Now, the actual max torque is found from the torque.crv and the engine really goes to the optimal rpm (max torque) for acceleration. So basically it's the same thing, only the RPM is now higher. The idea is to get maximum thrust when driving off.

Also, what's wrong with it in higher gears?

 

[Ruud]

It's everything. Racer looks terrible for me now, the lighting is f**ked beyond belief, no matter the combination of car, track, or CG/No_CG

All content needs to be updated to use Cg. Missing skydomes can hurt autoexposure. v0.8.29 (or v0.9.0RC1) updates the version checking from 050b6 to 090, so any car/track that hasn't got 'version=090' will fail to load.
Compatibility with non-Cg is just not possible, shadows won't work, klux lighting will not work, so that is not an option. Anyway, it'd be good to get all the content updated to the new level. Something that must be done, perhaps painful, but the result is much better looking stuff and no more 'my downloaded track from 2005 doesn't work'. That's why v0.9 should be the new stable version, for a new non-moving target which can be used for a few years as a baseline.

 

[Ruud]

Whoa.. I really don't feel like doing the shaders all over again.. is it only this track or does it apply to the others as well?

I don't see a problem, see attachment. Except for weird Z shadowmap trouble (why would smCor need to be tweaked for this, hm).

 

[DavidI]

The pit building btw floats a bit above the grass, giving funny shadows perhaps on the right side of that big yellow wall.

Its a new earthquake-proof concept.

No, I noticed it too, I'll add it to the todo. fix one thing break another, u know the drill.

 

[krsman3d]

Also, what's wrong with it in higher gears?

The problem is that when I set autoclutch_rpm to 0 the autoclutch is set to 3000 rpm for my particular car. That is great for fast starts, but if I am in 4th gear at 2000 rpm for example and press the throttle the revs go up to 3000. That is not normal behaviour but what could pehaps be done is to define a second autoclutch_rpm for higher gears where you don't want the clutch engaged at highers revs.
This way we would have fast takeffs and good behaviour in higher gears.

 

[DavidI]

What don't you like about it specifically? The older method also found an rpm at which to accelerate (depending on how much throttle you gave it; the less throttle, the lower the rpm). It would stick to that rpm, holding the clutch partially up, giving a balance of car acceleration forces and engine torque.
Now, the actual max torque is found from the torque.crv and the engine really goes to the optimal rpm (max torque) for acceleration. So basically it's the same thing, only the RPM is now higher. The idea is to get maximum thrust when driving off.

Also, what's wrong with it in higher gears?

Maybe the lambo needs something changed, but from still it revs to 5k rpm then launches. If your on an incline (not them again.. sorry lol) the car rolls back/forward until the 5k is hit.

it also seems to rev to 5k, drop back a bit then go thru to redline. It just doesn't act like any auto I've ever driven or seen. It's more what I'd expect from a CVT type transmission, trying to hold the revs at a determined power range & varying the drive ratio to accelerate. I know it's not, but thats what it seems like.

Maybe it's like launch control? I've never had the pleasure of driving a beast new enogh to have somethin fancy like that (yet) to know what it's like?

 

[Ruud]

The problem is that when I set autoclutch_rpm to 0 the autoclutch is set to 3000 rpm for my particular car. That is great for fast starts, but if I am in 4th gear at 2000 rpm for example and press the throttle the revs go up to 3000.

Ok that makes sense indeed. I've changed it so it only works in 1st gear, as a sort of launch control. Other gears then use the old method (also hunting the rpm, only less so).

 

[DavidI]

Ok that makes sense indeed. I've changed it so it only works in 1st gear, as a sort of launch control. Other gears then use the old method (also hunting the rpm, only less so).

Maybe it should be a car.ini switchable? even controller key/button assigned (toggles on then after one launch or after changing out of 1st resets to off?) or both? Thats more like IRL launch controls work imho.
It may well be a good way to fake CVT as well.

 

[Alexander Knoll]

Ok that makes sense indeed. I've changed it so it only works in 1st gear, as a sort of launch control. Other gears then use the old method (also hunting the rpm, only less so).

THX! That was what i want to say, but can't find the right words....

 

[Cosmo°]

The older method also found an rpm at which to accelerate (depending on how much throttle you gave it; the less throttle, the lower the rpm). It would stick to that rpm, holding the clutch partially up, giving a balance of car acceleration forces and engine torque.

That's what I'd expect/want a simple autoclutch to do actually - I agree with the others in saying that the newer iteration is a step back in functionality. One example I have is a road car that now requires 50% throttle input before it even starts of the line. If I wanted hard acceleration, I'd push the throttle pedal further down and expect the engine to rev close to it's torque peak. If I just brush the pedal for some 15% throttle, I don't want the revs to rise massively before the clutch locks though.

On top of that it also creates a slight delay in throttle response that wasn't there with the old method.

If the system is only looking for peak torque engine speed, but the actual curve is for instance one of those modern, artificially flat ones (still peaking at higher rpm, but >90% there for all the relevant rev range), it's going to waste time slipping the clutch as well.

I don't think this is something that desperately needs to be done right before v0.9.0, the old method was plenty good for our needs so far. Autoclutch should really just be an automated clutch mechanism - whether that's a substitute for when the user lacks a proper clutch controller input, or for vehicle transmissions that simply do the clutching for the driver. Autoclutch should not try to emulate an automatic transmission/torque converter or a launch control.

 

[KS95]

I like DavidI's suggestion, although if we have proper launch control in the future it might be a bit pointless.. hmm.

 

[luthobu]

I don't see a problem, see attachment. Except for weird Z shadowmap trouble (why would smCor need to be tweaked for this, hm).

Cool. Thanks for the update.