WIP 2000 Mazda RX-7 Type RZ

[Stereo]

Still haven't managed to replicate the problem :/ There's something very odd going on when the collision box hits the grass on Mashup though, I wonder if I can catch a video. It seems like a positive feedback loop, hits the ground and bounces higher each time until it flips over.

 

[Mr Whippy]

Didn't get the whole list checked off, but my vacation's coming to an end and I did hit the major points, so I thought I'd clean it up for distribution.

Mazda RX-7 (3MB)
window2_f.cg shader
Put the shader in racer/data/renderer/shaders/ if you haven't got it, the car in racer/data/cars/.

If you find something wrong that's not listed, let me know. Also feel free to advise changes to the .ini.

Wasn't there a way to reference CG shaders in the car folder.

This car is immediately gonna break when Ruud updates Racer, which isn't ideal at all.

Not looked, but curious what the difference is in this shader? I've had great results with the window shader here and it does everything I need it to.

Cheers

Dave

 

[Mr Whippy]

Still haven't managed to replicate the problem :/ There's something very odd going on when the collision box hits the grass on Mashup though, I wonder if I can catch a video. It seems like a positive feedback loop, hits the ground and bounces higher each time until it flips over.

Not using a newton mesh for the car body?

I think the console command is something like physmesh on or show physmesh or something... then set Racer realtime scale to about 0.25 and go for a drive, everything can be seen nicely in slow motion but with the realtime physics in that way.


Dave

 

[skybh]

Not using a newton mesh for the car body?

I think the console command is something like physmesh on or show physmesh or something... then set Racer realtime scale to about 0.25 and go for a drive, everything can be seen nicely in slow motion but with the realtime physics in that way.


Dave

The console command is "wireframephys on|off"

 

[Stereo]

It does have a Newton collision mesh (collision_2012.dof I think)
wireframephys on, realtime 0.25.
Looks like maybe the rotational inertia numbers are too low? When it takes off, if one corner clips, it sends it spinning around the front-back axis much faster than seems right. I'm only going 50-60km/h so I wouldn't expect enough energy to do multiple barrel rolls in the air.
EDIT:

An example Ferrari of 1280 kg has a X inertia value of 1950. Lower weight cars have lower inertiae. A good empirical method that is in use for Sports/Formula style car is the following:
Gather information on sprung mass ('preLoad'), wheelbase (front to rear) and front distance to CG ('attF').
Iyy = body.inertia.x = 0.8*(frontPreload*(attF^2) + rearPreload*((wheelBase-attF)^2))
Izz = body.inertia.y = Iyy*1.2
Ixx = body.inertia.z = minimally Izz*0.5 (difficult to obtain, but at least half of Izz).

Gonna try to apply this info from Racer.nl, see how it shifts the values.
EDIT 2: body.inertia.z should be 2.5 times larger by this measure (from 300 to at least 750). This improves the gross characteristics but it's still much too elastic. Will investigate more.

I'm using the same glass shader on multiple vehicles so having it separate works out for me (can reference it just by copying the car.shd vf_ sections) I'll make a note to put it in the car folder to allow a standalone version though. For a while it was also better to have it crash immediately because shaders were changing from version to version and sometimes it'd be incompatible in weird ways.

The major difference is that it uses fresnel to affect opacity, at 90 degrees it's gonna be 100% opaque if the reflections are 100%. Just based this on the fact that if all light is reflected, none is going to be transmitted through from behind to start with. From what I can tell this makes sense with irl glass too.

 

[Mr Whippy]

Yep, I always applied a fresnel 'copy' in my scanline/RT 3DS Max materials alongside the reflective properties, so the glass in theory was black at a steep angle, but replaced with reflection...

Makes sense for our glass too these days... I didn't realise it didn't do that

Dave

 

[Alexander Knoll]

Still haven't managed to replicate the problem :/ There's something very odd going on when the collision box hits the grass on Mashup though, I wonder if I can catch a video. It seems like a positive feedback loop, hits the ground and bounces higher each time until it flips over.

hmm...wired...to day i tried to replicate it and i fly to the moon but no racer crash...
maybe the crash was a fault from my sys....sorry...

 

[foxo]

If Stereo doesn't mind, I could upload my small improvement )

 

[foxo]

Well since i've got plenty of free time, I thought that i could help more
I didn't thought that it would be so hard to find some stock Potenza S07 images, but unfortunately google lost and i won. I don't wanna show the reference picture because it was a real challenge to make this texture. The real challenge would be the sidewall.

 

[johnthelemon]

The man. You are it.

 

[Stereo]

If Stereo doesn't mind, I could upload my small improvement )

Go for it, I'll try to do some stuff myself but I don't have huge amounts of time.

 

[Stereo]

Found the normalmap plugin, got a start on the tire texture. Still need to tweak the "heights" somewhat but just the basics are giving me a good indication that it's worth it.

Need to remap the tires on the left side of the car cause their sidewall's backwards... wasn't obvious with the old blurry texture. For that matter all the tires could do with remapping. Anyway, this is kinda a precursor to doing blur models, cause I want the texture to be finalized before I tackle that.

 

[Stereo]

Progress on the things I'd have liked to improve was approximately 0, so I'm just going to call it final for now with the new wheels and ini.

http://nmurdoch.ca/7z/mazda_rx7-2013-11-05.7z
As mentioned in README.txt you can improve the quality of the wheel shaders by copying some code to utils.cg. While you're at it, dyn_standard_bump_reflect_f.cg and dyn_standard_reflect_f.cg would benefit from the same code - just find+replace 'texCUBElod' with 'texCUBElod_soft' to get a slightly more expensive but nicer looking reflection.

 

[foxo]

The README file is empty

 

[Alexander Knoll]

THX dude...testing^^

 

[Stereo]

The README file is empty

Hmm that's the trouble with testing it over several versions (was trying to find a version that would actually report shader errors in qlog)... I saved readme to a different version than I uploaded.
There's nothing too important in the readme anyway, just pointing out cg/utils_cg_update.txt (contents to be copied to utils.cg) and mentioning the bugs I know about. Uploaded a new version of the archive including it.

 

[Mr Whippy]

utils.cg injection is a bit of a pain because utils.cg might change, as might other stuff.

I was making a batch file to change a lot of Racer files to do nice things with Racer skies, but now we have a new version again the files the batch swaps around are changed, so they kinda won't work without more work again from the author at every update (ie, maintenance)

I wonder if it's possible to dump the soft envmap stuff in a car/cg file, that the car.shd can then look at instead?

Pros and cons each way but maybe a bit more elegant?

Dave

 

[Stereo]

Using custom shaders (to reference the new code) so really it needs changes for updates regularly anyway.

I have it in utils.cg cause I can change the default cg files and affect every vehicle more easily. Would be nicer to get it included in Racer, or to override cubemapbias directly. Haven't tried doing that.

I probably need to update the cg code to include the new LogZ2 line (and maybe extinction? Haven't looked at what changed there) anyway, though. Currently I have 2 shaders included:
- different windows shader (I think this is mostly to make the frame around the glass look ok due to how the texture is laid out)
- directional reflection shader (using this to stop the wheels from reflecting quite so badly - lets me turn up specular+reflections on them to realistic amounts)

 

[Cosmo°]

Great to see the update released. I'd like to point out that the qlog errors showing up in Racer v0.9.0 RC8 are due to the changes from RC5, which this release was developed for in the beginning of the year. I'll gladly provide a revised version once I feel that the current suspension related bugs have been ironed out and I have a better understanding of the friction circle methods used. Even so, I tested the car in RC8 and it works and feels fine for the most part, so it shouldn't stop any of you enjoying the release.

 

[Mr Whippy]

The new extinction/sky stuff is all in an intermediate state right now it seems.

Try get an extinction colour other than minty green hehe.

Hopefully it's all be fixed soon, but for now I'd still target RC5c (imo) because it's the last robust and solid version we've had for a long time.

Not to say the newer ones don't bring good stuff to the table, but it's often with other stuff that makes it not as ideal overall for a day to day "Racer" platform. Ie, this latest version with bump stops broken and a limbo extinction/sky system.


I agree, would be good for Ruud to add the required soft cube envmap and stuff into the utils.cg...

I'd like to have a play with making a modular shader where you generally have all the key controls added and then you can trim out the stuff you don't want. I guess it'd be a highly commented pair of shader files but fundamental stuff to do with materials/lighting generally stays consistent from shader to shader.
All we really should say is which things we want to control and how (ie, floats, LUT's etc)

Dave