Online Racing Championship

[Randy G]

Why is Gregor and Tony working on this?

http://www.siroccoracing.com/racing/public/page?id=1

No time to drop a line once in a while here, but plenty of time for this?

[MickeyMouse]

Thanks for pointing this out. I've never heard of it before. Very interesting indeed....

[JPS]

First I've seen of it, too.

I am glad they are keeping themselves busy (and hopefully paid) with something they have always been very interested in.

ORC appears to be in a very primitive state.  But perhaps the underlying physics are far more advanced than the graphics/sound/UI.  One of the news posts says the shadows are looking good, but I disagree; shadows under that F1 car in the video are WAY too dark and solid looking. Vid was posted March 3, 2010. Hopefully more progress since then. Lighting is very important to get right before proceeding with any texturing, shaders, etc.

"We are hoping to have an open beta in 2010 to test that everything works correctly on all sensible hardware and weed out those last few (hopefully!) bugs."

DeJa Vu   I'll try not to get too interested in it until something is available to download this time.  Broke a few teeth last time; not looking to be taken on that ride a second time.    It's already May in a few days; 2010 is approaching half-way point.

Wonder if they are going with DirectX 10 (doubt they will base it on 11 so soon).

[JPS]

@ Randy G

Just noticed the old Pontiac link in your sig.
I'm suprised the website is still up.  End of an era, you know.  They're up in Heaven playing cards with Oldsmobile now.  Sad so many good ones had to go so that lower lifeforms could survive.

[stenyak]

They seem to be using Ogre3D (open source graphics library). Ogre3D allows to use D3D or OpenGL, but they could very well have decide to restrict to D3D for ease of development.

Also, they use Hikari, which is a library for Ogre3D which allows to embed flash. This probably means that the UI is made in flash, while the game itself is rendered with Ogre (as I say, either d3d or ogl are possible).

They also use Awesomium, which allows to embed websites into applications, as in, embedding a browser with a very light interface.

Looks interesting, hope it's possible to betatest it soon :-D

[ashman]

Hi Folks,

Was looking at my website stats and noticed there are a few clicks coming from the Westies forums.

I am indeed using Ogre3d, and I will probably stick to D3D for the minute to simplify things and try to get something released as quick as possible. 

I am also using flash (flex) for my UI - <a href="http://www.youtube.com/v/kmdFbPterKY" target="_blank">http://www.youtube.com/v/kmdFbPterKY</a>

@JPS The shadows are programmer art and haven't been blessed by the hand of the artist, therefore they are perhaps a little dark, this can be very easily tuned.

It'll be a while yet until there is a public beta, things are taking a bit longer than I'd hope, I am working 10-12 hours around 6 days a week to get it out, if there is one person in this world who wants to see a beta its me! .  Just relax, pop in every so often to see how it's progressing, if it comes out and you find that it's fun, play it   No need to get worked up over a racing sim imo.

If you want to see the horrible nitty-gritty progress then you can follow me on http://twitter.com/AshMcConnell

All the best,
Ash

[MickeyMouse]

Thanks for the info. How are the West's and Gregor involved in the project?

[ashman]

Thanks for the info. How are the West's and Gregor involved in the project?

Gregor and Tony are friends that are helping out the project when they can.  There is no need for Gregor to be working full time (or anywhere near full time) on the project, the really hard work was in creating the physics library, which was pretty much completed 3-4 years ago.  We have tweaked it to make it faster by some optimization, but mostly using SSE.  Any work Gregor does on the physics engine for ORC will feed directly into RL, so it's not a case of me stealing him away from RL .  He does offer advice and help when I'm testing things, but i'm not filling his time.

Tony has been finding it hard to get any time to work on ORC yet as he is very busy on his other projects at the minute.  All the 2d art (Menus / Website) has been completed by my wife who is a graphic designer.

All the best,
Ash

[MickeyMouse]

Gregor and Tony are friends that are helping out the project when they can.  There is no need for Gregor to be working full time (or anywhere near full time) on the project, the really hard work was in creating the physics library, which was pretty much completed 3-4 years ago.  We have tweaked it to make it faster by some optimization, but mostly using SSE.

That's interesting, did the SSE work? Most of the post's I've seen on Gamedev about implementing SSE for anything other than image processing seem to be rather negative because of the alignment issues.
There is one thing I've always been curious about, when computing suspension movements is it best to take an analytic approach (look ups from offline simulation), an explicit constraint solver approach (Danzig, Lemke, Baraff, etc.), or iterative approach (PGS), do you know how Gregor does it?

Any work Gregor does on the physics engine for ORC will feed directly into RL, so it's not a case of me stealing him away from RL .  He does offer advice and help when I'm testing things, but i'm not filling his time.

I (and I'm sure most here) wasn't aware there was anything to steal him away from, since RL appears to be all but dead. 

[ashman]

Hi Mickey,

Yeah, I'd heard conflicting reports about SSE too.  We started by using SSE (intrinsics) with the Vectors class and it seemed to be around 20% faster (when profiling the amount of time the physics took as a whole).  We added SSE for the matrices and it slowed down.  Not sure what the reason was, but we've kept it in the Vector class now.  I've put the physics on another thread now which helps a lot with overall game speed, assuming most people have dual core (82% according to http://store.steampowered.com/hwsurvey/ )

We were worried about accuracy when moving from doubles to floats (kinda required for SSE), but it seems to work absolutely fine.  We'll keep an aye on it though in case it starts to get a bit jaggy on big tracks.

I suspect Gregor is taking an iterative approach (just from the params I need to set for the physics engine), but hopefully he'll be able to tell you in better detail, I'll try to ask him tonight, he is offline atm.

*edit* He says he implemented PCG (preconditioned conjugate gradient)

All the best,
Ash