All these tracks being made for rf2 has me itching to create a track of my own. I want to do something different though. Not the usual. I'm seriously contemplating creating a track based off the streets of my hometown, The Bronx! Reason I think great tracks could come from those streets is because although people think NY is probably just flat streets the reality is it's anything but. There are "hilly" areas all around in the Bronx with decent elevation changes through out. Infact when I used to go to high school the walk from the train to my house had to be at least a good 100 foot change in elevation. Might be more than that, I don't know to be honest. lol But there are some very interesting streets and avenues that would be great for all kinds of different racing. Where would one start with something like this?
Aside from needing to know all about track creation and editing, your idea brings with it a problem: simulating all those buildings would kill the performance of the track for 99% of users. Even Silverstone, a relative desert, brings many computers and video cards to their knees. And that's assuming you were a super-hero of modelling and graphics. Read some threads here about how long it has taken the most talented track making and editing people here to do one small building like a shed or some such. A city would take weeks or months or even years.
To further your thinking, did you happen to see the August edition of Inside Sim Racing TV? They show a low-cost track release for one of the other sims; it's a version of Long Beach, with everything that is not track surface not fully modeled. All of the buildings and everything beyond the barriers is just the laser-scan data. As a counterpoint to fully modeled buildings being a performance killer, this kind of shows where you end up if you go in the other direction. The performance, framerate-wise, is said to be incredible. What I found so interesting is that the buildings rendered this way are really kind of beautiful... they look like they're invisible, but being defined by rain running down them (ala the Invisible Man), or their outlines are sketched in snowfall or something--- it's sort of ethereal and pretty. Anyway, for the same reasons (off-track stuff being hard on machines if beautifully-modeled), I was thinking about doing some of the tracks in Arizona. Since there really isn't much around (it being a desert and all), I think Silverstone levels of quality could be done while retaining that WIP level of performance. For example, have a look at this one: http://arizonamotorsportspark.com/
A professional laserscan isn't something accessible for the public, the team of iRacing is kind of unique for doing all of their scanning work in-house so hey can pull those things off. Marvin, you are probably constrained to google earth paths which usually are pretty crappy and only give you the large outlines of things. Secondly if someone can drive around the track with a good data acquisition system driving the left and right path, that is a big plus as well. Even then, the data will be far from laser scan quality and contains a lot of noise, but it is a big step beyong google earth paths... All in all, it all comes down to source material. If you just accept the fact that without a 10mm resolution laserscan you just won't get the banking and elevation just right, you will be just fine. After all, 99% of the sim community won't ever know if you are off, even if it's half a meter, as long as it drives good. Having a good base to work from, a clean road, lots of reference, really can go a long way, but then again, only so far. Big cities are a no-no with the gMotor visual engine I think. Probably the biggest reason for that is that it is built the way it is, game engines are built way differently in the way they handle assets and they are more suited for this thing. Building a city track would be a breeze in the Unreal Development Kit for example, but it would make no sense edit: Actually, I believe mainly in the USA, your government has a free DEM library for some areas, that is definitely worth looking into. It won't be the accuracy you see in iRacing but it sure as heck beats google earth any day
No they are not, there is not trouble for gMotor to handle such environment. I've seen some converted tracks from Shift Street (I believe) and while they look great, they also perform great. And they are just that, half ass conversions. Imagine what you could build from scratch just gMotor engine in mind. Since rFactor is not sandbox game, you only really need the high details on buildings next to track. Buildings behind those and more far away, can be simple boxes with appropriate tiling textures. Very easy on performance.
I wouldn't want to touch it though, decorating the world without a good instancing system and speedtree, the sheer horror... Anyway, I just found this, may be of use: http://www.youtube.com/watch?v=HtjJ91Q4kX8
To create a track with some accuracy i do the following: Build a center path in Google Earth and import to 3ds MAX with importGEO script. The kml line is not to build the track but to get the right scale and orientation. After import the kml use SASPlanet to get a Hi resolution background images from GoogleMaps or Bing. Next import all tile images using importGEO script. The background images have the size of 1024x1024 to get a clean background image in 3DSmax. After get the a full track image background can create a center line to create the track.Can use my script SplineTrackMaker. Alltitude can be added at any time and in various ways. GPS data of key points, I US can use http://www.usgs.gov/ Workflow videos:
Thanks for the replies and opinions people. I'll take it all in and go from there. Thanks for this info. Those are some great resources there. Looks daunting but starting anything from scratch always is I guess. I'll see where and how it goes.
Can't have paid much attention to the Santander London GP promo then? Sent from a mobile device using Tapatalk
If you are referring to my comment about big cities, notice the 'I think' in there With big cities, I mean Grand Theft Auto size big cities. Ever since the legendary (cough) Midtown Madness series I dream of a current generation open world sim using rF1/rF2 physics and visuals, but I have never seen anything that comes close. Having worked with UDK, and now a bit with rF2, I see fundamental differences in how both engines use and render their assets and that is why I think it's more likely a GTA sized city would not run as well in rF2 vs UDK. Not meant to talk rF2 down or anything, just an observation of mine based on impressions of two different platforms. This (http://www.youtube.com/watch?v=jDCtKDnaOuI) is looking very good but just not what I mean with big city open world type of driving
But what is the point of doing large open world /sandbox city to rF2 as it would be completely lifeless and empty? And you can't also make AI cars to drive randomly around or follow traffic rules. But still, I would not rule out such map with rF2 engine.
That's the difference between other engines and this one, decorating and bringing life to a virtual world is something those things are aimed at and do a great job at. Having AI follow rules concerning traffic and stuff is a totally different thing as well. There is this Russian driving instructor thingie, but boy that sucked... Test Drive unlimited apparently was a bust as well, great looks, crap physics.. Still, would be great...
What I am doing is kind of a compromise of all that. I am making track which is section of highways and forest roads, but does not have cities. But the major feature is that AI can drive right lane of the road and the route goes so that they will drive opposite directions.
Is that big enough for you? It's GTA:SA map in rFactor: I'm not posting it to promote its convertion but to show you what can be run in rfactor the first.
It's a start, and the problems are evident straight away sadly, but how awesome would it be to have that fully 'alive' within our favorite sim Cool ^^