I think it is just the opposite : the higher detail mesh is rendered and the lower detail one is driven on. This is because the details in the road surface mesh are used for contact detection with tires and the CPU is used to do that logic. I could be wrong though.
Without any clarification from Tim then can we assume rFactor 2 cannot recreate the kind of surface detail iRacing can?
Yes, you are wrong The whole point is to have low poly visible track (but not collidable) that is easy on GPU resources and high detailed invisible layer, collidable, on which cars are actually racing over. That's how you have it in iRacing.
I haven't tried it yet, but I've seen someone recently released such a track for rF1 (called Thruxton beta 0.9 or something). Not sure if it's LS, but the author says there's a high-poly invisible mesh to drive on. It's on RFC I think.
Yes this is how it is normally done, the visual is much lower resolution than the underlying LIDAR data.
It's not laser scanned, the track surface was really just an experiment to see if it was a viable way of doing things in rFactor, it also has differing grip levels across the track surface (via the tdf file) and I was going to play with it a bit more before releasing it on rfc but it was just stagnating on my HDD so thought I'd release it before rF2 was released.
I remember way back an announcement that rF2 would be capable of using point cloud data. Can't remember if that's still true or not. Personally I would love it if there was some homebrew Lidar project out there that might be cheap for us amateurs to use.
Well rF 1 could but is that tied to the 3d mesh of that track therefore restricting how much of that point cloud can be used because of the huge amount of processing required for the end user or can that point cloud data be used to create an invisible surface layer like iRacing apparently has that allows much better use of the point cloud data to create accurate bumps etc
Yep, you are. iRacing couldn't do it before they did it. Is there some overwhelming belief that there's only one brain capable of doing things like that? And once they've done it, nobody else will ever be able to get their brain to think anywhere close? Just seems like such a weird premise for discussion.
Thats not what I was saying at all but thanks for taking the time to come back here to post the sarcastic reply
Hm, not really sure I would say it was sarcastic... More an expression of feeling perplexed is how I'd describe it. Apologies if it comes across that way.
Tim - there are few thing I would like to have clarified if possible. What exactly is the HAT file ? From what I understand, it contains the mesh that the car actually drives on. Is this just a collection of the polys from objects marked "HATtarget" in the SCN file ? I believe some filtering happens because I occasionally see messages about Gaussian elimination failures in trace files but is this smoothing or just just redundant polygon elimination ? Also, are collision targets contained in the HAT file ? I would appreciate any info you can give about this.
One of the improvements in rF2 is reduced (and configurable) track surface smoothing/filtering. Without getting into too much detail, essentially, as polys have become smaller than at the time the system was designed our coders have been able to reduce the filtering. As larger polys tend to have sharp unrealistic edges on road surfaces. So yes, you should get a more direct feel in rF2. You've always been able to run different hat surfaces vs. graphical surfaces at your leisure, but this is not something we persue.
I'm glad Michael responded. Unfortunately my knowledge with ISI stuff is still growing. If I were to line them up side by side, I currently know more about other laser scanned track tech construction - and it's problems (because of my work prior to ISI), than ISI.
Actually, Michael gave the answer to one question I was after. He said, "One of the improvements in rF2 is reduced (and configurable) track surface smoothing/filtering" which is exactly what I suspected was going on. The game (RF1) smooths the drivable surface (which is contained in the HAT file.) This means that the highly detailed surface is not exactly what cars drive on - it is filtered and smoothed by the game engine. What is not clear is how the detail of the smoothed surface compares to the surface that is rendered. For most courses the two surfaces are the same model so this means the surface driven on is less detailed than the one that is rendered. He also said that the amount of smoothing is less and configurable in the new game which is very good news ! For users that is. It means a more work is ahead for (some) track makers.
I believe it's the other way around, the smoothing algorithm adds detail rather than taking it away, it was primarily designed to smooth out any nasty transitions in tracks which use very large triangles. It won't remove any detail from a mesh which uses small triangles.
This. As I understand it. There's no detail being 'removed'. It's about blending on the tracks where they isn't enough detail (transitions being too large due to large tris).