(07-29-2021, 04:09 PM)checkpoint10 Wrote: Honestly I'd love to move on from Nashville and just want to make it "good enough." But I think it is important to fix this one, because ultimately the track needs to be fun, and nothing is less fun than a game-ending crash.
What about going back to the tighter version of the MBC? I mean the first one where we realized we had a problem with lateral speed. This never crashed for me. Your new tool should allow easy making of the Lp files. Only drawback is that we will have a single-file start (it is a cleaner start anyways)
(11-09-2021, 06:29 AM)samsepi0l Wrote: What about going back to the tighter version of the MBC? I mean the first one where we realized we had a problem with lateral speed. This never crashed for me. Your new tool should allow easy making of the Lp files. Only drawback is that we will have a single-file start (it is a cleaner start anyways)
I didn't have great version control so I'd have to see if I could dig it up. But I am hoping that undoing the changes systematically will help. I think the most likely culprit now is the extremely short sections I had, maybe the game gets confused when a car skips over a section from one frame to the next?
Related to my theory from my previous post, I have a math problem for you all.
Assuming that ICR2 crashes because a car completely skips over one section from one frame to the next, for example, a car is in Section 2 and then 1/30 seconds later, it is at Section 4. I am assuming ICR2 calculates position of a car 30 times a second, since that is the maximum frame rate of the game.
In my problematic version of Nashville, Section 3 was "only" 16869 500ths long (that's 2.8 feet) which I believe makes this a plausible situation. Imagine if a car happens to be near the end of Section 2 in one frame, and then travels 3 feet for the next frame, it could find itself having skipped over Section 3 directly to Section 4.
1) What is the minimum speed that a car can travel and theoretically skip over the section from one frame to the next?
2) At that minimum speed, what is the probability that a car driving through the section would skip over the section?
3) How does the probability change as the speed of the car increases?
4) What is the minimum length of a Section we should build, in order to eliminate such a possibility?
I should also point out that Section 7 is actually shorter (12487 500ths long) but it is at the exit of the chicane - perhaps is normally driven at such an angle or speed that I have never encountered a crash there.
11-09-2021, 05:29 PM (This post was last modified: 11-09-2021, 05:37 PM by samsepi0l.)
Dennis is way better at this kind of thing than me, but I will give it a try.
Quote:1) What is the minimum speed that a car can travel and theoretically skip over the section from one frame to the next?
2.8ft/5280ft = .00053 miles
in 1 frame the time of 1/30 = .033333 seconds passes
.015909 miles per second X 3600 is 57.27 MPH.
When I was driving in the original "MBC" design Nashville, I was going about 50mph through the chicane.. I bet if you watch the youtube videos you can confirm.
Quote:2) At that minimum speed, what is the probability that a car driving through the section would skip over the section?
I'm not sure but it will depend on where you were in the previous sections of the track.
Quote:3) How does the probability change as the speed of the car increases?
It will get worse because you are traveling more distance per 1/30s frame.
Quote:4) What is the minimum length of a Section we should build, in order to eliminate such a possibility?
It depends on the speed.
This made me nervous, so I checked Denver. My shortest section is 21 feet. If I do the calculation I need to be going 409mph to skip over the section (phew!)
11-10-2021, 03:26 AM (This post was last modified: 11-10-2021, 04:14 AM by Dennis.)
Here my thoughts about the problem at the rpy2lp tool, with 1/15s frames.
(10-31-2021, 03:37 AM)Dennis Wrote: There is also a limitation in the track layout. There can be only one radius change in a RPY and LP sector, so avoid short sectors. LP sector is about 11ft long,
RPY sector length depends on speed and in corners also on the lateral distance to the centerline, so the limit increases above 120mph linear (22ft at 240mph), and also at large lateral distance to the centerline in corners, like in Nashville. Adjoining straight sectors count in this case as one.
In your case, each straight section count as a separate section.
1) 57,51mph longitudinal (aka LP) speed on a straight, in a corner multiplied with (r-dlat)/r
2) at the minimum speed, the chance is also minimum
3) the chance at speed v is 1-(v[min]/v)
4) at straights 293,333papy/mph longitudinal speed, if it is part of a longer straight
else 586,666papy/mph longitudinal speed, at least 65536papy
at corners 586,666*r/(r-dlat) papy/mph longitudinal speed, at least 65536*r/(r-dlat) papy
because the rpy2lp tool uses 1/15s frames
Corner radius and dlat are signed, like in the script.
If we want to come up with a rule of thumb for minimum length of a section, should we agree on about 11 feet - which seems like the distance that a car may travel at 240 mph in 1/30 seconds? You could also shorten this on a road course where you are 100% sure that a car cannot reach the speed at which it would skip a section, but approaching 2 feet may become a problem.
My guess is that the game crashing bug is a type of warping bug that we normally see when a car suddenly "jumps" from one section to another. In the case of Nashville, instead of jumping backwards, the car is jumping forward. This would explain why shortly before a game crashes, the car warps through the track in a crazy way.
- I removed the second "extremely" short section, so the shortest section is now 10.4 feet, which theoretically you have to go 214 mph to skip across within a frame. If my theory is correct, this should eliminate the game crashes near the chicane.
- I made an edit to the uphill section to minimize the disappearing cars there.
Known bugs:
- I still haven't figured out how to prevent cars ahead from disappearing momentarily at the exit of the chicane.
(11-11-2021, 11:51 AM)checkpoint10 Wrote: Known bugs:
- I still haven't figured out how to prevent cars ahead from disappearing momentarily at the exit of the chicane.
Have you tried to set back the drawing order to standard there? Because you don't have a hill there, you don't have to deal with the see through issue.
(11-11-2021, 03:46 PM)Dennis Wrote: Have you tried to set back the drawing order to standard there? Because you don't have a hill there, you don't have to deal with the see through issue.
Yes, I have tried that, even by rebuilding the track using TRK23D and just going with the default output.
Call me crazy, but I wonder if there is something going on with the track.txt PIT numbers that is causing this. I have seen cars disappear near pit entrances/exits and I think I fixed it this way before.
