Quote from ljuba :

Yes but I keep getting this message... what could be wrong?

D:\Igre\LFS\ghostcar>ghost_offset BL1.ljuba.ghost 1
The system cannot execute the specified program.

I don't know, it works for me.

I just tried downloading it again to check if the upload was corrupt and it's working here. Only thing I can think of is that I accidentally posted the debug build and you don't have the msvcrt libraries. But shouldn't it say so if that's the case?

I tried to bind a button to the Ctrl key, but the profiler wouldn't let me do that. Is it possible to bind the Ctrl key in the profiler or do I have to use Joy2Key?

I don't normally have the connection list open as the position list is a lot more useful when spectating. Because coloured names don't highlight properly, an arrow should be added to the position list as well.

It would make watching multiplayer replays easier too.

An aside, but still related: Why can't we change roll-bar/brake-bias/etc when we are in single player? It could make setup testing a bit easier.

The only reason I can think of is that some (crazy!) drivers would change their settings every corner while hotlapping.

While I commend your efforts and I am not suggesting that you stop what you are doing, I have some reservations about this project. These may not be 'game breakers', but they may lead to unexpected results.

1. The relationship between car behaviour/performance and its setup is a multi-dimensional nonlinear function. There is no mathematical reason to expect that the 'average' setup will result in 'average' behaviour. i.e., combining an understeering set with an oversteering set will not necessarily result in a set with less pronounced understeer & oversteer.

2. How do you 'average' different differential types? i.e., if set A has an open diff and set B has a clutch-pack diff, what is the average of these two sets?

The first point is what I see as the biggest problem. Like I said, it may not be a problem at all, but if things don't turn out like you expect, then it could be the cause.

Yes, I have noticed the same thing. I think it's because LFS changed how it reported the names of the cars via InSim in the latest official patch. Ghost doesn't understand the new names, so it defaults to a blank string.

AFAIK it's not currently possible, but it would be nice if we could run LFS from the command line and tell it to output a RAF of the fastest lap in a SPR.

As a convenience for people living in more enlightened countries, could you put the speed in km/h as well as mph?

wheel4hummer's model is the best way to do it since proxy dll's 3, 4 and 5 don't need to be coded in any special way. Is it a lot tougher to do it his way?

Why tell us? We can't help you.

This forum post might help you guys that aren't getting the ghost path.

http://www.lfsforum.net/showthread.php?t=19600

I agree that racing and hotlapping are quite different, but I also think that the skills involved are very similar. I find that the people who are fast in hotlapping will generally be fast in the race. They may lose a position or two because they don't have the racecraft of the better "racers", but they will still do very well.

We weren't supposed to be discussing ideas in this thread, it was supposed to be for finalised ideas that had been discussed and agreed on.

I suggest for a mod to move these posts (and delete this one).

Quote from wsinda :

Ease of use.

F1PerfView offers a stunning amount of stats, but when I use it I spend most of the time moving, resizing and zooming windows, instead of studying the graphs.

I was skeptical to begin with, but it's very nicely done. It improves on many of the things that I don't like about F1PerfView and has a much nicer "workflow".

One thing, are the panes resizable vertically? That is, I can drag the divider between the graphs and the track map left and right, but I can't seem to drag the divider between the graphs up and down.

This looks very similar to F1PerfView. Apart from showing setup data, does it offer any functionality that F1PerfView does not?

It's a problem with the converter. I made up a little utility to offset the Z co-ordinate in a ghost file.

I've attached it, but I'm not supporting it as it doesn't do any error checking or anything.

Great work DaveWS, I'm really liking your sounds. They make me feel like I'm driving a race car

One thing though, the MRT sounds a little too "meaty" to me.

Tyre pressure really affects top speed and acceleration in the MRT. It has such low power that tyre resistance really slows you down.

My frame rates drop dramatically in windowed mode when Anti-Aliasing is enabled. If AA is turned off the frame rate drops, but not so dramatically.

I still fail to see how a "skills test" will stop someone from acting offensively online. You're not going to stop offensive behaviour with a test. The idea is such a knee-jerk reaction with no basis in reality.

I would say that pretty much everyone knows that they shouldn't crash into other racers, some just don't care.

Instead of trying to keep the new players off the servers perhaps we could try and help them get better. We'll build a much better community working together than trying to create some sort of tiered elitist structure.

Basically there are two types of people online: 1) The sorts who want to "play nice"; and 2) Those who only want to win/drive fast and don't care about whose race they wreck.

By helping out the first type of racer we can build a better community. Nothing that has been said in this thread will stop the second type of racer from being offensive, that is just how they are. Furthermore, they will surely be able to pass the licensing system (or lessons, or whatever) and will just continue to act in the same way.

Quote from Bob Smith :

Mathematically it's near impossible to generate a curve that will stay perfectly flat through a given range yet be, well, curvy elsewhere.

Bob, you're an intelligent guy, you know what he meant. He didn't say that the curve needed to be 'perfectly flat', just that there was an 'optimum range'.

Quote from Bob Smith :

Do tyres really cool that fast? I've not seen hard data but I know rubber can hold a lot of heat (high specific heat capacity), I remember doing cool down tests on 500ml or so of boiling water at school. It takes a good few minutes before you'd want to put your hand in. Admittedly that's stationary, but that bit shouldn't be too difficult to get right (in my not not particularly knowledgable oppinion). So perhaps it's the cooling effect of moving air that is a bit weak?

Two points:

1. The specific heat of rubber tyres (between 1.2 and 1.7 J/K.g) is less than half that of water (about 4.1 J/K.g).

2. From what I can remember about the thermodynamics that I learnt at university, the effect of moving air on cooling something is VERY strong. It has to do with replacing the (warming up) air that surrounds the hot object with cool air; hence the fan inside computer cases.

My gut feeling is that the tyres do cool too slowly, but one would really need to generate a model to make sure.

I tried doing a search but haven't seen this mentioned before.

If I put the line '/view custom' in a script then it does not change to the custom view. However, the same command does work from within a textbox in the game.

Tested using U35 in both the MRT and XFG.

Is the ghost directory set to read only? I've seen programs crash when they attempt to write to a file that they can't.