Acceleration

It's pretty self-explanatory. Have a look at the CC lines I did for Valencia and Sepang, you should see the differences in the affected corners.

=====================================================


Intel NUC 8i3, 8GB RAM, MS Sidewinder Wheel

TomMK Wrote:
-------------------------------------------------------
> It's pretty self-explanatory. Have a look at the
> CC lines I did for Valencia and Sepang, you should
> see the differences in the affected corners.


@RHill: ...Especially so, if you really know how to build a good CC-line.

---

My workthread - [www.grandprixgames.org]
Full of classic F1/non-F1 track layouts

My blog about F1 performance analysis - [thef1formbook.wordpress.com]

I just tried making the entry to the final corner at Melbourne loads tighter and instantly there were more overtakes on the pit straight.

When the ai follow me through there now, they get within .2 on the main straight and look for an overtake, where before it would always be .5 if I got a good exit.

So making the corner entry super tight seems to work

I would be curious if the modifications result in an overall improvement in laptime, or not.

Exits should certainly be better, but I'm wondering if you guys now overbrake the entry and the middle.

(Looking at how curious your problems are, RHill, I'm still inclined you are just simply too good at controlling traction on the exit of the corners compared to the average player. That, or CC-lines which fit 'normal' mod physics are simply inadequate for some of the more 'extreme' ones.)

---

My workthread - [www.grandprixgames.org]
Full of classic F1/non-F1 track layouts

My blog about F1 performance analysis - [thef1formbook.wordpress.com]

I've now tested with ofiginal and modified lines and the ai certainly stay closer on the straight.

The lap times are only decreased a tiny bit, but the ai are far better matched with the player onto the main straight. I'm testing Melbourne with a relatively short pit straight, so with longer straights the ai will def have more opportunity to pass the player.

The trick seems to be to ignore the logical line you would take and make the ai enter the turn as quickly and tightly as possible, to maintain speed onto the straight vs the player

@Atticus - Very good point, so I did a test: multiple quickraces on Pro level. For Sepang, my CC line (inspired by yours, thanks!!) is definitely faster than the original. It seems the technique of opening up the exits is certainly not detrimental:

Average AI laptime - Original CC Line = 1:37.816
Average AI laptime - TomMK's CC Line = 1:35.520

Here's an observation: this is part of a larger pattern. As good as GP4 is, the developers made a bit of a hash of certain aspects (probably in a rush to hit a release deadline). CC lines are - as we've demonstrated here - one of those aspects.

I also think the original Magic Data is confusing and nonsensical. And the original TV cameras are not in line with the real-life ones in a lot of cases.

I've thought for a while now that we need a concerted effort to "tidy up" the stock version of GP4 so it's a just a better overall game. But now we're getting into a much larger project...

=====================================================


Intel NUC 8i3, 8GB RAM, MS Sidewinder Wheel

Maybe the first step is to redo all the 'stock' cc lines but only the corners that lead to long straights?

TomMK Wrote:
-------------------------------------------------------
> @Atticus - Very good point, so I did a test:
> multiple quickraces on Pro level. For Sepang, my
> CC line (inspired by yours, thanks!!) is
> definitely faster than the original. It seems the
> technique of opening up the exits is certainly not
> detrimental:
>
> Average AI laptime - Original CC Line = 1:37.816
> Average AI laptime - TomMK's CC Line = 1:35.520
>
> Here's an observation: this is part of a larger
> pattern. As good as GP4 is, the developers made a
> bit of a hash of certain aspects (probably in a
> rush to hit a release deadline). CC lines are - as
> we've demonstrated here - one of those aspects.
>
> I also think the original Magic Data is confusing
> and nonsensical. And the original TV cameras are
> not in line with the real-life ones in a lot of
> cases.
>
> I've thought for a while now that we need a
> concerted effort to "tidy up" the stock version of
> GP4 so it's a just a better overall game. But now
> we're getting into a much larger project...

No problem.

If you modified all the other corners as well they may make it quicker overall, but I'm pretty sure squareing corners out that much is slower for those particular corners. That's what I was trying to suggest to RHill - without much success, I fear.

And yeah, your observation of the CC-lines and track cams are right - precisely that's why I've begun to provide new lines and cams for some of the original tracks, although the project is not completed by any means. I don't know if RHill tried my version of Sepang (he most probably didn't), looking at it, I think his problem in the last corner should go away even with my line, but I'm not sure about the penultimate turn.

---

My workthread - [www.grandprixgames.org]
Full of classic F1/non-F1 track layouts

My blog about F1 performance analysis - [thef1formbook.wordpress.com]



Edited 1 time(s). Last edit at 01/24/2016 11:52AM by Atticus..

Making the entry tighter is definitely making the ai stay closer on the straights. The ai traction loss is masked when the corner is squared off and more speed is carried to the straight.

There is no way the corner is slower because of the cc line change, because the laptime improved very marginally and it was the only corner changed.

It seems less about squaring the corner and more about getting the ai to get their turn in done and dusted early to make for a straighter line, rather than a smooth ark

I've redone a few 2001 cclines to see how much scope for improvement there is. I think Sepang was a bit of an extreme example as I haven't been able to make the same kind of gains elsewhere. Here are the average improvements to the fastest laptime I've made so far:

2001 Melbourne = -0.468s
2001 Sepang = -2.296s
2001 Interlagos = -0.358s
2001 Barcelona = -0.641s

So a worthwhile exercise I think, to do all of them at some point.

=====================================================


Intel NUC 8i3, 8GB RAM, MS Sidewinder Wheel

TomMK Wrote:
-------------------------------------------------------
> I've redone a few 2001 cclines to see how much
> scope for improvement there is. I think Sepang was
> a bit of an extreme example as I haven't been able
> to make the same kind of gains elsewhere. Here are
> the average improvements to the fastest laptime
> I've made so far:
>
> 2001 Melbourne = -0.468s
> 2001 Sepang = -2.296s
> 2001 Interlagos = -0.358s
> 2001 Barcelona = -0.641s
>
> So a worthwhile exercise I think, to do all of
> them at some point.

I thought Melbourne would be bigger. When I did it, I've found the original to be one of the messiest of all. (Whereas others, e. g. the A1-Ring, were almost fine as they were.)

Did you re-do the whole thing or just the low-speed hairpins as RHill suggested?

---

My workthread - [www.grandprixgames.org]
Full of classic F1/non-F1 track layouts

My blog about F1 performance analysis - [thef1formbook.wordpress.com]

Random question guys but bare with me...

Gp4 was released at a time when a keyboard was the joystick of choice right? The ai also accelerate in the same way as a keyboard player would (less feathering of the throttle).

So my question is what calibration range would most replicate the ai keyboard style acceleration, but using a pedal set? Basically when you calibrate the accelerator, what range would most closely replicate the ai?

My thought process is that the ai are more cautious on corner exit and later on the throttle because they have less 'range', but this would be no disadvantage if the player used a keyboard, but a wheel allows a greater 'range' of control. I just wanted to try limiting the throttle movement, but what calibration would full throttle be to best replicate the ai throttle habits

Amazing Topic, Guys. I'm sure most of the Players find the acceleration of the AI out of tight Corners too slow.

This is an amazing Achivement for the Game. Thank you so much for the Contribution.


Regarding the Laptimes: In order to stay close to the original Laptimes, would it make sense if you lower the Track Grip
in the Magic Data after improving the CC Lines?


Many Greetings and a lot of Respect.