Kurt "Yoda" Kalbfleisch
Scuderia Sprezzatura
Well, I'm there at last. When I started horsing around with AC in April, I thought I'd never be fast enough to run with the AI, even at 80%. Practice, everyone said. Practice will make you fast. And I thought setups would do it, so I focused on setting up the car. And I got frustrated, because nothing I did seemed to make the car go faster. I tried all sorts of setups, too. Downloaded miracle setups, expecting my lap times would go from abysmal to lightning in an afternoon.
Of course, I practiced. But mainly with an eye to improving my setups, because that HAD TO BE the secret to going fast, right? Had to be.
And I did get faster. Slightly. A little bit at a time.
I compared my lap times with real race drivers in similar cars, as posted on the WEC results pages -- and finally accepted that I couldn't ever measure up. No way. I'll never be that fast.
Inevitably, I started playing around with AI. Hot lapping was fun, but frustrating, and I thought maybe I'd have more fun if there were other cars on the track. The prospect of racing with cars going faster than me was daunting, though. There was no way I'd ever be able to race with those AI cars, even at 80%.
And then I realized that, by creating my own AC career, I could adjust the AI to BELOW 80%, and hey, that would be a great way to inject some fun into the simulation. I did a little research, found a racing series that looked like it would be fun to (sort of) emulate, and I built it: Corse Clienti's Ferrari Challenge Europe. Because I could include all the tracks from the '15 season. I adjusted the AI percentages for all the opponents to 40%. As a starting point. Because, you know, I'm slow.
In the practice session at Monza, I quickly discovered that the AI cars would brake too early for each corner, and I learned to anticipate their braking while watching for my own braking point. I learned to read the other car's line and to set up the pass on a less-than-ideal line.
I ended up not saving that first practice. I'd discovered that I was faster than all the AI cars, so I adjusted the opponent.ini file to set the AI cars at 60%, and went back to it. This time, the AI didn't brake as early, but they were still easy to out brake. I got better at cornering off line. There were a couple times when I thought I'd been held up by traffic and still drove a faster lap than I'd ever been able to before. What's happening here?
I also discovered that the AI cars would bunch up and run in groups, making it impossible to pass them all at once and that forced me to occasionally run in the middle of a bunch of slower cars. It occurred to me that if I adjusted the AI opponents individually, it might help break up that bunching, so I went to the Corse Clienti website and compared the drivers in the Coppa Shell series, ranking them on a scale of 1-9 according to their average performance in the season so far. I was getting faster, too, and the 60% wasn't really cutting it any more, so I subtracted each driver's ranking from 70%, giving me AI strengths from 61-69%.
And of course, I noticed the actual lap times that the real world Coppa Shell drivers were turning in. There was no way I'd ever be that fast, I thought. No way. By Saturday afternoon, I was within about five seconds of those real world times, and as far as I was concerned, that was as good as I'd ever get.
My setups? I wasn't putting much effort into them. Default, mainly. Soften the front springs a little.
By Sunday morning, it was clear that even the 69% AI drivers weren't fast enough to offer me much competition. I still hadn't saved that first practice at Monza, though. My efforts were focused on dialing in the AI. The problem was that every time I bumped up the AI percentage and drove some more, my lap times would come down, and I'd have to adjust them some more.
Because, you know, I kept getting faster.
Late Sunday morning, I took the plunge and set the range for the AI drivers from 81-89%. I didn't think I'd be able to keep up, but I was clearly much faster than the AI had been, so I figured I'd give it a try.
That's when the magic happened.
The AI cars were now putting in lap times that averaged the same as their real world counterparts in the Coppa Shell series. AND SO WAS I.
When I was just hot lapping at Monza in the Ferrari 458 GT2, I'd been consistently turning in 1:59s, but now? 1:52s. Not alien fast, to be sure, but that's what the real Coppa Shell drivers are doing. Not the pack. The fast guys, Duyver and Prinoth and yeah, those guys.
I took the pole for Race 1 on Sunday afternoon, and was in second place two laps in when I got called away from the computer and had to DNF. On Monday, I took the pole again for Race 2, lost two places in a bad start, but fought my way to the lead by lap 3. By lap 11, I had stretched out my lead to 7 seconds when I locked up going into Variante della Reggia, wound up on the run-off, and incurred a 20 second slow down penalty. (I'd have protested that, since I was alone on that section of the track and didn't gain any advantage in cutting the chicane, but the Assetto Corsa AI officials don't really listen to protests.)
I finished tenth.
And yesterday afternoon at Mugello? I was second fastest in practice, took the pole for Race 1, and finished second. All at speeds matching the real world Coppa Shell series. My setup? Default with slightly softened front springs and two notches additional splitter and wing.
So yeah. Practice. Practice, practice, practice.
Of course, I practiced. But mainly with an eye to improving my setups, because that HAD TO BE the secret to going fast, right? Had to be.
And I did get faster. Slightly. A little bit at a time.
I compared my lap times with real race drivers in similar cars, as posted on the WEC results pages -- and finally accepted that I couldn't ever measure up. No way. I'll never be that fast.
Inevitably, I started playing around with AI. Hot lapping was fun, but frustrating, and I thought maybe I'd have more fun if there were other cars on the track. The prospect of racing with cars going faster than me was daunting, though. There was no way I'd ever be able to race with those AI cars, even at 80%.
And then I realized that, by creating my own AC career, I could adjust the AI to BELOW 80%, and hey, that would be a great way to inject some fun into the simulation. I did a little research, found a racing series that looked like it would be fun to (sort of) emulate, and I built it: Corse Clienti's Ferrari Challenge Europe. Because I could include all the tracks from the '15 season. I adjusted the AI percentages for all the opponents to 40%. As a starting point. Because, you know, I'm slow.
In the practice session at Monza, I quickly discovered that the AI cars would brake too early for each corner, and I learned to anticipate their braking while watching for my own braking point. I learned to read the other car's line and to set up the pass on a less-than-ideal line.
I ended up not saving that first practice. I'd discovered that I was faster than all the AI cars, so I adjusted the opponent.ini file to set the AI cars at 60%, and went back to it. This time, the AI didn't brake as early, but they were still easy to out brake. I got better at cornering off line. There were a couple times when I thought I'd been held up by traffic and still drove a faster lap than I'd ever been able to before. What's happening here?
I also discovered that the AI cars would bunch up and run in groups, making it impossible to pass them all at once and that forced me to occasionally run in the middle of a bunch of slower cars. It occurred to me that if I adjusted the AI opponents individually, it might help break up that bunching, so I went to the Corse Clienti website and compared the drivers in the Coppa Shell series, ranking them on a scale of 1-9 according to their average performance in the season so far. I was getting faster, too, and the 60% wasn't really cutting it any more, so I subtracted each driver's ranking from 70%, giving me AI strengths from 61-69%.
And of course, I noticed the actual lap times that the real world Coppa Shell drivers were turning in. There was no way I'd ever be that fast, I thought. No way. By Saturday afternoon, I was within about five seconds of those real world times, and as far as I was concerned, that was as good as I'd ever get.
My setups? I wasn't putting much effort into them. Default, mainly. Soften the front springs a little.
By Sunday morning, it was clear that even the 69% AI drivers weren't fast enough to offer me much competition. I still hadn't saved that first practice at Monza, though. My efforts were focused on dialing in the AI. The problem was that every time I bumped up the AI percentage and drove some more, my lap times would come down, and I'd have to adjust them some more.
Because, you know, I kept getting faster.
Late Sunday morning, I took the plunge and set the range for the AI drivers from 81-89%. I didn't think I'd be able to keep up, but I was clearly much faster than the AI had been, so I figured I'd give it a try.
That's when the magic happened.
The AI cars were now putting in lap times that averaged the same as their real world counterparts in the Coppa Shell series. AND SO WAS I.
When I was just hot lapping at Monza in the Ferrari 458 GT2, I'd been consistently turning in 1:59s, but now? 1:52s. Not alien fast, to be sure, but that's what the real Coppa Shell drivers are doing. Not the pack. The fast guys, Duyver and Prinoth and yeah, those guys.
I took the pole for Race 1 on Sunday afternoon, and was in second place two laps in when I got called away from the computer and had to DNF. On Monday, I took the pole again for Race 2, lost two places in a bad start, but fought my way to the lead by lap 3. By lap 11, I had stretched out my lead to 7 seconds when I locked up going into Variante della Reggia, wound up on the run-off, and incurred a 20 second slow down penalty. (I'd have protested that, since I was alone on that section of the track and didn't gain any advantage in cutting the chicane, but the Assetto Corsa AI officials don't really listen to protests.)
I finished tenth.
And yesterday afternoon at Mugello? I was second fastest in practice, took the pole for Race 1, and finished second. All at speeds matching the real world Coppa Shell series. My setup? Default with slightly softened front springs and two notches additional splitter and wing.
So yeah. Practice. Practice, practice, practice.
