Posted: 15 Dec 2011, 06:44
Just for something to do I decided (using just the rookie cars for now) to run some speed tests on cars using Wolf R4, 30 cars of the same time, no collisions, no weapons, and racing on Toys In The Hood 2 (the one with the bridges, since its shorter and has more terrain variation than 1.)
Then I averaged the scores from the 29 cars, took the scores from the top 3 and the bottom 3, and checks to see how much variation there was between the top 3 and bottom three scores. So each of the cars I tested ended up with four values: Average, Top Scores, Bottom Score and Variation.
The Average score is really a rating for how good this car is, at least when driven by the AI. The Top Score is how good this car can be, and the Bottom Score is how bad it can be. The Variation shows how consistent this car is at reaching its average score.
(I got lazy doing the math and neglected the milliseconds on some of these, which can influence the scores a point or two.)
The first three scores are time in seconds, so a lower score means a faster car. The last value is the difference between the second value and the third value.
So based on that I can see which cars are fastest, which cars are consistent, and how these two custom cars compare. We can tell that Volken Turbo and Dr. Grudge are the fastest of the stock cars tested, but Volken Turbo is much more consistent, while Dr. Grudge has the potential to do better. So Dr. Grudge is more of a wildcard.
Phat Slug is also quite slow on average, if you didn't know that already.
RC Bandit and Harvester perform roughly the same, but Harvester is much more erratic with the potential to be faster.
Now the fun part, we can compare the custom cars. Conquer 2.0 was the only custom rookie car that I had on hand at the time, besides my Grimace car, so those are the two being tested.
We can tell from this data that Conquer 2.0 is very erratic, which shouldn't come as a surprise if you give it a test drive. It also scores rather badly, meaning for it to be challenging either the AI needs to be tweaked or it needs some parameter enhancements. Which change is needed is probably better determined by test driving. But since my own race time with this car (2:34) was about equal to Dr. Grudge (2:35) and Grimace (2:34), I'm going to say that the AI probably needs tweaking.
Grimace on the other hand is a little more consistent, but has the best rating on this board by a small amount. That means its reasonably balanced, but probably a little fast. However since my race time with this car (2:34) wasn't amazing, it might just be easier for the AI to drive.
My best race times were with Harvester and RC Bandit (both 2:30)
Of course, for anything to be reasonably based on my personal race times, I'd have to do many more races to get a better average. And that would change for everybody.
Then I averaged the scores from the 29 cars, took the scores from the top 3 and the bottom 3, and checks to see how much variation there was between the top 3 and bottom three scores. So each of the cars I tested ended up with four values: Average, Top Scores, Bottom Score and Variation.
The Average score is really a rating for how good this car is, at least when driven by the AI. The Top Score is how good this car can be, and the Bottom Score is how bad it can be. The Variation shows how consistent this car is at reaching its average score.
(I got lazy doing the math and neglected the milliseconds on some of these, which can influence the scores a point or two.)
The first three scores are time in seconds, so a lower score means a faster car. The last value is the difference between the second value and the third value.
Code: Select all
Phat Slug 181, 172, 193, 21
Volken Turbo 167, 161, 178, 17
Dr. Grudge 167, 158, 189, 31
RC Bandit 171, 164, 182, 18
Harvester 170, 159, 190, 31
Conquer 2.0 187, 170, 201, 31
Grimace 166, 159, 183, 24Phat Slug is also quite slow on average, if you didn't know that already.
RC Bandit and Harvester perform roughly the same, but Harvester is much more erratic with the potential to be faster.
Now the fun part, we can compare the custom cars. Conquer 2.0 was the only custom rookie car that I had on hand at the time, besides my Grimace car, so those are the two being tested.
We can tell from this data that Conquer 2.0 is very erratic, which shouldn't come as a surprise if you give it a test drive. It also scores rather badly, meaning for it to be challenging either the AI needs to be tweaked or it needs some parameter enhancements. Which change is needed is probably better determined by test driving. But since my own race time with this car (2:34) was about equal to Dr. Grudge (2:35) and Grimace (2:34), I'm going to say that the AI probably needs tweaking.
Grimace on the other hand is a little more consistent, but has the best rating on this board by a small amount. That means its reasonably balanced, but probably a little fast. However since my race time with this car (2:34) wasn't amazing, it might just be easier for the AI to drive.
My best race times were with Harvester and RC Bandit (both 2:30)
Of course, for anything to be reasonably based on my personal race times, I'd have to do many more races to get a better average. And that would change for everybody.