Gingiba
for me, i think the Mazda 787B is the winner for me now, even though i was in the pesky seat for awhile. I'm now a Mazda 787B driver
driving the pesky for quite sometime was fun, but now i'm driving the 787B like crazy and was it a good choice. the car is easy to handle and sounds nice 2
well, the sauber in my hands seem to be very well under controlso, the car is great with me 👍
Law93ok guys.....lol. i don't think the sauber isn't really meant for me....so, i think i'm going with the 787B and never leaving it again
That is sig worthy material, my friend.Digital Nitrate
Digital-NitrateAs for Team666 . I thought we agreed to disagree
I will say that if you look at my tests youll see I have logged thousands of miles on the cars that I have tested. Furthermore, I have repeated the same line-up to see how consistent the results would be and they have been within 99% accurate and reproducible. I purposefully left off those repeat line-ups form the data as they not only didn't offer any new information, but because the results of those repeats were always the same, it would possibly weight the average results unfairly.
In fact, I go out of my way to make sure ALL cars are equally represented in the data, are tested against their closest rivals, and with different starting grid positions. In other words, I am quite aware of how certain variables can affect the results; such that I do everything, I can to minimize them.
In addition, I have compared both A-spec and AI performance, and with only a few exceptions, while the times are obviously better under A-spec, the rankings are unchanged. I post, and prefer AI test results as they are verifiable by everyone willing to take the time to run the same test(s), and most importantly reveal the capabilities of the AI for a specific event so that you can select a car based on the AI's abilities.
The advantage of comparisons from AI driven cars is that they are extremely consistent (lap times within 99% of average), easily monitored, and can be verified by EVERYONE Not just the person driving, possibly making specific adjustments that perhaps cater to their specific driving technique/preferences and that can only be accurately reproduced, thus verified by the original tester.
(If you really feel the need to continue discussing the merits of my testing techniques, feel free to send me a PM)
I´ll run these AI tests myself, since I have such a hard time believing the consistancy of your findings! Wish me luck, and I´ll be back with a PM in a few weeks!
Yes, and please pause the game at the 24hr mark and take a photo, similar to what I did, that shows the current position, number of laps, number of pit stops, and fastest lap.Team666No, I really don´t feel like it!
No one can recreate that field.... for several reasons. First of all half of those cars are never used by the AI on 24hr La Sarthe II. Second of all, there are only 5 AI cars in a field, and because running your own car risks interfering with the results it makes no difference what it is as you MUST either park it in the pits or off the track for the entire race.Team666
I´ll do that! Any more tips are welcome in my messagebox!Digital-Nitrate
Have you actually looked that up? AFAIK, all of the cars in the example are in use by the AI on that track. IIRC the possible startingfields for Cdls I and II are the exact same. Eligable (AI) cars are AFAIK:Digital-Nitrate
Digital-NitrateHowever, as mentioned before, on several occasions I have repeated the exact same line-up and starting positions from the previous race and the results (both in times & placing) have always been within 99% of the previous, sometimes even less than that.
Moreover, as for any doubt on the consistency of the AI, feel free to look at the race results for each testing session from my 24hr CdlS II ratings:
You'll see that in most cases, despite racing each time for 24 hours and over 3,600 miles, that for each car's session on the 24hr mark, the AI had driven that car within just a couple miles difference from one session to the next. In many cases the difference was less than a mile!
Of the five sessions with the Minolta 88C-V, the worst it did was to go 3,644.3 miles by the 24hr mark, while the best was 3,645.4 - that's a difference of only 1.1 miles from worst to best after five full races and 15,213 miles! From worst finish to best, the AI was off by only 0.03%
I'm sorry, but no A-spec driver is going to be that consistent... over all those miles... ever!
I seriously have no idea where the rumors came from that the AI drivers are wildly inconsistent. Possibly it is from those watching B-spec Bob who does struggle more than the AI, certainly when the skill points are low. Possibly it is from those watching on short races where grid position plays a far greater role in the outcome of a race.
Are the AI wildly slow? Perhaps. Do they have some difficulty with some cars on some tracks? Yes. However, they are bloody spot on in the consistency department... and their "go-easy" speeds probably help insure their consistent lap and race times.
BTW: You mentioned that the AI doesn't know how to drive the Bentley, thus their results with that car on CdlS I are very inconsistent.
My tests show something entirely different.
After six sessions, 144 hours, 19,936 miles, the average difference between each session was just 1.5 miles! That's a 99.96% accuracy... and that's when it was starting in different grid positions and up against different opponents, which can either slow down its progress when being lapped (bumping into the slower car before making the pass), or speed it up when getting lapped (bumped from behind).
Here are the distances the AI drove the Bentley by the 24hr mark of each of the six races:
- 3,326.8
- 3,325.4
- 3,319.5
- 3,323.1
- 3,320.5
- 3,320.9
I'm sorry, but any notion that the AI is inconsistent from one race to another in the same car, over distances that would not be affected by grid position is absurdly inaccurate. The AI is definitely slower than any marginally skilled A-spec driver, but they are brutally consistent over the length of a race.
Have you actually even read my test results?Team666
YDNRC (You did not recall correctly)
I guess I'm lucky because after more than half a million miles of monitoring AI cars on CdlS no AI car has ever experienced it.
Thank you, I stand corrected! Allthough I wasn´t completley off with the cars... I must have mixed in the Infineon and Fuji endurances somehow...Digital-NitrateHave you actually even read my test results?
YDNRC (You did not recall correctly)
The eligible cars are listed in each of my threads on CdlS I & CdlS II, but I'll correct your list for you if you like.
The cars in BLUE are available to the AI on CdlS I, but NOT on CdlS II.
The cars in RED are available to the AI on CdlS II, but NOT on CdlS I. (which also makes sense as this was the track used by Group C Le Mans types, and not by modern LM cars.)
The cars in PURPLE are available to the AI on both CdlS tracks:
then you are indeed lucky! Those things happen to me all the time at both Le Mans tracks. I may have to use another testbed to get better results, since these glitches are more consistant than the AI...Digital-Nitrate
I think not, since like I said, the glitches are too abundant. I would never get to experience an entire race that way!Digital-Nitrate
Don´t get me wrong, I´ve done very extensive testing myself, both in GT3 and GT4, but I rely on myself rather than, in my experience, a flawed and highly unreliable AI. The reason I do all the testing myself is that the AI are highly inconsistant and the fact that they don´t use the cars full potentials. I simply don´t trust them to do anything other than drive poorly and most likely crash at various occations.Digital-NitrateBTW: You may want to try and familiarize yourself more with the game before you go around exerting a great deal of time and energy casting doubt on someone elses work. However, I'm glad to hear that you will be running these tests yourself, as this should put an end to this once and for all, and perhaps you'll finally admit that the AI results are very accurate, very consistent, and easily reproducible.
Well, I can understand your view, since my own is the opposite. I´ve spent countless hours testing things in GT3 & 4, and that has given me a lot of fun and a lot of experience, atleast in the settings area. The fact that you say my method is wrong is the exact thing as if I say your method is wrong. Wich I don´t, I simply question it from my experience with the AI. And your AI experience are obviously much more positive than mine!Digital-Nitrate
Like I said, I´m very familiar with the game, and a very experienced tuner, and I surely will not tolerate such a remark. I will do my own testing from your preferences, but I have a feeling my results may differ somewhat, since there are obvious differences between AI and GT4 copies. I will also have a look at the crashes, to see if the AI are preprogrammed to do certain things in certain cars. If that is the case I don´t have to bother with their crashes and runoffs, since the results will be the same anyway!Digital-Nitrate
No, it does not.Gingiba
Nitrate? I think you misunderstood my point. The A.I.'s overall speed when compared to B-Spec is roughly between 2-3. My theory was that since the Pesky can use R3's in this race and thus a higher speed setting (in addition to overall speeds when compared to the A.I. even on the "3" setting), if you were to B-Spec the race with R3 tires F/R you could concievably win because of higher cornering speeds if it was set to 4 (and possibly even 3). The Pesky '03 can use R4 tires in this race, and possibly be used in B-Spec even higher to make up for it's BHP deficit when compared to the Goup C cars and '04 Pesky.
Sounds like both you and Team666 have VERY buggy copies of GT4, which may explain some of your observations.Gingiba
Law93
GT Pro
