GT4 AI: OMG, it's even worse than I thought.

[RaidoGT]

Right, I'm sure we all know that GT4 AI is braindead. That it makes no effort to avoid you whatsoever, that it always tries to push you off the road, that it is ridiculously fast in some corners (often bouncing into the guardrails as well) et cetera.


But I wonder if it's universally known that PD has also *deliberately* programmed it to cheat.


First, a short story recap: so there I was, entering the Japanese 80's competition, ready to sweep the field in the '89 Skyline GT-R 4WD even on N2 tyres. By comparing several 80's cars I was planning to see which one would give the most fun competition, and I wanted to start with an 'easy' one first to see the lay of the land, so to speak.

Well, no dice. The AI ran away from me like there's no tomorrow, producing 'hare' after 'hare' which got miles ahead of the rest of the field, time after time. Didn't matter if it was a small 140BHP low-horsepower car like the '86 MR2: they all acted like they had near 400 BHP.

So far, I had thought this was a random occurrence; often you get a rabbit, sometimes you do not. Bad enough, since it's not very realistic (especially for underpowered cars) but then the GT4 AI is known to be bad. Leading to the well-known habit of having to hunt for a good lineup. Which, in the end, I found I had to resort to as well. After a while I found one in which the 4WD Galant wouldn't run too far ahead in the first lap, it seemed. In the end, after trying a few times I got ahead of it and won, but it was far less easy than it should have been, given there were no other GT-R Skylines in the field.

And then, in doing so I started to notice a pattern. One that seemed familiar, and vaguely disturbing.** Ran the field again to confirm my suspicions, and they were. The AI is deliberately programmed to cheat, to become artificially fast if certain conditions are met.

It turns out that the 'rabbit' or 'hare' AI isn't so random at all. I found (perhaps others have noticed as well) that if you manage to pass the lead AI early on, you can stay in the lead fairly easily. The (former) lead car will not get ahead of the field all that much, in general, in my test case about one second or so.

But when I deliberately put my Skyline right behind the leading Galant and stayed there as if I was giving chase, it suddenly turned into an outright rocket. After a few laps we were miles ahead of the field (about six seconds) and I had to drive much faster than I had to do the other times, when I got in the lead early on.
*In other words, PD deliberately programmed the AI to cheat and become artificially faster in such conditions - to make sure you always stay behind.* Or at least have to drive like a maniac to get past. It has nothing to do with your driving performance in itself; it just gets an unreal speed boost when you're still behind. This is akin to the infamous "Rubberband AI" in more unrealistic games like Need For Speed, where the AI almost always gets an artifical advantage to stay ahead or catch up. Once you get ahead, on the other hand, the AI's speed stays fairly consistent.

I'm sure the general idea was to "put up a challenge" and such fluff (as if the AI doesn't already do that by putting up faster lineups if you show up in a fast car yourself. I'm sure there are dimwits out there who even support such a fake speedup, ignoring that it's technically no problem at all to have the AI zip around an entire lap in 0.2 seconds, to make you lose no matter what).
In the end though, the AI was programmed to cheat and suddenly make itself artificially faster than otherwise (after all, if I got in the lead and get away from the field early on, this didn't happen - measured this several times, and the AI's behaviour was easily repeatable). Doesn't matter if it's a small city car: suddenly it will become just as fast as your big-bore GT-R and you'll need a bigger turbo or nitro to get by. Needless to say, this is totally unrealistic and a deliberate ploy by PD to have the AI cheat, speedwise.


I don't know about others, but personally I find this deeply offensive (especially in the past people paid good money for GT4 and in return, they get shafted by the already boneheaded AI coders - deliberately) and unacceptable. If KY and/or the AI's programmers would stand right in front of me right now, I'd punch them in the face. Hard. And I'm not even kidding. Because this is not only dumb AI, it's also been deliberately programmed to fake things and make human gamers lose.
That wouldn't have been quite so bad if this had been a pc game and you'd be able to have a patch come out to correct things - but in a PS2 game that's just not possible. The cheat has been deliberately put in there by PD, and it will never go away.

In other words: PD delivered a cheating game here, they knew it, and made gamers pay real money for it. 👎

Totally. Freaking. Unacceptable. Imho at least.
Personally, in the future, before I even *consider* looking at GT5, I'll now make darn sure I know that PD didn't deliberately mess up the AI - again.


**A bit of history lesson for the newcomer drivers: In 1998, when Grand Prix Legends arrived as the now-legendary Godfather of serious racing sims, it came with a system called 'Global Hype'. This made the AI slow at first to match your beginner laptimes but automatically speed up as you got faster. Doesn't sound so bad at first, - but it never stopped getting faster: no matter how good you got, you'd always end up losing since AI is always able to speed up indefinitely. Needless to say, this system was almost universally hated and neutralised by the gamer community through various patches later on.

 

[daan]

You do know the AI can run modified cars?

 

[Greycap]

Had the case been that the AI always runs faster I probably wouldn't have ended up with a 100% completed game with a 100% win ratio. Also worth noticing is the notoriously slow pace the AI exhibits when on cold tyres, the sudden increase of speed is often a result of them getting the tyres warmed up.

 

[Parnelli Bone]

^yea, what daan said...

Also, running N2 tires in a race full of cars on S2s is a ridiculous idea. I know people do it sometimes for an additional challenge, but it's not an easy endeavor. This sort of practice should be taken up by only those drivers who are really looking forward to a super-challenge, like I said. If you use cheaper tiers, you'll need to take up a lot more of a cornering path. You won't have nearly as much control. The Ai will.

I'd have to disagree with the "rubberband Ai" statement as well. Rubberband effects are definitely in GT1 & 2, but not in 3 or 4. In the earlier games, Kazunori has admitted the Ai could magically gain more grip in corners if it fell behind (which is why you'd see cars wipe out, fall far behind, and 30 seconds later they'd be back fighting you again). I think power could also surge and fall in these earlier games. It gives GT1 & 2 a super-exciting feel to be surrounded by all these cars fighting with you 80% of the time, but it ain't realistic at all.

Instead, the cars in GT3 and 4 are tuned to some extent, and yes, there are some jackrabbits; but there is no rubberband effect, sorry to tell you. Cars in GT3 and 4 can fall far behind if they wipe out, and in many cases that's the end of the race for them...you won't find a car wipe out, fall many seconds behind everyone else, and seconds later be fighting you again.

Finally, you should know this: cars in GT4 CAN become faster in later laps because their tires get warmer. As their tires warm, they drive more confidently. Get used to it and stop whining.

If KY and/or the AI's programmers would stand right in front of me right now, I'd punch them in the face. Hard

...ooookay. Uh, welcome to GTP>

 

[RaidoGT]

I know about those, but that's not the issue here. It's about the relative - artificial - speedup of the AI cars in relation to yours. That's not random, that's deliberate - and the very definition of cheating.

If a human player came out (online) and put up a car that'd always be 0.5 seconds a lap faster no matter how brilliantly and faultless your drive, you'd call him a cheater. That's what I'm calling PD right now, too. Wouldn't have been necessary to put it in, either.

I'd have to disagree with the "rubberband Ai" statement as well. Rubberband effects are definately in GT1 & 2, but not in 3 or 4.
Instead, the cars in GT3 and 4 are tuned to some extent, and yes, there are some jackrabbits; but there is no rubberband effect, sorry to tell you.

Oh yes there are. I just seen 'em, measured it, it was repeatable. There is no denying that it's basically rubberband AI, just more applied to the car in front.

Also, running N2 tires in a race full of cars on S2s is a ridiculous idea. I know people do it sometimes for an additional challenge, but it's not an easy endeavor. This sort of practice should be taken up by only those drivers who are really looking forward to a super-challenge, like I said. If you use cheaper tiers, you'll need to take up alot more of a cornering path. You won't have nearly as much control. The Ai will.

It's not ridiculous, I won all my series/cups that way, so far (no driving aids or downforce either). Because S2's are simply nowhere realistic for non-race cars and make things way too easy (for my taste in any case). Sometimes with difficulty, but it's doable.

That said, it has nothing to do with the issue I noted here: no matter which tires you're on, the lead AI will in such cases always get that artificial speedup; I'm sure that if you put R1 racing tires on, they'd be even faster.

As their tires warm, they drive more confidently. Get used to it and stop whining.

It has nothing to do with tires warming up. Only the lead car exhibits this behaviour. The others speed up a bit as well, and that's indeed fairly normal; but they don't turn into rockets as you chase behind 'em.

Oh, and if feeling shafted at being cheated by a product is 'whining', I have some subprime mortgages or swamp land in Florida to sell ya.

 

[Parnelli Bone]

I know about those, but that's not the issue here. It's about the relative - artificial - speedup of the AI cars in relation to yours. That's not random, that's deliberate - and the very definition of cheating.

If a human player came out (online) and put up a car that'd always be 0.5 seconds a lap faster no matter how brilliantly and faultless your drive, you'd call him a cheater. That's what I'm calling PD right now, too. Wouldn't have been necessary to put it in, either.

You're whining. Stop it. It's embarrassing.

If a human player came online and drove half a second lap faster than me no matter how brilliantly I drove, I'd compliment him (or her) on his (or her) skillz. 💡 Then I would practice, assuming I want to be a better driver, that is....

I'd say most of the "regular" people here in the GT4 forums have some complaints about GT4's Ai. Certainly the fact that they ram you and don't recognize where you are 100% of the time is one of these complaints. But the fact that they drive faster in later laps is never one of these.

 

[the Interceptor]

@RaidoGT: I don't really see where you're trying to get with this discussion.

The AI in GT4 is not unbeatable. There may be certain effects in its behavior which alter its consistency in terms of lap times. If you personally think that this is unacceptable, then stop playing the game and choose another one. Also, of course, you are free to wait and find out whether GT5 shows the same effects, and only then decide whether to buy it or not.

I just don't see what this thread is for, since there seems to basis for a discussion.

 

[Tornado]

GT4s AI doesn't rubberband. They are useless and uncompetitive most of the time, and they are complete FAIL at nearly half of the tracks in the game depending on the cars they are driving (especially in FF cars, which raises it to about 3/4), but they don't rubberband at all. In fact, I played GT4 first and then played GT3 and was surprised at how much harder that game was because they do rubberband. And I've since noticed it in GT2 as well, something like 9 years after first getting the game. But they don't in GT4.

 

[Parnelli Bone]

@RaidoGT: I don't really see where you're trying to get with this discussion.

He wants someone to commiserate with. Some sympathy? He won't get it from most of us, I'm afraid.

The AI in GT4 is not unbeatable. There may be certain effects in its behavior which alter its consistency in terms of lap times. If you personally think that thius is unacceptable, then stop playing the game and choose another one. Also, of course, you are free to wait and find out whether GT5 shows the same effects, and only then decide whether to buy it or not.

I just don't see what this thread is for, since there seems to basis for a discussion.

It's funny, most people when they complain about Gran Turismo complain about how easy it is to beat the Ai. It's rare to see the opposite happening!

 

[RaidoGT]

You're whining. Stop it. It's embarassing.

Happy to be the embarrassed one, in this case. But rather, I'm angry that programmers would go this far.

If a human player came online and drove half a second lap faster than me no matter how brilliantly I drove, I'd compliment him (or her) on his (or her) skillz.

Nonsense, and you know it. If your drive was flawless, you should at least cross the finish line at the same time. Anything faster from the other party, and you just know they're cheating. I've been witness to some such behaviour online sometimes, and we all knew what was going on.

But the fact that they drive faster in later laps is never one of these.

Like I said, this has nothing to do with that (you can see and measure it in the behaviour of the other AI, after all). You're confusing the issue, since it doesn't have anything to do with tires at all, basically, only with relative speedup of certain AI.

It's funny, most people when they complain about Gran Turismo complain about how easy it is to beat the Ai.

Yeah, but most of GT players do so by having S2 tyres (or better) and driving aids on. Then it gets easy, sure... That said, even then the game shouldn't ever cheat. Even if you were the world's best GT driver you'd still get left behind in such cases, since AI can after all speed up virtually indefinitely.

GT4s AI doesn't rubberband. They are useless and uncompetitive most of the time, and they are complete FAIL at nearly half of the tracks in the game depending on the cars they are driving (especially in FF cars, which raises it to about 3/4), but they don't rubberband at all.

Well, you are correct in that it doesn't always rubberband in GT4, and it's not nearly as common as in other games. And sure, they do fail miserably sometimes. But rubberbanding does happen *sometimes* in GT4 as well, from what I've noticed today.

 

[Greycap]

Nonsense, and you know it. If your drive was flawless, you should at least cross the finish line at the same time. Anything faster from the other party, and you just know they're cheating. I've been witness to some such behaviour online sometimes, and we all knew what was going on.

Ever heard of a term "faster driver than you"? During the last couple of years I've seen dozens of situations in which guy X has driven a lap he thinks to have been flawless, yet gets beaten by me by a good second and a half. Neither of us knows where the difference is. And I'm not the best out there by any means.

Yeah, but most of GT players do so by having S2 tyres (or better) and driving aids on. Then it gets easy, sure... That said, even then the game shouldn't ever cheat. Even if you were the world's best GT driver you'd still get left behind in such cases, since AI can after all speed up virtually indefinitely.

So... you mean that even when you're running on N2 tyres and the AI is using S2 they still shouldn't be faster than you because that means they're cheating? That's not rubberbanding, that's you having an inferior car. And that speeding up sounds very much like tyres warming up, the difference between cold and warm tyres when you're driving is perhaps one second per lap but with the AI we're talking about three or four seconds a lap. Yes, it really happens. Believe me or not. AI with cold S2 tyres is slightly slower than you on cold N2 tyres because you can actually drive with cold tyres but when the AI gets those tyres into their operating temperature you'll be eating dust, plain and simple.

 

[the Interceptor]

I have a hard time believing that the programmers at PD deliberately made the AI this way to cheat. I'd rather suspect the effect, given it is there, to be something that was supposed to be a "give the player a run for his money" feature which was not tuned very well when the game was released.

Have you ever looked at a replay from the point of view of the leading AI car and observed its behaviour directly? Like actual corner speeds, sector times and these sorts of things?

 

[Parnelli Bone]

Happy to be the embarrassed one, in this case. But rather, I'm angry that programmers would go this far.

Nonsense, and you know it. If your drive was flawless, you should at least cross the finish line at the same time. Anything faster from the other party, and you just know they're cheating. I've been witness to some such behaviour online sometimes, and we all knew what was going on.

Okay, smart guy. Recently there was a contest here in the GT4 tuning section, called the Tuner Challenge Version 5. It's down a page or two, but shouldn't be too hard to find. Anyways, there are four judges to this challenge. I am one of them. I'm not getting into all the fine details, but the challenge is being held at Twin Ring Motegi East. One track, which keeps things consistent.


There's a guy named McSqueegy (also called Macca) who was also one of the judges. In all cars we tested, McSqueeegy is a second or more faster than me. "How can this be?" I think to myself. It seems impossible, but when I think about it, it certainly IS possible someone could be faster than me.

Do I sit there going "Oh, you cheater!" No, I congratulate him on his efforts. Have a look at the thread if you must (and also the TCV5 discussion). Here...I'll make it convenient for you. The top link is the contest, the bottom link is the discussion.



https://www.gtplanet.net/forum/showthread.php?t=117848

https://www.gtplanet.net/forum/showthread.php?t=113949

How many times have I lambasted those who are faster than me? None.

Like I said, this has nothing to do with that (you can see and measure it in the behaviour of the other AI, after all). You're confusing the issue, since it doesn't have anything to do with tires at all, basically, only with relative speedup of certain AI.

Of course it has to do with tires, genius! You do know that as tires warm, they become grippier, right? 💡 You do know that as tires become grippier, the car has less of a chance of sliding off, right? Thus, it can become faster.


Again, stop whining...start practicing. 💡

Yeah, but most of GT players do so by having S2 tyres (or better) and driving aids on. Then it gets easy, sure... That said, even then the game shouldn't ever cheat. Even if you were the world's best GT driver you'd still get left behind in such cases, since AI can after all speed up virtually indefinitely.

I'm still at a loss here. I'm not the world's best GT driver, I allow myself to lose plenty of races. I think a 100% win ratio is a waste of time, yet I STILL manage to win plenty, too. You don't need to be the best to win in this game, trust me...

What you call "cheating", we call "challenge". I'm always glad to have a challenge in my game, otherwise I'd get bored quick.

Well, you are correct in that it doesn't always rubberband in GT4, and it's not nearly as common as in other games. And sure, they do fail miserably sometimes. But rubberbanding does happen sometimes in GT4 as well, from what I've noticed today.

Of course I'm (actually we) are correct. PD did not program Ai in GT4 to rubberband. If a car is making good time in this game, it's usually because it has more horsepower and/or it is supposed to go faster than the rest of the Ai. To some extent, PD tried to make sure the fastest, best-handling cars in real-life are also fast & good-handling in the game.

Case in point: last nite I'm doing the Silver Arrow series...for Mercedes Benzes. Uh oh....there's a McLaren SLR in the mix. I know beforehand he's got around 600 horses, which means this guy will jackrabbit. and sure enough he does. Do I sit there whining about it? Do I sit there thinking I should slam Kazunori in the face because I can't keep up? NO! I practice. I get to the point that I can keep up with the McLaren.

 

[Parnelli Bone]

Sorry for the double-post. But I just wanted to return to this statement.

If KY and/or the AI's programmers would stand right in front of me right now, I'd punch them in the face. Hard. And I'm not even kidding.

Hey, I believe you. You're not kidding. You sound like a downright scary dude with some serious issues.

Have you ever been mentally committed? Taken anger management classes? Seriously. Think about what you're saying. You'd "punch Kazunori in the face. Hard"

Take it easy. It's a GAME fer cryin' out loud.

The funny thing about this uh....statement, is that your fist would raise about 2 inches towards Kaz's face before he'd press a button and whatever security detail he's got present would bring your ass to the ground. I'd love to see what Japan's justice system would do to you...hopefully your incarceration would happen quicker and last longer than it would in the Netherlands.

But in all seriousness, you need help. For real. I'm not even sure you should be freely posting here at GTP, although that's for the mods to decide.

 

[Tornado]

Well, you are correct in that it doesn't always rubberband in GT4, and it's not nearly as common as in other games. And sure, they do fail miserably sometimes. But rubberbanding does happen *sometimes* in GT4 as well, from what I've noticed today.

I can enter a race (let's say Schwarzwald Liga B at The 'Ring) with anything from a 330 HP BMW M3 to a 1000 HP Mercedes SL65 AMG. And whether I finish ahead by a few seconds or a few minutes, the AI lap and sector times will not change.

 

[Lain]

Interesting read, but if the AI does have a rubber band effect as you described, Polyphony implemented it quite backwards.

To me it seems the AI should get faster if you can easily take 1st, and get slower if it's difficult.

 

[Giygas]

Yes, it does seem like there's a "rubber band" effect, but i doubt that is the case. if you chase a Bentley speed 8 with a BMW Maclaren , more than half of the time you would destroy it in the first lap or 2, but after that they come with the jaws of doom.

Probably due to the tires being warmed. I doubt there is an actual effect, I've trailed in 2nd many times and still beat the ahead car by a second or two. And trust me, IM NOT THE GREATEST DRIVER.

 

[Parnelli Bone]

Interesting read, but if the AI does have a rubber band effect as you described, Polyphony implemented it quite backwards.

To me it seems the AI should get faster if you can easily take 1st, and get slower if it's difficult.

GT2 is the worst with rubberband, especially in the easiest of races.

One time I did the Sunday Cup at High Speed Ring. I was driving a Honda Accord Wagon, and forgot to tune it. I was quite slower than I wanted to be. Didn't matter, the Ai swarmed around me as usual....not getting too far ahead or too far behind.

I even did a little experiment. Slowed way down...way below the usual light-speed pace HSR demands. Guess what? The Ai slowed down with me, all-a-sudden driving as tho they had all just taken a dose of Sominex.

On the other hand, I can remember some races I did at Laguna Seca...great example, since there's tons of situations where the Ai screws up the corkscrew area...literally twirling and spinning into each other. Doesn't matter if I get thru this mess unscathed and (let's say) get a 5 second lead....by the time I'm rounding Andretti's turns, those same Ai will have caught up to me.

...now that's rubberband folks.

 

[HEAVYSET]

The point that was made about the game throwing a tuned car at you without you knowing it was a good one. The rally events in GT3 were easy to gold until I got to the wet S.S. one. The AI Suby identical (or so I thought) to my own would dart away on the straights. I could corner with it in the early part of the first lap, but it would rocket out of the corners and be gone. I had to put another 200hp under the hood to match its straightaway speed. Sneaky, sneaky, sneaky, the game was.

Also, I always considered when a seemingly unmodified AI car gets the lead and tends to pull away from the rest of the AI, part of the reason is because it doesn't have to deal with any other AI cars ahead of it. As was mentioned, the main flaw with the AI is they get, let's say, a little confused around one another. Their brake timing is off and they take odd lines, etc. The lead car can just book it with near perfect braking and lines until you catch it. Then it has to deal with you.

 

[Seismica]

You see when you get a bunny in a race, they leave the rest of the field for dust.

Some races i have been in, over 5 laps i've been 30+ seconds ahead of the guy in 3rd, but the guy in 1st will be right on my tail. If i don't pick the right car, or if i don't get my lines right, it seems like the guy keeps pulling away.

The reality is there are subtle differences between your times each lap, usually because you took a slightly different line through a corner, which resulted losing about a second due to the straight afterwards, and then braking too early for the next corner (More than likely you will brake at the same point, even if you are going 20mph slower than you usually do).

I find that when racing on the Nurburgring with an M3 against a varying field of German or European supercars, i blitz the field in the first half, but the second half is very bumpy and the AI always seem to close a 10 second gap very easily (more often than not due to my mistakes). This is not rubber banding because racing a stock M3 against an SLR, of course the SLR is going to be faster. It just turns out in the first half of the 'Ring, the AI is very slow.

Most events restrict you to Normal/Sports tyres, and i think the difference between S2 and S3 tyres is so small its not worth the cash to buy the softs (unless you're tuning the car fully, or you have a stock car with softs, like a RUF RGT).

This game can be very challenging. I think that makes it more fun.

 

[Tornado]

GT2 is the worst with rubberband, especially in the easiest of races.

*anecdotes*

...now that's rubberband folks. *tips hat*

I still remember the race that I noticed how bad it was. It was the 80s car Cup at Tahiti Road, and I entered a fully tuned Subaru 360 (yes, I do see the problem with that statement). I blew away by about 4 seconds at the first checkpoint. Then I spun out at the 90 right before the long sweeping left turn. The AI blew by me going way faster than any crappy 80s cars could possibly be made to go in game, then they proceeded to slow waaaayyyy down while I collected myself from the wall. Good times.

At least the rubberband AI in GT3 was much more stealthy. At times you could barely tell if they were cheating or not.

 

[Smallhorses]

RaidoGT, you may want to do a little research before you come in with such an accusatory post. While it's true that the AI aren't the best programmed drivers in the world, they're limited by the vehicles they're assigned to drive too.

You can use the "Preview" function before a race to find out the HP and tyres that your opposition run before you race. (On the 6th placed car in the preview)
One member (jdw) who used to be very active in the 200pt Races forum did this for just about every event in GT4, previewing again & again & again until he'd been through lineups that had every available car starting in 6th place and noting down their HP values. He was kind enough to share his data with a few people that requested it from him.
While it's impossible to tell whether they've used other modifications such as weight reduction or drivetrain or suspension upgrades, it's certainly apparent in many cases that the AI has power boosts in many cases.
In this case for Japanese 80s event, here are all the possible runners, and it's readily apparent that almost all, if not all have power upgrades above and beyond their stock HP values [Power in square brackets]:

Nissan Skyline HT Turbo RS '83 [235],
Toyota Sprinter Trueno GT-Apex '83 [229],
Mitsubishi Galant VR-4 '89 [228],
Mitsubishi Lancer 1800GSR '83 [228],
Toyota Carina ED '89 [228],
Toyota Celica 2000GT-FOUR (ST165) '86 [228],
Toyota Celica 2000GT-R (ST162) '86 [228],
Mazda RX-7 GT-Lim '85 [228],
Honda Accord '88 [228]
Nissan Skyline GTS-R (R31) '87 [227],
Isuzu Piazza XE '81 [226],
Toyota Celica GT-R (ST183,4WS) '91 [226],
Toyota Celica XX 2800GT '81 [226],
Nissan Exa Canopy '88 [225],
Nissan Silvia Qs '88 [225],
Toyota MR2 G-Lim SC '86 [224],
Nissan Silvia Ks '88 [224],
Toyota MR2 1600G '86 [223],
Toyota Corolla Levin GT-Apex '83 [222],
Nissan Skyline HT 2000 RS-X '84 [221],
Honda Ballade Sports CR-X '83 [217],
Honda City Turbo II '83 [216],
Honda Civic 1500 3door 25i '83 [197]

If you aim for a grid which has some of the heavier FF cars such as the Accord and Carina on the front row, it should prove easier as the AI in faster cars are not wonderfully adept at passing even their animated bretheren, as you've no doubt noted.
Looking at this data, it's not surprising at all that an initially light MR car such as the MR2 (1070kg) with almost 100HP of upgrades is going to significantly outperform a heavy FF car such as the Carina (1200kg) or FR car such as the RX7 (1280kg) when they're aligned at much the same power rating (give or take an HP or 2.) Similarly one wouldn't expect the Nissan Skylines (~1250kg)which're only given a 30-or-so HP boost over their stock form, to be competetive with a Honda City Turbo (735kg) that's had over 100HP of upgrades even with the AI's ineptitude at driving tuned FF cars.

If you need further info about the full AI fields for any given event, I can provide it to you if you shoot me a PM. (Allow me some time to get back to you as I'm mega-busy with work at the moment.)

 

[oldmodelt]

Yeah, but most of GT players do so by having S2 tyres (or better) and driving aids on.

It's my understanding that the driving aids (asm/tcs) actually slow you down if you are a good driver. Very few use them that I know.

 

[SuperCobraJet]

Like in real life, you get surprised sometimes.

Back in the old muscle car days, many of the fastest cars were the old plain Jane models, with modified engines and drivetrains.

Hustler cars looking for an easy mark. They always found some too.

 

[PF]

I'm not yet ready to dismiss your theory RaidoGT, but am sceptical. If it is, as you say, repeatable, then you should be able to do the following:
* Repeat the race with the same lineup by getting to first early (with no AI contact, cause we all know how much you love bumping the AI!), and photograph the results screen with second place's time.
* Repeat the race by getting to second and tailing the lead car, and photograph blah blah blah.
* Post the photos showing a significant difference in the AI times between second and first place.

Even then it wouldn't be 'proven' but would be good supporting evidence.

 

[RaidoGT]

PF, yeah, it was totally repeatable. That's why I got suspicious in the first place.

It seems some of you have some trouble with reading comprehension though, attributing the phenomenon on tyres, warming up or overall AI speed/performance, when I already said those had nothing to do with it. (Had they been, I wouldn't have thought it was the AI speeding up as it did.)

It's really very simple, and should be repeatable:

- In the same race, if I got into the lead early on, the leading AI didn't get away from the pack.
- Start same race again, got into 2nd and pretended to give chase to the lead AI. Which then speeds up and gets away from the pack (me following) like there's no tomorrow.

- Started the same race again (twice) to see if the behaviour's repeatable.
Turns out it was. Hence, the AI does cheat and turns into a rocket (to keep you behind) if you're not in the lead, but doesn't get away from the field if you are.

- Note, it doesn't *always* work like this; at least in GT4 in some races it seems to take effect, in others it doesn't. May depend on choice of cars or AI lineup.

It's not pure rubberbanding as in NFSU, strictly speaking it's more of a 'reverse rubberband', but the effect is about the same. But it's there nonetheless, at least in some races.




PS. Of course, in that first post, I needed to vent a bit Feel better now, at least a little. Naturally, if I met KY and the responsible coders in real life, I'd do a sincere "sumimasen" followed by a "失礼します" to apologize for my rudeness and give a courteous explanation of my objections, supported by some game footage.

And then I'd deck 'em both. Totally.

Afterwards I would of course utter a most sincere and polite "moshiwake arimasen" (have been in Japan for a while so had to learn the keigo lingo for a bit) and offer both a pint in the local pub (it's only a game, no need to be unpolite, right? <grin> )

 

[Austin343]

I agree that quite a lot of people read your post incorrectly, but the criticism that your findings might be based on perception and not on facts is still justified. As someone suggested, you should repeat your "experiment" and take photos to confirm that you were experiencing a real effect. And the AI most certainly does not "cheat" like it did in GT2, but rather increases its pace to be closer to that of a human, assuming that such a phenomenon exists in the first place.

 

[Parnelli Bone]

PF, yeah, it was totally repeatable. That's why I got suspicious in the first place.

It seems some of you have some trouble with reading comprehension though, attributing the phenomenon on tyres, warming up or overall AI speed/performance, when I already said those had nothing to do with it. (Had they been, I wouldn't have thought it was the AI speeding up as it did.)

It's really very simple, and should be repeatable:

- In the same race, if I got into the lead early on, the leading AI didn't get away from the pack.
- Start same race again, got into 2nd and pretended to give chase to the lead AI. Which then speeds up and gets away from the pack (me following) like there's no tomorrow.

- Started the same race again (twice) to see if the behaviour's repeatable.
Turns out it was. Hence, the AI does cheat and turns into a rocket (to keep you behind) if you're not in the lead, but doesn't get away from the field if you are.

- Note, it doesn't *always* work like this; at least in GT4 in some races it seems to take effect, in others it doesn't. May depend on choice of cars or AI lineup.

It's not pure rubberbanding as in NFSU, strictly speaking it's more of a 'reverse rubberband', but the effect is about the same. But it's there nonetheless, at least in some races.




PS. Of course, in that first post, I needed to vent a bit Feel better now, at least a little. Naturally, if I met KY and the responsible coders in real life, I'd do a sincere "sumimasen" followed by a "&#22833;&#31036;&#12375;&#12414;&#12377;" to apologize for my rudeness and give a courteous explanation of my objections, supported by some game footage.

And then I'd deck 'em both. Totally.

Afterwards I would of course utter a most sincere and polite "moshiwake arimasen" (have been in Japan for a while so had to learn the keigo lingo for a bit) and offer both a pint in the local pub (it's only a game, no need to be unpolite, right? <grin> )

Whatever. 👎

Are you still trying all this on N2 tires? '80s races?

 

[Seismica]

It's my understanding that the driving aids (asm/tcs) actually slow you down if you are a good driver. Very few use them that I know.

Wrong. If set up correctly, the electronic aids can make you much faster. TCS stops all wheelspin, but keep it low enough to stop understeer. ASM to prevent Understeer, i used to set this at 15, and it meant i could go around corners fast with no skill. But eventually i got past all that, i now use manual gears and turn all aids off, but i still can't beat my 'Ring laptime in the 88cv that i made around 3 years ago...

With good skill and consistency i think aids off would be better, but we don't know for sure, i don't think anyone has done in depth experiments with different driving aid setups. My example is probably just a one-off.

EDIT:

You are right in saying most people don't use them At least anyone whos serious about the game, instead of those who just pick up and play and don't bother completing the game.

 

[PF]

Hey RaidoGT,

It seems some of you have some trouble with reading comprehension

you should be able to do the following:
* Repeat the race with the same lineup by getting to first early (with no AI contact, cause we all know how much you love bumping the AI!), and photograph the results screen with second place's time.
* Repeat the race by getting to second and tailing the lead car, and photograph blah blah blah.
* Post the photos showing a significant difference in the AI times between second and first place.

Even then it wouldn't be 'proven' but would be good supporting evidence.

And let me reiterate that I'm not dismissing your opinion outright, as seen here:

I'm not yet ready to dismiss your theory RaidoGT