The subject has been discussed to death I know but this is meant to discuss the direction it should go or hopefully will go in the future. Some people like it some people hate it in the current state. I’m in the group that does not care for the current version but feel that with some new tweaks see it being a better idea than right now. A few things I hope Turn 10 has looked into or are going to implement in the future.
Make sure that only the tracks that can be run circuit style or drag style are counted for the % of completion.
Reduce the the amount of laps required to “first sync up” your Drivatar.( We don’t need to do 10 laps per track to collect early data)
Make it so a total 100% sync is only possible with every class of car used on all qualifying tracks( You can drive a F1 car way different than a D class car on a given track)
Any lap that uses rewind is not counted in any way.(This might already be used but if not it should be)
Regardless of your friends list if they do not drive or have collected data for a certain class of Drivatar they will not be used.( Little Jonny that uses your XB1 to play and is on your friends list should not be a unbeatable driver)
Faster or more frequent data collection to advance the system.(Daily updates of qualifying data to improve the system also the reason you should not have to do as many laps at first to sync up)
That is just a little of what I hope to see used to better the system. If you dont race C class you should not have a Drivatar in game in C class. I think the current idea is good but if someone just used the E21 to rush through to sync up to the max is not getting the 100% best data to use when it comes to his or her D class Drivatar.
I think the drivatar system can benefit from longer-ie: more laps per race. That way you don’t have to smash and bash your way to the front. That should take care of the crash fest that was FM5!!
Indeed! One of the biggest pains in the a$$ in Forza 5 was the fact that you were ALWAYS placed on the very last position in the beginning of the race, no matter how you performed in the last heat/race. That’s not making it more challenging, that’s just frustrating and cheap.
Why don’t you take your time and make clean passes. If you cant pass a car then learn to finish 7th or whatever position you are in. This is after all a racing sim and not a demolition derby.
Because the races are only 2 or 3 freakin’ laps! Unless you’ve dumbed them way down (or are playing LeMans or the 'ring) there just isn’t enough time to pick & choose where to pass. An option to do a qualifying lap would work wonders.
Hence my statement “If you cant pass a car then learn to finish 7th or whatever position you are in. This is after all a racing sim and not a demolition derby.”
Its not a racing sim in career mode. Its just a pointless repetitive task with an implied goal to finish third.
Nothing about the actual racing in the career is sim. Its a free for all until you get around 5th on the grid and I find hard to believe people haven’t had drivatar issues unless maybe they purchased the game late.
Day 1 career race 1 I was destroyed numerous times by idiotic drivatars. Race clean all you want and you still got taken out. It wasn’t until t10 nerffed the drivatars that at least half the field were somewhat predictable/logical in their race craft.
I could take a #1 tune in multiplayer and cleanly win the race after starting last in 3 laps. In career it was exceptionally rare to have that happen. The short career races are a big reason for the terrible drivatars.
Firstly we’d have to know what the Drivater actually is and how it’s programmed. In order to be a virtual copy of each of us then data would have to be collected on every track in every class and possibly in different cars. The system would need to know how you drive in traffic, in clear air, your different lines, braking points how you pass and how you defend then compute a virtual you in every race and scenario. This would have to be done for every player - perhaps 2.5 million by this stage of FM5s life. This would be a monumental task and even then the data is constantly changing as people improve or play differently. In my experience Drivatars don’t mimick the player they’re representing very closely at all. It may be that there is a much broader range of AI and your Gamertag is applied to the one closest to you or that there is an attempt at true replication but FM5 is early stages and there’s a long way to go to refine it. I agree with T Rex in that the more data you collect the more accurate the model. To do it properly you’d never stop collecting data.
I long ago stopped thinking of Drivatars as people and just think of them as regular AI.
In both FM5 and FH2 there is a statistics section in one of the menus. If you take a look at this data, it is a good chunk of what goes into your drivatar data.
i am of the opinion that it was better to race against the AI than these drivatars. I don’t know why but even on the easiest setting, the front drivatar controlled car would always seem to be just a tad bit faster than me, even though I consistently set times in the top 1000. (this might be because someone on my friends list is just a tad bit faster than I am.)
sometimes I am just in the mood to set the game to easy and see how many times I can lap these cars before the race is over, this stopped with fm5. one of the few disappointments I have with this game.
I personally think the Drivatars work pretty well depending on the environment. I tried career mode a few week ago to practice using the wheel, and it was savage. Properly brutal. Drivatars would dive bomb each other, and basically brake-check on slight bends.
However, if You set up a private race, with some friends if you like, and fill the rest of the grid with AI on unbeatable, for say 15 laps, the Drivatars come into their own.
There isn’t the frustration of trying to pass them all in 3 laps, and as long as you remember they aren’t actually real people the racing can be very enjoyable. I consider myself quite a good driver, but a few times I’ve been caught napping and the Drivatar pulled off a good clean overtake on me.
Also, as with the hoppers, it seems that the further forward in the pack you get, the cleaner the opposition are.
How glad I am to see you providing some meaningful ways to improve the Drivatar system, phoxxy10. Traditional AI is antiquated and could only provide so much of a challenge before something like this would have to be implemented and, without understanding the algorithms behind the scenes, you point out some possible ways that Drivatars might be improved.
I am confident that with some of the ideas you’ve raised and of course the developer’s knowledge of how Drivatars operate, we’re going to end up with some pretty darn good offline competition sooner rather than later.
True, the game does store quite a few stats and it would be interesting to know how or if this info is applied. I’ve raced against Drivatars of guys who are quicker than me but I can still beat they’re Drivatar. There’s also the problems associated with erratic and unpredictable Drivatar behaviour and this occurs even when the player himself is typically smooth and consistent. Having said that there is the odd race where the Drivatars are reasonably clean and not too wayward in the corners. At the end of the day it’s all AI so it’ll never behave exactly like the person but it’s a cool idea and I hope they stick with it and continue to improve it.
Exactly. A misconception about Drivatars seems to be that they drive exactly as their real life counterparts do. This is just isn’t true because the Drivatars are still throttled according to the user’s difficulty settings. Drivatars merely attempt to mimic the driving style of their respective Gamertags … otherwise Turn 10 could just use ghost data to replicate the gamer’s driving to a “T”, although that would make it rather hectic with multiple Drivatars attempting to assume the same driving line at one time.
Well, sure. Drivatars are associated with the difficulty that their counterparts drive at. I suppose what I should have said is that Drivatars, on all levels, are not a one-for-one representation of their associated Gamertag drivers and are throttled to make the gamer more competitive. That is to say, if I raced against the users associated with the Drivatars I faced in single player, I would not be finishing with gold medals every race as I did in the career mode.
One thing we need is for Drivatars to take mechanical damage. When last I played FM5, they could take damage in multiplayer but only cosmetic in the championships and not even that in free play. After being able to be damaged, they then need to care about damage, not wanting to bang up and break their cars. That should help the chaos quite a bit. Further, we need to segregate Drivatars based on whether or not their players run with damage on. Players with damage off would likely behave differently from players who will suffer from playing bumper-cars.
It’s not a lap-by-lap thing. It’s isolated instances of behavior. With that said, in an interview they said it logs player data up until they rewind. If the player brakes too late, locks up and slides way off into armco, then rewinds, it logs the late brake, skid, and crash, not what they do after rewinding.
I’ve always suspected the short races to be part of the problem. We start out at the back and get five minutes to try to cleanly overtake the entire field. While we’re stuck behind slowpokes running three across, we are faced with the choice of either ramming our way through or patiently waiting while the front-runners put half a lap between us. I think if we had more time to wait for clean passes it would improve Drivatar behavior.
Drivatar skill varies not by player skill but by difficulty setting. You can have a top-1% and bottom-1% player Drivatar in the same race, both on matching Drivatar skill settings. However, the game seems to select Drivatars to be your primary competition and sets them a bit higher than the field average.
I’d swear in 2013 my main rivals were my friends who started near me at the back, and they also had to overtake the field, but then it seemed like it changed and friend Drivatars stayed at the back but my rivals were selected now from the front. If so, I like the old way better because it’s more interesting and more fair. I hate being stuck at the back knowing my main rival is up front with nobody in his way.
Excellent points, IceMan PJN. The Drivatar’s nonchalant attitudes in regards to their driving machine does not do the system any favors in creating the sense of realism.
there were serious flaws in fm1 sat, especially since racing drivatar vs drivatar was an actual form of competition. it went off rmps and following the suggested line, not speed, lap times etc… to get the fastest drivatar you had to select a manual transmission, never shift out of 1st gear, and keep it as redlined as you could without blowing the engine all the way around king cobra. but drivatars are no worse than ai in other games as far as i can tell, i dont have a problem with them. my problems with fm1 drivatars weren’t with the drivatars, it was the way data for them was collected for them. to me not shifting is something i never thought of until i read it on the net. it made no sense. it was like a glitched based learning system where if you didn’t know the glitch you were handicapped in drivatar vs drivatar competitions. the fastest guy in forza should not have lost races to some noob who was bored enough to drive lap after lap in 1st gear to discover he had the fastest drivatar. i was not the fastest nor the noob, but i was left wondering why it was programmed that way. rpms win races? then 4 cylinder engines would beat any v8 and thats just not real life, not sim.