Forums

My PGG for the week was my shortest yet. The matchmaker arranged a 5 vs 4 series, us having the numerical advantage. 2 of the opponents saw 3 RS200s and my Porsche Prodrive in our team, and quit immediately, then another, and the last opponent quit 55 seconds in the second half of the Infected game. They were so convinced their chances were zero, it became a self-fulfilling prophecy, and they handed us the win.

It's all in the mind.

Edited by user Saturday, May 18, 2019 12:48:18 PM(UTC)  | Reason: Not specified

Quite a lot on my mind. I have been wondering few things between FH2 and FH4. I have been trying the same cars in both games on similar and different type of routes. Some cars performance and handling is very alike between two games but there are exceptions. I did FH2 championship in Nice with the same '46 Jaguar D-Type I used on FH4 Ashbrook Classic Endurance circuit. In FH2 D-Type is absolutely beast while FH4 version is very powerful, but also slippery. That's not an issue though, FH2 version stock is B 618 while in FH3 it's C 513. That's 103 points of difference in PI. I don't think Horizon series need to be about total historical accuracy and realism, it's IMO fine line it is, there is Forza Motorsports for serious approach.

Yet there are edge cases. I prefer driving these stock and I haven't even looked if any of the Classic Racers have much tuning options available in stock form but stock D-type Jag in FH2 can be a challenge for many modern cars. In FH4 there is much less grip and as very powerful car there is lot of wheel spinning with Jag even if player is being careful. It has other characteristics but I think that might be the most striking when trying the car for the first time. Thing is no tyres from that era or even modern era can take that much of stress and even Horizon is about fantasy, some players may feel that this sort of sticks out like sore thumb.

I don't feel there's need and I don't even want developers to start reworking PI system and this isn't anything that route creators should stress about. But came to think something else while I was doing yet another race on Ashbrook Endurance short today with '57 Maserati 300 S (B 668 stock). Best lap 5:04,204 (I think might be actually decent for change) and Time: 25:55,072. My best lap with D-Type was 5:26 but I couldn't but think how fast I could make D-Type yet keeping it in C-Class. And I came to think something from FH3 where it was possible to pick individual cars into race or championship. I didn't used that feature much though and I can't remember if it was possible to create a race / championship with mixed class. Of course there's no way of restricting players upgrading vehicles to maximum allowed, but that's not an issue, that's about freedom of choice. I do wonder if current tools allow making a race like that, because Drivatar PI appears to be dynamic. If I pick B-class Classic Racers (or any division) and select C-class racers, will dynamic PI downgrade B-class cars to C. I think I'm going to try few things and that's post for another day.

Of course I could use Drag Strip or simply race against my own time but not a race in sense what I'm looking for. Somehow I think I'm not the only one.

Oh, this is getting embarrassing. I wrote earlier how players visual experience from route may vary simply because of different sort of cars, driving styles, etc. The curve where I spotted one deer (it was right in the middle of road on the first lap) and then spotted two and today when I was going around the curve in slide front towards inside I saw there's entire herd of deer in the woods. This is what I get for not using 3rd person view but in Playground games. I think that's enough of Ashbrook deer, I'll stop, I promise.


CUSTOM ROUTE REVIEW

SIN CITY 2

Map: https://i.redd.it/ju2fw0r6qew21.png

Creator: DMFDXGAMER
Starting location: North City Cross Country Circuit
Note: Despite starting location being Cross Country event, this is all tarmack city circuit.
Note 2: There should be 3 Blueprints available for this circuit with different lap count, but search came up only with this.

Event class: Road Racing
Type: Circuit
Restrictions: Anything Goes
Length: 6,9 miles / lap. 2 laps total.
Season: Summer
Time: Night
Weather: Rain

Creator's topic in his sub: https://www.reddit.com/r...through_the_city_i_made/
Active discussion at Forza Reddit: https://www.reddit.com/r...t_race_through_the_city/

I wrote earlier in this topic how I had great time with custom route called Dirtschleife (Dirt racing) and how many custom routes I tried felt perhaps somehow convoluted for one reason or another. Ashbrook Endurance is great route, but Drivatars are way off on it. This circuit, Sin City 2 was again one of the better racing experiences for me in FH4 for a long time.

I did this with lower class car (again) '70 Chevy Chevella SS (Super Sport) 454 (C 542 stock). I play on Expert, use manual without clutch and have driving aids off except for braking line. I keep Rewind on and this time actually had to use it couple of time. I don't use 3rd person view but during Playgroud Games. If possible I made different choices and made another run on this, but I wanted to share this today simply because this is a really great circuit!

Circuit makes great use of Edinburgh. There are various kinds of curves and despite not perhaps looking like that on map, some are almost like hairpins, tricky too because there are non destructible on both sides of the road at some places some your crash head on if turn gets too wide. There's uphill, downhill. Check points are placed pretty well. Even this is a long route there weren't but couple of places where I didn't figure out the turn before it was too late. On second lap I came to think that if I were using 3rd person view, I think I might have gotten them earlier.

I didn't try it but for example, I think the very first right - left - right after the start where you come more or less fast when starting second lap, I think those checkpoint markers may align quite nicely in 3rd person. That said, I felt this was a bit of technical track, despite that, I don't remember single blind turn even I use first person view. There was one checkpoint which looked quite narrow but I didn't noticed it causing any issues at least on couple of laps I did.

One things I felt attributed much to my experience was sections on cobblestone. If you checked the map and know Edinburgh, some bends leading in and out and 4/5th circle looking part are cobblestone road. So they are nice bends per se, but what adds to the experience, they are wide and cobblestone feels different from tarmac. I can't but make short comparison to FH2 where there is force feedback and all on cobblestone but I haven't really felt if it does anything to grip really. In FH4 I'm not so sure that it doesn't have some impact to handling. Lost control while pulling drifts on those sections but that might be also because of rain. One thing I'm trying to keep in mind from this circuit is how to avoid repetive design. Some section like in this may have several attributes, 1. Wideness, a bit of time to relax _or_ 2 drift, look for pass, these are opportunities. 3. Surface gives different response to player.

Drivatars and Race

I didn't noticed any bottlenecks made for Drivatars. This was actually tight race and I which I have surely lost if I haven't used Rewind on couple of places I messed up during second lap.

I messed up my first lap pretty badly, spinned couple of time, was rammed by Drivatars coming behind. This was Anything Goes type race and there were cars that I thought were better than my stock Chevelle SS, at least had higher PI. I kept making some small mistakes and even fell to 9th position at one point during first lap. This rarely happens but it was actually great! Especially because of certain feature concerning Classic Muscle division. LOL

Luckily this is a long route and I managed to gain positions again. I kept making some small mistakes, sidelining this or that and was on second position when but came too fast to fist curve, had to rewind and there were other situation but it was quite a close battle for the first position vs. '60 Chevrolet Nova and I barely made it.

Numbers:
'70 Chevrolet Chevelle SS 454 (C 542 stock)
Best lap: 5:32,725
Time: 11:27,255

Best Drivatar: '60 Chevrolet Nova SS (C 582 which appears to be stock)
Best lap: 5:40,448
Time: 11:27,890

So even my best lap was a quite better than the best Drivatar in the end it was close race and I even had to use Rewind. This is IMO because this is a great circuit. LOL

Before I did this route raced Haybrook Park Circuit in Edinburgh and other long circuit on southeast outskirts of Edinburgh in Classic Muscle division with the same Chevelle. Results were that I was seconds ahead and while listening to music in rewards screens (and my mind filling in the blanks ("... all you want is to play at playing god"). I thought how ironic is that.

I seen stock Oldsmobile Hurst/Olds 442 mentioned being some sort of God mode in it's division and it definitely is in FH2 on any road race, but so is stock '70's Chevelle SS at least up to Pro. They have different kind of character but once you learn them, you are practically playing in God Mode. I don't know how many of these exists in Classic Muscle division. The issue lies with Drivatars.
In FH4 it's much the same, only difference being, not the cars, but routes. Drivatars benefit from narrow short courses where it's difficult to overtake to begin with. To make interesting races, routes which are perhaps a bit more technical than default routes like this Sin City 2 are needed.

I really liked this track and hope others will try it too and perhaps post their own impressions. For cons, despite this being Anything Goes and lengthy route, I think this may not be great experience with 1993 McLaren F1 (S1 826) for example. There is rain, there are some curves which might be frustrating with anything that has a lot of understeering. I could imagine this being really great with up to B, perhaps lower A class Japanese cars for example.

I noticed this circuit has quite a few likes so I'm not the only one liking this. I hope everyone who find this to be good experience gives their likes to author too.

I'm not going to directly quote your post, but the PI changes based on the upper limit of cars. As manufacturers progress, and modern sports cars become the super cars of the past, it pushes that boundary.
.100-500 for D class has a huge performance gap between the cars but the reason above is why.

The D Type is still ridiculous in a straight line but it is becoming increasingly less grippy as time goes on. Even with the racing slicks.

As for the Ashbrook deer, I'm glad to hear that their population is steadily climbing 😂

I'm not sure I got that. So Drivar PI can go upwards but no downwards? Like if create Anything Goes type of race and I can select individual cars from categories D, C and B for example, then in actual line-up Drivatars cars, Like D class Porsche Spyder become C or B class with rest of Drivatar line up depending from car I choose for race? That would also mean that's there's no way to create mixed class race?

Checked Forza Wiki and it appears that there's perhaps some history developers having fun with D-Type.

Easy to believe it still being monster on straights in with better tyres FH4, it just had that feel.

Tried to do some tests on cobblestone and tarmac if there's any difference in grip.

I chose formidable weapon Abarth 595 esseesse stock (D 100) to burn some rubber, for exact reason that not being perhaps not it's strongest suite. Drove to Edinburgh and tried going from full stop to full rews on tarmac and cobblestone. If there's any difference, it's that front wheels spun maybe a fraction of second longer on tarmac. Tried the same test with manual clutch by getting about max stable revs and releasing clutch, same result.

Did same tests with Honda Prelude stock, no notable difference between tarmac and cobblestone.

Those are just two cars and just my ability though. Anyone else interested giving cobblestone vs. tarmac a try?

Edited by user Tuesday, May 21, 2019 12:35:08 PM(UTC)  | Reason: Not specified

Did another run on Sin City 2 earlier this week.

I used the same stock '70 Chevrolet Chevelle as before. No rewind, just wanted to see how it goes. Won it but it was actually pretty close.

Best lap: 5:35,575
Time: 11:21,838

Best Drivatar '90 Mazda RX 7 (C 651)
Best lap: 5:34,001
Time: 11:22,533

I will definitely get back on this circuit. I did some mistakes but this is what I actually like, learning new things and for me this circuit takes me out from my comfort zone. I think I may try this on pro too. In previous weekly playlist topics there were people telling how Drivatars messed up on Community Route Championship, or whatever it was called. Saw that myself too and then players who tried them on Pro or Unbeatable difficulty reported that Drivatars were even worse on higher difficulties. It's interesting to see if that happens on this circuit too.

Still can't get enough of Ashbrook Classic Racers Endurance Short. This time I got '57 Ferrari 250 Testa Rossa stock (A 704). Best lap 4:59,933. Great vehicle. I really liked to try long version of circuit but it's 15 laps and even I really learned to like these Classic Racers and like lower class cars in general, 15 laps on longer route is bit much for lower class cars. This car might be it though.

Funny, one of the races I'm putting into my racing series is a take on le mans. Obviously not going to run it for 24 hours and will only be a two or three lap race, but I think you would thoroughly enjoy it.

Trying to take in a mix of everything and hit every theme once. I'll shoot you a message start of June to give it a shot if you are free 👍

Be sure you do, I'll do my best to find time. :-D What is your approach on Mulsanne straight? I mean history is what it is, but what's your take on what would make a better Forza Horizon experience?

Regarding Classic Racers, or Vintage GT or division like that in FM7. It's sort of obvious thing to try. So I took '53 Ferrari 500 Mondial and '56 Jaguar D-Type to track just to see how they feel like in supposed simulation.

The first thing is homogulation. Didn't recall how to make races work without homogulation (or if it's even possible). So I raced then with whatever parts, I think both cars were defaulted to B-class in events I tried. I was quite disappointed. Mondial felt a bit stiff like in FH4, D-Type felt much the same with weird gear ratios. Actually it comes down to the Car Model which is Turn 10 is basing all the cars in game. That is sensible and clever approach. However there's a point where every car starts to feel like Go-Kart with R14 to size R17 wheels, weird. It reminds me of old PC game, Sports Car GT which sort of gave impression of real car, but in FM7 it's like it's there some sort of way and and then disappears and I'm left with a weird Go-Kart. Idea is kind of genius but weird in a sense that I don't get a sense of a real Go-Kart but something between real one and one and sort of which has a look of perhaps Formula One car via car plastic body, has an engine and all, but doesn't actually have driving features of real Go-Kart.

To make something fun, it doesn't always take perfect simulation anyway and I think Horizon is in the sweet spot regarding physics for what it's trying to do. Many cars actually have character and whatever creative freedom Playground Games took, how emphasised something may be or whatever compromises they did, I'm having fun with this. No need to take this to Go-Kart and what would be the point making '50's racers feel like... chicken at various speeds anyway. Or moles, maybe one day we could get a Gears of War racer where we ride on different on ferrets trying to catch moles, but then ferrets turn out to be bad guys too. Oh, everything is lost, but here comes the player, charging in with VW Bug Chainsaw combo (don't ask) to save the day!

Anyway, for custom routes, maybe there were more players interested some niches, like Classic Racers if there were good routes to race them, some hook to get more players interested.

Obviously the terrain isn't nearly as flat as la sarthe but it's something I have given some serious thought.

I'm currently thinking of making the race 330 P4 Vs GT40 but I think a vintage racer event including cars a few decades before the aforementioned would offer a wider field. I think the GT40 would dominate the circuit I've made and take the magic out of it.

Anyway thanks for being a contributor on this thread.

Edited by user Friday, May 24, 2019 12:48:52 AM(UTC)  | Reason: Not specified

No problem. I very much appreciate someone is putting some effort for creating better Forza experience.

Single player wise, Mulsanne straight is IMO difficult to get work without chicanes, like it used to be in real life. I haven't tried GT40 in FH4 but it has that habit being quite awesome across different racing game franchises. Just thinking out loud here: Let's assume for the sake of conversation that in SP and MP especially, MP, GT40 would be the best option. There's IMO not much that could be done to that in MP side of things. It might come to creating different kind of opportunities for different builds in general, different sorts of technical sections but issue remains if car is sorta more than sum of it's parts, like getting into corners fast and accelerating from them fast. Given pretty decent top speed too, difficult.

For single player it's perfectly valid choice to get car that isn't super car in it's class / division just se see how it goes. And I have been running more laps on Sin City 2 couple of things in mind. I did couple of races against Pro Drivatars and then dropped again back on Expert and made more laps aiming with best times I could. I'm not going to post lap times yet, it's quite late for me when I post these things and without Rivals it's pen and paper but I post them later. Note however, no rewind was used on but one run. It matters because it may mess up Drivatars.

I can't say this is conclusive but....

Expert and Pro Drivatars behave much the same. They didn't mess up anything on my Sin City 2 races even after I increased difficulty (like happened with some Community championship routes on Weekly playlist).
They appear to be sort of rubber banding, but not in a way people commonly think. I run slower race on Pro, about same lap times as expert and it was a very close race I won only perhaps instead Drivatar going to clean overtake coming to final straight rammed me instead of getting on vacant lane to the right.

I did second run and was able to shave seconds from my lap times, yet Drivatars were able to keep up. When I decreased difficulty back to Expert Drivatars and kept lapping same times Expert Drivatars weren't able to keep up with those times anymore EVEN when Expert Drivatars had more PI advantage than in races run on Pro difficulty.

One more thing I noticed. There was indeed couple of sections where I gained quite a bit to Drivatars. Pro Drivatars where for my experience faster on long curves on Pro than on Expert, which I think is one of the differences that would make a lot sense but other than that, I have come to think that certain kind of series of bends a bit like S turn to player advantage if player only knows how to use weight transfer and gained momentum for advantage. For me it looks now that Drivatars achilles heel might be there, Drivatars may not really be able to do that.

For route creators, who think of single player aspect too, this could perhaps used as advantage, not to take Drivatars out of the picture, but give wider variety of vehicles a real fighting chance. Perhaps other thing to think about is that for me Drivatars at least up to Pro may not rubber band the way people commonly may think. They appear more dynamic when difficulty is increased but they may work sort of opportunity window for player win / Drivatar win and when difficulty goes up player is perhaps given less opportunities to win, or window for that is smaller. Having many sections that fall out of Drivatar AI strongest features may make race trivial for human player, even if that's not intended.

But like I said, this isn't anything conclusive, just something I have been theorising. I hope I can get back with numbers and even more numbers later on this week.

Winter is almost over and we can all get back to tuning cars again 😂

So RetroKrystal asked in a topic "what does FH4 mean to you?" I wrote a looong wall of text before I figured out the hard way she wanted only 3 words. I thought that, instead of erasing them from memory, I could recycle my thoughts by posting them in this very topic, so consider this post the answer to a "Can you elaborate on that?" that nobody asked. Enjoy.

Edited by user Thursday, May 30, 2019 6:31:42 AM(UTC)  | Reason: Not specified

That is considerably longer than 3 words. Made me cry laughing though. So thanks for that? 😂

Took a while longer than expected to get back to this. In my previous post I gave some impressions how Drivatar AI work in general and some thoughts regarding Route Creators. I promised more numbers but before going to that I just feel this need to open up about things in the real life. Going to visit family, seeing kids getting their grades and starting their summer holidays, older kids graduations, meeting friends, chatter, good food, board games, late night barbeque, red wine. Lot's of really wonderful and awesome things in life.

Then there are things like going to loop 60 miles per hour in a race in the middle of the city, engine roaring and tyres screaming in four wheel slide. So do Forza (Horizon) come even close to things that really matter in my book? Not even close. But it sure has it moments. :-D

Then back to Sin City 2 circuit one more time.

So numbers thing. I did few races last week and more on this week races to try things out. With my '70 Chevrolet Chevelle SS 454 (C 542 stock) both on Expert and Pro.

On Expert my first times I posted were: Best lap: 5:32,725. Time: 11:27,255.
Best Drivatar being: '60 Chevrolet Nova SS (C 582 stock) Best lap: 5:40,448 Time: 11:27,890.

And my times stayed there for a while but since I then I started to get lap my laps around 5:26 - 5:27. Nothing really dramatic here, it's a long circuit and Chevelle isn't sort of optimal car for the race but that's exactly what makes it fun. LOL

My best time on Expert was 5:26,660 and 11:17,528 for the race (slipped to the wall on first lap). Best Drivatar was Lancia Delta C 533. Best lap: 5:34,948. Time: 4:20,765. The important thing is that I finally got a different Drivatar line up than during past week.

On Pro (again against different Drivatar line up than last week) I did couple of races, was pushing it too hard at times and best lap being 5:27,711 overall time being 11:08,422. I managed to lose that one major mistake was all it took to lose to best Drivatar Mitsubishi Galant VR-4 C 547 (so slightly tuned, stock is C 532), which best lap was 5:31,014 and time: 11:07,701.

On second try on Pro I won the race with best lap being 5:27,870 (slightly worse) but time 11:05,918. One rewind but I was leading already so nothing to mess Drivatars up. This time best Drivatar was Mitsubishi Eclipse GSX C 543 (stock) which best lap was 5:32,691 and time: 11:09,129.

Over time I of course learned more about the circuit and car (it's about discovery! :-D ) and while following Drivatars closely I learned that weight transfer etc. doesn't necessarily gimp Drivatar AI as bad as I originally thought (some Divisions may have it worse than others though). One additional note regarding track I missed originally, there's pretty good short cut straight at the start which doesn't matter that much on first lap but on second lap player can benefit from it a lot but not as much as what happened next.

I thought I had enough data regarding Drivatar line up and finally took the same Mitsubishi GSX (stock) to circuit that came second on my previous race. The result wasn't really a race anymore, it was a slaughter. My best lap being 5:16,995 and time: 10,45,574, no rewind. Best Drivatar was Triumph TR 6 C 549 but numbers regarding that don't matter. I got too much lead and in the end game compressed Drivatars race times so there's no way of knowing what their actual times were, only that they really weren't in the race.

My conclusion is that Drivatars do rubber band indeed, but not in a way it's usually associated and that rubber band may break at least on Expert and Pro. Drivatar AI's main goal isn't winning the race, or the best time. Certain time bracket where AI aims my be present, but main function of Drivatar AI is to challenge players in entertaining way.

What happens in Horizon Drivatars under the hood is out of hands of route creators, though when designing routes for certain divisions, Classic Muscle and Classic Racers design choices, emphasising design choices which doesn't hinder Drivatar AI and doesn't bottleneck the player either might be quite important for creating enjoyable Horizon experience.


Summary (I lost about third of notes but this is IMO enough to get a picture)

Chevelle '70 SS Stock Expert
11:13,624 Won
11:09,803 Won
11:17,528 Won

Chevelle '70 SS Stock Pro
11:18,079 Won
11:08,488 Second
11:05,918 Won

Mitsubishi Eclipse GSX Stock Pro
10:45,574 Won

Hello again.

Quick question for anyone using them, I've never owned a force feedback wheel and pedals. I'm looking at the G920.

Is it worth the price and would you say it greatly improved your playing experience?

I'm a super competitive driver and I am fully aware it will be like learning to walk again, but is it worth it?

Thanks

Planned to highlight some new custom route for today. Didn't had time to try longer ones I have in my bookmarks but found interestingly looking Road Racing Trail from Reddit. Problem is that it doesn't show up in my game no matter what search parameters I try and all that comes up with Gamer Tag search is older blueprint from him which is a race on default circuit.
This is the Custom Route I mean: https://www.reddit.com/r...r_around_edinburgh_city/

Why this route doesn't show up? Could anybody else confirm or deny if this is a bug or something else? Because if it is a bug I will submit a ticket which then does or doesn't lead to something.


But I did some racing in Edinburgh just to check how FH4 default routes feel against Sin City 2 and really. Took stock '70 Corvette ZR-1 against Pro Drivatars and it was a bit of catchup at first but then in the end piece of cake. Princes Street Garden Circuit isn't bad IMO, it's just that these get a bit old over time. I guess Sin City 2 is bit long for those who look for fast races but IMO it's a great circuit. Noticed yesterday that it has got some more likes. Don't know if this topic has anything to do with it, but it's great to see that there are players who express their support for those who spend time and effort to create races with a bit more variety and technical aspects.

I still kept thinking of Drivatars after my experience with Sin City. I left out that I had suspected Drivatar behaviour for a long time based on my experience on default routes. This is for turtleCZ whom I hailed from other topic but of course anyone can offer their insights.

So rubber band is reality which IMO isn't a bad design choice per se. I think the goals developers are trying to achieve are okay. Drivatars may need some tweaking but in the end it's easy to create an AI that is perfect in winning or losing but anything between is difficult. In the end my general experience is that FH4 Drivatars reward clean and fast driving, I think it also good tuning.

This is all just pure speculation.

Maybe races within Edinburgh city may turn murder to Drivatars because they can't utilise ability to pass soft obstacles (fences, cones, etc. breakable objects) because they aren't really a factor in city races where there are lot of solid obstacles (buildings, iron fences, etc.) so if Drivatar AI depends a lot from ability to pass soft obstacles unhindered under certain circumstances and that benefit is rendered null by environment and that while Drivatars appear to be able to pick closest to straight line in low degrees S-bend for example and use perhaps limited weight transfer without going to two wheels or something ridiculous and has limited ability to slide and recover but perhaps no ability to powerslide, then there are suddenly quite a many things against them. On positive side racing with cars that may not be ideal regardless of PI might turn out to be very fun, but negative aspect is that picking car which is close to ideal may turn out to be plain murder to AI.

Open question: What are within Drivatar AI skillset regarding alternate racing lines? For me it has appeared something quite random.

Tuning aspect of this is that IMO which works even when players don't necessarily always like the results. I do have little experience of this though but for me it looks like players who complain about impossible Drivatars may have the right PI but high PI number per se, doesn't mean fast car in every situation.

This weeks Trial is IMO a good example. Took '92 Honda NSX, upgraded it but I didn't pay attention to series and went to it with race tyres. Won the first race and was terrible at two others. Second try without with proper tyres but without AWD swap, not very good either. Last effort was with AWD swap, bit less horsepower, tweaked suspension and in the 3 races I was 1st, 2nd (leader was on our team) and 1st. last win being probably the largest lead I have ever had to Unbeatable Drivatars. So in MP events performance of others players in team are very good indicator if tune is good or not, but in single player it's more difficult, there isn't perhaps such a clear indicator.

Myself, I do things like some past seasons Dodge vs. Chevrolet Playlist events with Dodge SRT Demon, which has the look and feel I sort of like and which is also sort of terrible 3 ton tank with too much power and too little grip in it's stock form. LOL

Tyres are always a big upgrade, then some other things, but I didn't went for top PI, just for something manageable. Needed to tweak it but in the end, these are risks I take for the sake of fun. And if it weren't worked out I could have just picked Viper to those races. I mean it's really awesome feature that we can take say '82 Porsche 911 and upgrade it to A class so we aren't always racing against just Lamborghini Jalpa and other 911's or that cars from C or B class can be upgraded to whatever but Drivatar system IMO works already towards not frustrating the players. It's not that players aren't good either but Drivatars may get unnecessary flak just because of tune isn't working for particular race or series.

What are within Drivatar AI skillset regarding alternate racing lines?

With their 13x mass they rarely go off the suggested line but when they do, my goodness.

I have seen them "panic" turn into walls, trees, drive down the wrong road and all sorts of other hijinx. It's really quite hilarious to watch.

I am here! I think many people have more experience. When I play on lower difficulty AI looks like cheater to me and I am not sure why. For instance, in the first Horizon if you are way ahead of AI and do some mistake AI is moved faster behind you. It's not in 4. What I found on unbeatable is AI doesn't cheat at all. Maybe it's because I see them around me. They have probably faster cars but worse driving line, so it's not so hard to beat them. Can somebody confirm they use faster cars with same PI? Maybe it's tuning only and maybe it's a cheat.

Overall it looks like unbeatable is time based than player based. If you have better time you can always beat them. It's probably different for lower difficulties. I haven't seen any waiting on me on unbeatable. There is a change unbeatable is normal racing with a bit faster cars and not so bad driving line. And lower difficulties are some mix of "cheating" for slower players. With a bad car you are not able to win because it's too slow, like some cars are too fast. I am not sure about it.

Well you are right and wrong, yes AI is time based for example those Focus etc cant run goliath around time 8.10 (their best laps are usually something like 8.14) so if use car like 599XX Evo can beat AI over minute per lap. But those are also in player based i remember long time ago when there was warthog trial someone show he can beat AI when racing against them in single player and he did that track abaout 2.09 and was able to win, while same time if i run same event and i do 2.09 my finnish position is maybe 10th i remember AI is beat me in that race even when my time was 2.04 and it take something like 20 trys before i was finally able to win that race.

Well, since I can't find a Team Adventure to play Games with to save my life, I made a Dirt circuit. The events are called Mountain Foot Rally Circuit and feature 2 laps.







It's a relatively long track so, lap times should range between around 3'30'' for S2 class to 4'15'' for B class. It has a little bit of everything: you start downhill, then climb your way back through water, long straights, hairpins, realistic-sized bumps (not the usual massive jumps...), even a rail tunnel.

Wow, so what do you think it is? Some random track time (sometimes better, sometimes worse)? Or random "cheating" power added?

And if it's player based why it was so hard? How the game knew it.

Edited by user Sunday, June 9, 2019 6:37:10 AM(UTC)  | Reason: Not specified

I have seen similar things happening but I don't have experience if it's the same on Unbeatable, I play against that only in Trials. On lower difficulty levels this appears to be another Achilles heel for Drivatar AI.

Hilarious things like Drivatars driving to a tree or not so fun sideswiping human players in Trials may perhaps also happen because of other reason. There is "hidden" clutch for players which is actually very obvious if you stop your car let it sit idle for a while and then give some revs. You can go up to 1000 - 1500 or so, may depend of a car before it moves anywhere. It's even more obvious if you are on first gear and park on uphill section, car may start rolling backwards downhill even though player gives some revs. It's not a bug but feature that contributes a lot towards what makes Horizon cars fun.

But what if Drivatars doesn't have any sort of clutch at all? I seen clips from FH4 where Drivatar going on old Aston Martin pulls wheelies accelerating with everything it has, eventually getting on bend with right front wheel on air and slipping to non destructible bridge support due to lack of steering. :-D

I really don't know if this is related or not but when I grinded 49 more Head to Head wins for Star Card I started seeing Drivatars sort of trembling, or jumping after they had sometimes gained a bit more lead to my car and I came to think if those little jumps or shocks matched speed, powerband and shifting pattern.

Then during this weeks trial on last race I was in 4th position shadowing my team mate who was 3rd behind two Drivatars and I was just following, guy had been constantly good and I guessed he looked for opening to pass. We were all accelerating from bend to straight and shifting on higher gears and then I saw Drivatar starting to have that familiar stutter, not block Drivatars on Unbeatable can try, but stutter and perhaps then it tried to block but result was like Drivatar car just stuttered sideways right to my team mate who was going for pass, knocked him right out of track to left and it looked like there was a tree. (We won the race however and guy driving black Honda, who got rammed, even scored 200 - 300 or so in the end IIRC).

If, and this is of course a big if, Drivatars don't use any sort of clutch but get always on peak power without delays, that would compensate for other things but it would also perhaps be a double edged sword causing other problems like going to wheelie and losing steering. That may lead again in another issue where Drivatar cars don't get full benefits from upgrades, and overall also creates hazards to human players in certain situations.

In the end human mind has tendency to rationalise and create associations to meet that goal, like associating odd Drivatar behaviour and shifting. In the end I don't know if there's any way of testing if that actually happens or not.




Interesting!

For Trial events I have experienced the same in Showcase Remix with Healey from few weeks ago. Took several tries from me but what I read from the forums there might have been other players who got it with worse time. I recall there were discussion that it was somehow inconsistent.

Wild guess: What if there's also something like "super rubber band" that works based either individual performance or actives when players enter in certain performance bracket? If that sort of thing would remain active during Showcase events that sure could mess them up for players who constantly get times in top 33% (could be anything) in races category for example.

I have played Horizon 3 and 2 and for my experience Drivatars are less ram happy in FH4. I actually got FH2 just couple of months ago and noticed that most of time driving stock was good enough to win against Expert and Pro. In FH3 used to create championships for lower class cars and race them stock and my memory is that races with say stock Porsche 944 on Expert could be tight. In FH4 I think there are more variables simply because how FH4 simulates driving appears to be somehow finer. That creates some learning curve but in the end it's also tempting and I feel I have more control I can use to my advantage taking alternate racing lines for example. I can't say anything really conclusive because I mainly race D and C class events with stock cars. Yet at least on Expert and Pro Drivatar AI gives player a crutch by not going anywhere as fast as car is able which is fine because this rubber band is dynamic other way around too, but when players competing against them are actually fast crutches AI use to it's advantage aren't always enough or doesn't perhaps work like intended.

I have wondered some times if it's like players have more options, we have better game in that sense, but Drivatars are still FH3 Drivatars or something like that.

The big context for all this pondering is actually Route Creator. I still keep wondering why Pro and Unbeatable Drivatars messed up some custom routes (from weekly playlists a month or so ago) worse than Expert Drivatars. If there's any sort of hint that can be measured regarding that crutches Drivatar AI uses doesn't really always work to their advantage I guess it's that. For route creators this may also mean that only way to make sure everything works like intended on all difficulty levels is to test them on Unbeatable.

Thanks! Will try.

Just pure cheating, AI can hit times that humans cant do or do something really funny like driving motorway 430km/h with Ariel Nomad. And i would guess that game tracks how well we race and fine tune difficulty base of that.

@BabySeal363 Nice track, it was really fun to drive :)

You're welcome. Judging from the number of likes it's getting, it seems you're not alone in this judgement. In other news, I finally managed to get ranked in both Team Adventure and Team Games.

Tried BabySeal363's Mountain Foot Rally Circuit's.

Pretty freaking good in comparison for many that are out there. I did Classic Rally, Retro Rally, Modern Rally, Rally Monsters and Anything Goes categories of circuit. All on Pro difficulty, didn't try Unbeatable yet. I Used variety of cars, all stock except Hoonigan RS200 which I had tuned to S2 981 for some Fortune Island PR stunt couple of weeks ago.

Visually circuit is very pleasing. I think players are more or less familiar with different sections, now we have them here combined in one longer route. That all also contributes towards good racing experience. Elevation changes, water crossings, small jumps, some tarmac, a bridge, different kind of bends and the tunnel. It's possible to get quite fast racing line into tunnel and accelerate all the way towards to end. Racing out from the tunnel to the light is quite awesome experience on higher class cars because there's wall still running on right side. It gives really good sense of speed how fast you are really going which is sort of difficult for games because of FOV limitations and that when we keep doing these races we get used to it over time. I felt circuit really shines on faster cars which I felt was some kind of bliss (See! Isn't this relaxing!). My best lap with S2 Hoonigan RS200 was 3:26,790 which is within time bracket author mentioned.

I noticed some issues too, which interestingly appeared to create more difficulties on Drivarars on D and C class, than on higher classes. More about that further in this post.

Took '68 Lancia Fulvia (D 490) to Classic Rally scramble. Some background about Fulvia line up. It was very successful line in real world rallying back in the day. I don't recall which years model was in Gran Turismo 5, but '68 Fulvia in FH3 and FH4 feel pretty much the same as in GT, which for me is awesome. For curious to try out my best lap was 4:43,350 which was about 11 seconds better than Drivatar Talbot coming to second.

I think quite a few of us can ever get any chance to test drive real '68 Fulvia and I'm very pleased with how it's modeled so I encourage anyone curious about racing history to try it out and this circuit offers so much variety that I think anyone can figure out why that car is such a legend after trying it even just once.


Retro Rally division circuit was most interesting and fun experience I have had for a long time. I took the worst car on purpose, '80 Renault Turbo 5 (C 512). It's important to know that Renault Turbo 5 is the only RWD car in Retro Rally division. All others are AWD cars and that of course creates actually interesting situation with Drivatar lineup in the race. It actually took two races before I managed to win with it. Turbo 5 is old fashioned engine and on uphill sections very dependant of keeping it on right powerband. One good lap isn't enough with it at least on Pro. I won when I got my best lap and my best overall time, lap being 4:44,801 and time 9:39,680. It was very close though Drivatar becoming second having time of 9:40,181 and I noticed that below on results there were Galant VR-4 with best lap of 4:40,915 * see General interest.

Doing races like this has certain perks. For me it's not about missing rivals or actually even about how good or bad I'm in FH4. For me this is about glass half full vs glass half empty. I think it's important that people communicate about their wishes and disappointments but in the end game is how it is, maybe something regarding end game may happen maybe not, but as now, anyone can take that RWD Renault 5 against AWD Drivatars on circuit that's everything but not easy with that car and see how close to say that 4:40,915 there is possible to get with stock build, or whatever. Or taking that Mitsubishi VR-4 and see how fast I can go with it (Best lap: 4:33,140). Definitely going to try again with Renault.

I raced this circuit also with B and A class. For B class taking Ford Focus '03 stock on this may be actually something like race but even better in Anything Goes version with '73 Porsche. For A class it's more difficult to say. Subaru WRX was fun, perhaps Porsche #185 959 Prodrive might be even more fun. If time is limited perhaps skipping A class and go trying S1 instead might be better experience.

Great work that there are possibility to race this circuit on many classes. My last race on this today was with '73 Porsche 911 Carrera and it was definitely interesting. I liked dirt trail called Dirtsleiffe or something like that I liked a lot, but it's also very long. This was one of the better experiences with '73 Carrera and another thing I recommend player who like challenges and lower classes with also a bit of historical accuracy to try.


General interest

Checkpoint 11 (if my count is any good) right after first water crossing is a bit too narrow. Drivatars which car can benefit from torque and weight momentum on crossing (coming from straight tarmac section) seek to pass player on there but slow down and fall behind because gate is a bit too narrow. * In other hand this may be saving grace for those who race this with '80 Renault 5 Turbo (stock).

Racing line starting near first water crossing going uphill and turning towards tunnel section is IMO set pretty close to optimal, however Drivatars don't appear to utilise it to it's full potential. Why is that happening?

Lost count of check points (and replays bug out on after 19 till show them again on long tarmac section) but after second water crossing and after short tarmac section turning to gravel to the right, Drivatars have issues on some checkpoints. I recall I gained a lot of lead on some of my B and C class races on that section and there may be two reasons. I think one reason may be that Drivatars going in group can't pass trough gates like that and lose time in negotiating their position or something.

Second issue which for some reason it doesn't appear to gimp Drivatars in upper class races but get's much worse on lowest class. In D class Anything Goes race I did with '65 Alfa Romeo GTA two Drivatar cars couldn't finish the race at all. When I lapped Drivatar Terradyne Gurkha I noticed it was going uphill right between the gate, then reset itself and tried the section again, and again. There were no other traffic, it had enough room. Other Drivatar with M-B Unimog got only one lap but then got stuck probably at the same place. DNF for both, I waited for minutes.

Section where circuit get back turns right downhill bridge on left side. Turn itself doesn't appear to be too steep for Drivatars but checkpoint is so narrow that Drivatars may slide out from track either because they collide with each other or try to avoid colliding with each other. I don't know if this may only occur on higher class (S1+) races.