The final frontier in our Project CARS coverage? I believe so.
To future-proof myself for the release of Gran Turismo Sport, and the inevitable inclusion of the Canadian Football League in EA’s Madden NFL franchise, I went out and purchased a used PlayStation 4 sometime last week. Now I’ve owned a PlayStation 4 on two separate occasions, but each time parted with the system due to an inherent lack of compelling games. Look, the PS4 is honestly a great platform and I feel it’s the one you should invest in during this console generation, but when I saw everything Need for Speed: Rivals had to offer in just under four hours, it was time to let someone else have a turn.
Armed with a Logitech G29 and a bit too much free time, I grabbed copies of both DriveClub and Project CARS – a pair of titles known for high sales numbers yet poor reception from their respective hardcore communities. Poor choices to begin my third phase of owning a PlayStation 4? Possibly, but while many PRC.net readers will crucify me for giving money to Slightly Mad Studios, Project CARS does give me the most racing action I can possibly receive on one disc for the $50 asking price of the Complete Edition, and that’s important during the initial startup phase where the cost of acquiring a new platform can skyrocket quite quickly if you’re not careful.
We’ve covered the various faults of Project CARS many times here on PRC.net since the title’s release – from the overzealous fanboys aggressively marketing the game on a ground level basis, to a full critical analysis – but it has always been with an extremely cynical tone. Sure, it’s done in part to appeal to the crowd who show up here seeking validation for their feelings of frustration with the game, but make no mistake, when I start to find cracks in a product, I usually tear it a new asshole and go on a classic tirade that’s since become a staple of PRC.net feature-length articles. But after spending something like $370 CDN on the startup of a new console, that mentality was no longer an option. I actually wanted to sit down and try to have a bit of fun with my new purchase, warts be damned.
With a five hundred kilometer off-road endurance race on the schedule this weekend, I figured I needed some practice sitting in front of a toy steering wheel for extended periods of time, and using Project CARS to run offline marathons in preparation for the event seemed like a good idea on paper. I’d get some use out of both my new console and new wheel, as well as understand why many people are still choosing to sink time into Project CARS as their preferred simulator of choice despite the laundry list of grievances even the most tolerant of sim racers seem to have. The race format would be fairly simple; a ten minute, balls to the wall qualifying session, followed by an authentic, full distance race at 100% difficulty in my series of choice.
And obviously, because this is an article on PRC.net, things didn’t go so well.
Race #1: 1996 SCCA Trans-Am @ Sears Point (39 Laps)
If bipolar disorder could manifest itself into any kind of top ranked auto racing series, the SCCA Trans-Am championship would fit the definition almost perfectly. With periods of popularity and prosperity contrasted by an equal amount irrelevant seasons, the American road racing championship featuring race spec pony cars from General Motors, Ford, and Chrysler has been a stomping ground for the likes of North American road racing aces such as Boris Said, Ron Fellows, and Scott Pruett. With no pit stops anywhere in sight, and races contested over a relatively short distance of one hundred miles – producing just under an hour of competition time – this is the closest the United States have ever gotten to reproducing a series akin V8 Supercars on the eastern side of the Pacific ocean.
Project CARS features the championship winning 1996 SCCA Trans-Am Ford Mustang as prepared by Roush Racing and piloted by Tommy Kendall, which pumps out upwards of 800 horsepower and is connected to the tarmac not by pure racing slicks, but instead with a set of bias ply tires – a combination that’s just a little south of sanity. In the past I’ve commented that while some cars in Project CARS are atrocious, others can be quite good, and the Trans-Am spec Mustang falls into the latter category. Make no mistake, this car is outright dangerous on cold tires and exhibits handling characteristics akin to a poor iRacing release, but properly warming up the tires on your out lap warrants an experience that feels like you’ve stepped into a GT3 entry that’s trying to unleash its inner American stock car – more power, more weight, and less grip, but still somewhere in the same ballpark as a Callaway Corvette.
Sonoma Raceway is a track I know like the back of my hand thanks to its inclusion on the NASCAR Sprint Cup Series schedule, and this created a situation where I utterly decimated the artificial intelligence in qualifying by over a second. There were no instances where the bots circling the track even stood a chance compared to the times I was putting up; hardly indicative of the 100% AI difficulty I had selected prior to commencing the event. Alas, there would be 39 laps – or 100 miles – to decide the event, and with a simulator engine inspired by isiMotor under the hood, you never quite know what the AI are going to do when the green flag drops.
This race was over before it began, and that’s not fun when you’re signed up for an hour’s worth of driving; picking off my first backmarker not even a quarter of the way into the race. From the moment the lights went out on the starting grid, I was basically driving in my own zip code, completely destroying the AI cars to the point where had it been an online event, I would have most certainly been accused of cheating by virtually everyone sporting a PlayStation 4 headset. I understand that some vehicles in Project CARS are optimized for the AI better than others, but the Trans-Am event just flat out wasn’t fun. You race for the thrill of racing, and this wasn’t much of a race – it was a parade of Mustangs around Sears Point, and I significantly dialed back my pace to ensure I wouldn’t choke the inevitable victory away.
As I made my way through the field and became involved in sporadic packs of opponents, it was hard not to question what was occurring out my front windshield. The AI cars had great difficulty maintaining their composure throughout all corners of the race track, occasionally wandering from the racing surface entirely, or deviating from the preferred racing line for no rhyme or reason. Unlike traditional isiMotor products, where the bots stick to the coded driving line as if their life depended on it, the AI drivers in Project CARS took a very creative approach to piloting a virtual race car, liberally interpreting what many of us would consider to be the proper line around Sonoma, maintaining a moderately quick pace only due to artificial velocity or grip boosts that the player is not privileged enough to receive. If I drove next to an AI car – not hitting them, but merely indicating I was in the process of putting them a lap down – they would throw their car into the sand as if they had suddenly gained sentience and given up in frustration. Provided they maintained their composure and actually put up some kind of a fight, their driving lines would be absolutely nonsensical, swerving across the track with the grace of a soccer mom late for her daughter’s dance recital, and on many occasions I was forced to take evasive action when it shouldn’t have been required in the first place.
None of this was enjoyable as a player.
Adding to my growing list of frustrations was the inherent lack of pit strategy displayed by the thirty bots I had been sharing the track with. As I said earlier, SCCA Trans-Am races don’t require a pit stop; it’s one tank of fuel and one set of tires until the very end – a true sprint series. Regardless, the entire field could be seen in the pits at one point or another, handing the victory to me on a silver platter and producing a gap of almost two minutes to the driver in second place. As if this wasn’t enough to put the nail in the coffin for this train wreck of a race, a majority of the field remained in the pits for good until the conclusion of the event – and I’ve documented this in the screenshot above. My race engineer did not make me aware of any incidents that had occurred on another part of the circuit, meaning many AI cars simply drove into the pits for fuel and never left.
By the time I’d completed my thirty ninth lap and took the checkered flag, I managed to nearly lap the entire grid twice, and was very close to two whole minutes ahead of second place on the game’s highest difficulty. I wasn’t even running a custom setup, and the only practice laps I had were the five or so laps registered in qualifying – two of which were spent fine-tuning the steering box.
Race #2: 2016 NASCAR Sprint Cup @ Watkins Glen (90 Laps)
America’s most prestigious auto racing series invades upstate New York every August for one of two road course stops on the Sprint Cup Series tour, visiting a neutered layout of Watkins Glen International – a circuit which for several years hosted a late-season round on the Formula One World Championship schedule. Watkins Glen has never been a particularly difficult race track, boasting massive sweeping corners that serve as the perfect tutorial circuit for stock car-oriented sim racers wanting to understand the concept of turning right without running into too much trouble. It’s honestly hard to fuck up a lap here, the track is just so wide, the turns so simple, and the layout incredibly easy to memorize, you usually don’t run into problems unless it’s literally your first time away from the confines of Daytona or Talladega.
Arriving as post-release downloadable content for Project CARS, inside the American-themed car pack, was the 2016 Ford Fusion stock car. Based on 1960’s technology, albeit finally dropping the carburetor in favor of fuel injection, modern NASCAR stock cars are sort of insane – rumors circulated only a few short years ago of top teams crafting engines which would produce nearly nine hundred horsepower – a combination that’s honestly a bit dangerous in a 3,300 pound brick on wheels that can’t turn and has a hard time slowing down, let alone completely stopping. Unlike the 1996 Trans-Am Mustang mentioned above, however, the virtual counterpart of the 2016 Ford Fusion stock car as seen in Project CARS is an abomination of the highest order. It’s certainly got the engine and tire specifications correct, but the vehicle’s weight feels as if someone cut the value in half, boasting the maneuverability and overall feel of a gymkhana car in DiRT 3.
It’s bad. Really bad.
Qualifying showed signs of potential. Rather than dominating the field for an easy pole award, a mere four tenths of a second separated myself from second place, and I didn’t feel as if I ran a bad lap. And when the race went green for what would eventually be ninety minutes of racing, the AI cars surprisingly kept up to my bumper, though none of them ever got a proper run and attempted an overtaking maneuver. Once again, I’d occasionally catch the bots deviating from the driving line for the sake of deviation rather than strategy, artificially assisted by an invisible master hand ensuring their ticks and twitches did not cost them speed, but this was not enough to prevent me from checking out.
Fuel strategy woes also plagued the AI cars, however these problems crept up from the very start rather than waiting until late in the race like the Trans-Am event. Long before I’d even began calculating how much fuel I was burning per lap – so not even halfway through the first fuel run – I noticed many bots were already filing onto pit road, effectively allowing me to lap nearly half of the field over an hour before track position would become an issue.
I honestly shit the bed during this event as a driver. My first pit stop, I had no idea how to actually execute a pit stop in Project CARS, meaning I left pit lane with only ten extra liters of fuel, and the same set of worn tires still on the car from when the race had started, resulting in a situation where I was forced to pit again and lost almost 45 seconds of track position due to my own incompetence. And around the midway point of the event, I cooked the left rear tire a bit too much and basically destroyed my vehicle only a few feet past pit entry while exiting the final corner, causing me to limp around the entire race track – in front of a live Twitch audience – back to the pits for an unscheduled stop. At this point, my estimation was that I’d lost nearly three minutes of track position in total, which should have taken me out of contention for the win right then and there.
It didn’t.
I drove like a madman through the field, and the hole I spent the back half of the race climbing out of really magnified the shortcomings of the artificial intelligence in Project CARS. These bots simply cannot drive, and while they don’t intentionally set out to destroy you or have any sort of designated gameplan behind their sporadic maneuvers, I’d feel a bit better if the realistic liveries were replaced by the paint jobs from the old Destruction Derby series.
Taking evasive action from one of many mammoth AI clusterfucks was something that happened once per lap, and their line through the bus stop frequently saw them cutting across the grass like a John Deere lawn tractor. If I happened to be in front of an AI car going into the backstretch chicane braking zone, I was shoved into the grass by their front bumper at mach one. It was a very sloppy, unpolished experience, and at the risk of ruining the otherwise impartial tone of the article, the game may as well have been raining rogue Austin Mini’s from the sky, because after a while I was expecting practically anything to happen if an AI car was in my general vicinity. With how frequent their meltdowns would occur, I’m surprised the official trailers for this game have managed to completely omit these instances from featured gameplay footage.
Like the Trans Am event I participated in a day earlier, the entire field of thirty nine AI cars threw the race away with their nonsensical pit strategy, allowing me to capture the victory by twenty seconds despite losing almost three minutes of on-track real estate thanks in no small part to my own lapses in concentration and unfamiliarity with operating the pit strategy menu. Had I ran a race where I didn’t perform an extra pitstop or smash the car into the front stretch wall, I would have lapped the entire field at least three times on the game’s highest difficulty. It’s not fun or even the least bit challenging to drive for almost two hours on what the game says is the ultimate challenge, ruin your race on multiple occasions, and still take home the win uncontested.
Race #3: 2012 IndyCar @ Road America (50 Laps)
Virtually unchanged from the layout which once hosted Can-Am Series races in the late 1960’s featuring the likes of Bruce McLaren, Dan Gurney, and Mark Donohue, the four mile permanent road course in Elkhart Lake, Wisconsin has established itself as one of America’s premiere purpose built auto racing facilities. Unlike the highly technical rhythm sections of the Austin Grand Prix circuit, or the claustrophobic concrete jungle of Long Beach California, Road America is a throwback to a time when big dreamers built insanely fast and incredibly dangerous tarmac paths through otherwise dense forests, and while there’s obviously been an effort to modernize the facility, the remnants of Road America’s roots lie just on the other side of the catch fence. After several years away from the property, the Verizon IndyCar Series returned to Road America for a pair of two hundred mile races in 2016 – one on Saturday, and another a mere twenty four hours later on Sunday.
Another post-release addition bundled with the American-themed car pack, the 2012 Dallara DW12 is the foundation of what was at the time known as the IZOD Indy Car Series. With the help of the late Dan Wheldon, Dallara constructed an open wheel race car that combined the turbocharged insanity of late 1980’s Formula One rides with modern advancements in downforce and mechanical grip, effectively chopping the balls off of a Lotus 98T so the children of past CART stars could have something to play around in. Dallara isn’t known for fantastic professional level race cars, and it certainly showed with the DW12; the car generally lacked torque exiting low speed corners, but would take off like a rocket once the driver entered the high RPM range and the boost accumulated, creating a car that looks cool but is kind of shitty to drive for an extended period of time. For those who own and still play rFactor 2 on a consistent basis, Image Space Incorporated have absolutely nailed the behavior of this car.
I believed the Dallara DW12 was yet another car the boys at Slightly Mad Studios didn’t get quite right, though some extensive lurking allowed me to stumble upon a setup for Road America on some Project CARS setup database that drastically changed how the car performed. So as a special shout-out to hkraft300, you’re the real MVP of this race. Your setup is awesome. The car isn’t perfect to drive, but it went from being fairly dog shit with the default setup, to significantly better, though the awful Project CARS garage layout made putting this into the car a real pain in the ass.
I snatched the top spot on the grid by only three hundredths of a second, and two AI cars flat-out walked away from me when we finally got down to business, which at this point was a really welcome change of pace compared to how the other two events began. Slowly but surely, I reeled in the leaders, yet my initial pleasure of seeing a pair of competent bots out in front of me quickly faded as I realized why I was able to eat away at the gap. Cones that were once present on the inside of several corners at the start of the race mysteriously vanished, and once the leaders were within a few car lengths, I was able to clearly see their vehicles monster trucking over some of the more treacherous rumble strips, losing precious time in the process that the artificial grip and speed hacks AI bots receive in Project CARS couldn’t completely overcome.
Like clockwork, I inherited the lead long before the pit window had opened, the AI cars opting to pull in at a time no human driver would consider to be reasonable. On the plus side, I was enjoying how the car felt with worn tires; one thing I’ve found odd about Project CARS, is that the overall driving experience really comes into its own about two thirds of the way through a fuel run. I’m still not sold that there’s this super advanced tire model operating behind the scenes as the marketing stuff has claimed, but there does come a time where you get into a rhythm and it doesn’t feel too bad.
Not completely recovered from my marathon at Watkins Glen a day earlier, another lapse in concentration saw me mow down a line of traffic cones in sector three as I approached the midway point of the race. This royally fucked up both the front wing as well as portions of the suspension, and for the second event in a row, I was forced to make an unscheduled pit stop. Unlike the NASCAR wreck at Watkins Glen, this wasn’t the result of pushing the car too hard and cooking the left rear; while the setup I’d found online was quite good and I felt I could push the car hard in many of the treacherous corners at Road America, the Dallara DW12 as seen in Project CARS is susceptible to random surges in lateral grip in high speed corners, akin to the same effect that creeps up on people in iRacing when they least expect it, and I couldn’t settle the car down quickly enough to avoid an off-track excursion.
On a positive note, I was impressed that smacking into a row of plastic traffic cones and bottoming out the vehicle on the grass had a noticeable effect on the car’s performance. Some other simulators don’t model the dangers of traffic cones.
Repaired, refueled, and rejuvenated, I hit the track on the same set of Firestone Reds – soft compound tires – that had been equipped during my first pit stop on lap 17 of 50, in an effort to prevent myself from bleeding real estate on track by sitting in the pits for too long. With the pack spread out completely around the four mile Road America circuit, I was now forced to deal with AI bots on a regular basis, and it turned out the increased downforce and improved handling characteristics of a professional level open wheel race car did little to prevent the AI from turning into kamikaze land missiles that made the Watkins Glen NASCAR event a frustrating endeavor.
By chance, I managed to record a clip of their antics with the PS4 share button, documenting an AI car shooting across my screen after entering turn three with an absurd driving line and basically failing to hit the brakes, and I can assure you this was not the only time this situation happened. Things got to the point where if I saw an AI car in my mirror, regardless of his standing in the race, I conceded the position immediately – it was not worth even bothering to try and race these guys properly. This is extremely disappointing to have a racing game where you are essentially forced to omit the whole racing aspect of the experience, waving by AI cars until they inevitably become the victims of their own ineptitude.
The clutch failed on the next pit stop, though I once again neglected to take new tires to save time in the pits and remain in contention with an AI car that had established itself as the lone bot to worry about when it came down to the overall race order. Unlike the Trans-Am and NASCAR events, the AI machine piloted by Izkander Liedia was fast enough to win the race if I didn’t put a bit of effort in during the closing laps. And it legitimately looked for a fifteen minute segment that I would have to settle for second place. Izkander was in the lead by a ridiculous margin, and I hadn’t physically caught him on track; only awarded the lead when he eventually dove into the pits. We were stomping the rest of the field as a unit, but Izkander was hauling ass and clicking off lap times that were basically inconceivable – a massive complement to a bot living in the world of Project CARS.
Izkander pulled into the pits with six laps remaining. This dude had the race won. I hadn’t even seen him on track in the previous eighty minutes of driving, that’s how quick this bot was running. Yet on a set of soft compound tires that held up for the final 133 miles of the event and had just a sliver of life remaining on the left front, I netted my third marathon victory in a row because the race leader pit for absolutely no reason.
100% difficulty, ladies and gentlemen.
Conclusion
A pretty vocal portion of the console sim racing community have claimed to invest a fair bit of time into Project CARS for either offline racing, online leagues, or both. If you travel around to various sim racing message boards where talk of Project CARS is not met with immediate hostility, there are stories of people completing half-length or full-length events with relative ease, treating the game as their own personal auto racing career thanks to the diverse array of both cars and tracks featured on the game disc.
After 520 miles driven across three full-length offline events, I’m honestly not sure how they’re managing to accomplish this in any fashion. Either these people are so incredibly bad at the game that the AI bots actually have a chance against them, or the WMD members did an exceptionally good job at fabricating stories about the game for their own specific communities that just aren’t true. The artificial intelligence in Project CARS just doesn’t function properly enough at any given point to provide a compelling offline racing experience. They can barely stick to the ideal driving line without unnecessarily deviating when it is not appropriate to do so, they constantly pit when it is simply not necessary, and blow corners at such high frequencies that the game’s cover athlete may as well be Vadim Kogay. Provided you’re not a complete beginner to the genre, it’s possible to run a race at 75% attack and finish multiple laps ahead of the competition on the highest difficulty.
It was indeed cool that I could bounce around to all these different series within the confines of a single game disc, but I can’t deny what I saw out my virtual windshield. The nuances you have every right to expect as a customer from a product that labels itself as a hardcore racing simulator basically don’t exist. The bots merely circle the track… poorly… and provided you can steer in the direction you’d like to proceed in, you’re declared the winner after completing the set race distance. It’s just not a rewarding or exhilarating experience, which is why people play racing games in the first place. Racing is intense and demanding; a chess match at 160 mph. You don’t play a racing game to progress through the story or level up your character, you race for the sheer thrill that racing provides. That thrill is non-existent in Project CARS. You pull away from the bots, and when you cicle the tail end of the field and proceed to work your way through the lapped cars, you get to watch them smash into walls, drive into the grass, spin out, sit in the pits, and generally do things they’re not supposed to be doing.
It’s a bit like watching that old Super Mario 128 tech demo from all those years ago, where the characters just aimlessly wander around the playing field, only this time, they’re stuck inside a race car and given a general direction to progress towards.
With Project CARS 2 to be released at some point in the not-so-distant future, it’s absolutely integral for Slightly Mad Studios to completely eradicate the AI issues documented above. Otherwise, improvements made in other areas of the game will simply not matter in the slightest.