Whilst IMSA content and reworked physics headlined changes included in the latest version 1.6 update to Automobilista 2, fans of offline racing also had a lot to celebrate. Throughout its development, Reiza Studios promised a whole heap of AI improvements from race ability to speed and even strategy logic, in theory placing the Brazilian title’s opponents among the best in the space.
Well, the benchmark for AI offline racing in 2024 for sportscar and endurance racing has surely been a toss-up between the incredibly solid iRacing and the highly immersive Le Mans Ultimate. So, let’s jump into each title to work out just how effective the AI is in various sportscar racing scenarios, so you know which to prioritise when recreating real-world races.
Comparing Sportscar AI in iRacing, LMU and AMS2
To compare the artificial intelligence in this trio of simulators, we will focus mainly on their shared core focus: sportscar racing. Unlike simple sprint races, this style of competition includes several facets unique to the genre, presenting numerous challenges for the AI.
With overall immersion for single-player running forming a critical focus of our test, we will also be diving into the AI’s ability to race convincingly against in-class opponents as well as their competence at passing through traffic and plan out strategies for those longer, more complex races.
Recreating real-world grids
But first, any offline racer will tell you that their race focuses on immersing themselves in a real-world series. We do not want to race John Smith, Mario Rossi, or any other placeholder names you can think of. Instead, we want accurate liveries with their real-world drivers. And depending on the title, putting together accurate grids is more complex than others.
Straight off the bat, the Downloads Section here at OverTake is full to the brim with accurate livery and AI packs for Automobilista 2. These focus on the Formula One scene, but several packs are also available for other series. These come with custom AI performance and character attributes and make the most of the different models available to drive in AMS2.
Moving over to iRacing, which is a similar story of total customisation, check out our guide here on how to put together your very own AI car set in the online racing service, which is fit for offline races. Creating one’s own livery pack takes a lot of time; however, it seems that searching for pre-made packs will require a lot of Google searches and scrolling.
Offline Sim Racing for Real Life Grids. Image credit: Porsche Press Site
Finally, there are two ways of looking at Le Mans Ultimate‘s real-world livery situation. As the official game of the FIA World Endurance Championship, it gets all the cars, liveries, and tracks from the 2023 season, with the 2024 collection also set for completion soon. But choose to race anything other than the WEC, and you will be forced to look elsewhere. In summary, LMU is either absolutely perfect for your needs or totally useless.
Setting up Races
Before we get into the best battles you can have against the AI in sim racing, we must first set each game up with the right session settings. Each title does this slightly differently, with iRacing, in particular, proving to be quite the problem child.
Setting the AI in iRacing
Yes, having a vast array of options available to you in the typically online racing game is nice. However, it can be tricky to know where you stand when the AI difficulty level slides between 0 per cent and well over 100. In general, this slider can be humbling to those used to other racing games as the majority of us that sit in the ‘normal’ range of racers – in the 2,000 iRating region – will be perfectly competitive with AI opponents set between 70 and 80 per cent.
iRacing’s AI Difficulty range can be tricky to get to grips with. Image credit: iRacing.com
The rest of your sliders are complicated and should realistically be set when putting together your AI grid set. Each racer’s personality can be changed with a quintet of bars that needn’t be touched. On the other hand, options such as Time Acceleration and start times are far less free to your choice, so perfectly simulating a time-accelerated endurance race isn’t as easy as you might think. Moreover, some elements like safety cars and the ensuing pass-around options are not even relevant to our case; they are only functional on ovals.
Setting up AI Races in Automobilista 2
Another title with a plethora of AI difficulty and setting options is Automobilista 2, with not only overall difficulty available for tweak-age but also their relative strength in the wet. Following the 1.6 update to the game last week, most car and track combinations should all sit at a similar AI difficulty for you, especially the IMSA content that has clearly been fine-tuned to avoid players spending hours testing for the right level.
AI Settings in Automobilista 2. Image credit: Reiza Studios
Below that is the wet weather slider, which you may want to test. Again, it should be at a similar level across all the content in the game, apart from if you have a particular affinity or hatred of a certain track. As for the aggression option, we recommend leaving this at Medium, allowing the game to choose aggression based on your AI livery pack.
Akin to iRacing’s session setup options, the time progression in Automobilista 2 is restricted to specific values. Equally, race durations always hit a five-minute mark, meaning you cannot make full use of our time progression guide.
Le Mans Ultimate AI Race Setup
Unlike its two rivals, Le Mans Ultimate allows great freedom for both race duration and time progression. As such, you can pretty much match any race duration to a real-world event length. However, the maximum length is that of the real-world event at each track. Therefore, Le Mans is the only track at which you can run a full-time 24-hour event.
LMU’s AI settings are in two different places. Image credit: Studio-397
Much like Automobilista 2, the AI difficulty strength across the board appears to be very stable, with a single setting working well for all circuits. One confusing slider in the game is the AI Limiter setting, which slows the overall leader down, ensuring the pack stays tight at the front. Unless you want a ten-way battle for the lead, we recommend dropping this to 0.
Best AI Racecraft in Sim Racing
Now, we get down to the nitty-gritty portion of the subject. Once your session is set up, it is time to go racing, and when it comes to battling opponents of a similar speed as your own, there are a few other games that can match the likes of iRacing, Automobilista 2 and Le Mans Ultimate.
Indeed, all three of these games feature AI that will think of switching up their lines when close to another car, be it in defence or whilst attacking. More complex moves like going for an undercut line and focusing on corner exit grip may be a bit too advanced today, but there is still a lot of fun to be had going wheel to wheel in each of the three games. That being said, there are notes to point out from each.
LMU: Clearly Scripted Defence
Moves scripted by AI lines are certainly to be expected from any racing game in the modern era, given when you look deep enough, each and every one of them is built upon a chain of 1s and 0s. But in Le Mans Ultimate, certain defensive moves are so clearly scripted that they become easy to predict and work around.
Le Mans Ultimate’s AI runs a set of clear lines. Image credit: Studio-397
Down many straights, namely the run into Turn One at Sebring, the AI will aggressively hold the inside line if a car behind gets close enough to activate the behaviour. This leads to somewhat unrealistic movements that no real driver would do in a real fight. In fact, the bots are so keen to hug the inside line that they stick too far to the left, making returning to the regular racing line impossible before the turn.
After a while, you will automatically look to go around the outside and not even think about the inside dive as you soon learn it to be a blocked space. Sure, aggressive defending also exists in IMSA racing, but lap after lap for up to 10 hours is a bit of a push.
AMS2 Grip Hacks
Overall, we suggest that the most fun to be had in pure racecraft situations offline is in Automobilista 2. The AI is very good at defending when it needs to and sticking to the faster line when it can get away with it. Moreover, the slipstream effect is evidently very strong in the game, with the AI actively looking for the toe down the straights, meaning they will make moves even on player cars without hesitation.
Multiclass racing in AMS2. Image credit: Reiza Studios
However, once you get to the corner after a straight, it appears a little dodgy in the Brazilian game. Once alongside you, it feels as though the AI can do things you simply cannot. Worse still, these moves also feel different to what the bots can do when running solo. Go to the inside, and suddenly, your opponent will appear to gain extra grip, making a round-the-outside defence an easy task, where the player would get caught out by dirt off-line, for example. This means that any move you make has to be clear-cut.
Whilst we cannot entirely prove that the AI gains grip when fighting each other, it is a sensation we get when in an intense battle.
By extension, the combination of impressive grip through the turns regardless of a car’s positioning, plus a high drafting effect, means cars follow each other incredibly closely. A normality in aero-less models like Formula Ford or Ginetta Juniors, nose-to-tail running should not be visible in GTP or Formula One racing, and yet it is in AMS2, a sight that dramatically ruins any sense of immersion.
The AI does tend to squeeze you on the apex in AMS2. Image credit: Reiza Studios
Finally, rooted in the Project Cars engine, elements of wheel-to-wheel battles shine through from the previous racing series. In many cases, contact with the AI will lead to a dramatic loss of grip, making turning and braking a tricky task. This is exacerbated when, on the inside, the AI does have a tendency to squeeze the apex perhaps a bit too much, turning in on you. As soon as the cars make contact, you lose all front-end grip, and both cars go ploughing off-track.
Lack of iRacing Aggression
With smooth-moving opponents that deviate from their line in defence, react to a player’s positioning and space out nicely on track, iRacing may sound like the perfect game for AI racing. However, we get the feeling that the computer opponents lack enough punch to feel alive.
Where the other games feature mistake-committing AI leading to fun exchanges, iRacing‘s opponents appear to be on rails in a drab way until you get close. Once more, even if you are far slower ahead of a same-class car, your chasing rival will struggle to commit to a move, no matter how dry-cut it appears. Whether they get thrown off by the player’s braking points or suffer from running a similar physics model to the real one, we do not know. But we do get the sense that a Turn 1, Lap 1 send can set you up for a fantastic result, provided you keep it on the road.
Racing Through Traffic
Inherently, multiclass racing is not only about dicing with one another at a similar speed. Instead, top class cars must weave through traffic all whilst jostling for position. Meanwhile, those in slower classes are always on the lookout to ensure they lose as little time as possible when passed by top-class cars.
Multiclass racing in Automobilista 2 has worked well since Version 1.6. Image credit: Reiza Studios
This added dynamic complicates AI’s competence in sportscar races, meaning only a select few titles can portray the sport with any sense of realism. Then again, across our three semi-official sportscar simulators, we see an impressive level of competence from the AI in managing traffic, especially with Automobilista 2‘s most recent update.
In fact, it is only when truly pushing overtakes to near-unrealistic proportions that many will ever gain significant ground on AI opponents in the top class. Better yet, this is the case for all three, with neither sim falling behind.
For the most part, Le Mans Ultimate is the more aggressive of the three in traffic, with three wide moments seemingly far from putting the AI off. However, at times, sequences of corners with the racing line moving from one side of the track to the other do confuse faster classes.
At its best, Le Mans Ultimate is the top Multiclass sim. Image credit: Studio-397
On the other hand, iRacing‘s top classes never really get stuck and relatively smoothly make their way through the pack without much drama. This can lead to the field closing up more than you would expect in larger battle packs. But it soon works itself.
In general, it is worth remembering that offline multiclass races are all against the AI, so they will never be quite as reactive as real players. Therefore, take your time in traffic; there is no need to send it quite as hard as you do online. Make sure to follow this chilled-out guideline, and you are sure to have a fun race-long experience.
Strategic Masterclass or Ferrari-esque Blunder?
The final element almost unique to sportscar racing is the inclusion of long-distance strategy. In the real world, teams typically run full tanks for the first portion of the race, switching tyres when necessary depending on allocation and only ever thinking about fuel strategy within a stop or two of the end.
Stoic iRacing Bots
Much like its performance on-track, the iRacing AI is very competent in the pits for the majority of a race. Similarly to LMU‘s opponents, your rivals will utilise full tanks for successive stints until the end of a race, at which point they will calculate the fuel they need to get to the end.
However, with tyre allocation inexistent and tyre warm-up not as significant a factor as in the other games, changing tyres at each pit stop is the done thing versus AI. They do it, so you might as well do it, too.
AMS2: Strategic Genius
Last year, Reiza Studios introduced an update to its AI’s strategy logic. While mainly focused on open-wheeler racing, the effects also lead to a fun challenge in longer-format racing. That being said, splitting a race into equal-length stints from the start does not lead to a totally accurate experience.
Furthermore, whilst there is no limited tyre allocation as part of Automobilista 2‘s IMSA content, the lack of tyre warmers does make double stinting a set of tyres beneficial in the opening laps of a stint. The AI recognises that and makes sure to keep a set of tyres as long as it can. A race of less than two hours with regular fuel use and tyre wear should, therefore, not require any tyre swapping unless you are ready to make some moves.
In addition to AI that plans out its tyre and fuel strategy from the get-go, the added complexity of Safety Cars putting a spanner in the works does lead to fun situations in which quick decisions are crucial – not only to winning the race but also not losing a lap.
Safety Cars can flip strategy on its head, even in sim racing. Image credit: Reiza Studios
That being said, in a recent wish list for Automobilista 2‘s endurance racing update, we were hoping for more extended safety car periods. Indeed, the sometimes even single-lap at a slower pace doesn’t allow the pack to close up in the way we would like. So, hopefully, one day, a 15-minute IMSA-style FCY option becomes available. Joking, obviously.
Le Mans Ultimate: So Close Yet So Far
Now, straight off the bat, Le Mans Ultimate is the only racing game in this comparison that features a tyre allocation system. However, the AI typically pays little attention to this until the latter part of a race, at which point they will re-fit old tyres. This means that to keep things interesting, you must not think about tyre strategy and certainly not attempt a double stint on the same set.
One thing LMU does well, however, is ensure AI cars run full tanks and pit at the very last minute. At times, you will notice them lifting and coasting and even extending their stints by a lap over others. Though this strategy rarely sticks, eventually, they revert back to standard stint lengths.
Best Offline Sportscar Simulator
Overall, if you can get past the frantic opening stint and somewhat unrealistic nature of Automobilista 2’s AI paths, we suggest the Brazilian sim as your best option for replicating most endurance races offline. It gets the smartest AI from a strategic standpoint and features opponents that will make your life complicated on track.
Of course, if you are looking to recreate a World Endurance Championship race, Le Mans Ultimate is perhaps just as good as long as you overlook the lack of tyre strategy. The field spreads out in a very realistic way and provides heaps of immersion as you chase famous faces down in the doom and gloom of a dark Le Mans.
Whilst we do enjoy iRacing as an offline simulator thanks to its highly competent AI, there are far too many hoops one must jump through even before getting to the track. From tricky settings that require fine-tuning at each circuit to the elephant in the room that is the game’s subscription service, without forgetting the soulless vibe, one gets when fighting the title’s bots. There are many reasons to look elsewhere.
Regardless of what our favourite simulator is for recreating longer format events against AI, it is fantastic that offline opponents are finally getting some love from developers. We cannot wait to see how far AI tech comes over the next 12 months.
Which is your favourite game for offline AI racing in a sportscar field? Let us know in the comments.
Nota: El contenido ha sido traducido por Google Translate, por lo que algunos términos pueden ser imprecisos
Fuente: https://www.overtake.gg/news/best-sim-racing-ai-for-sportscar-racing-automobilista-2-vs-le-mans-ultimate-vs-iracing.2663/