Testing a game like Project Cars 2 is an unusual process compared to most. Of course, there are some things that will always be familiar in games testing – bugs, glitches, graphical issues, memory leaks, etc. But with a simulation you aren’t just testing to make sure it works, you’re also testing to make sure it’s as realistic as possible.
In fact, it's fair to say that QA and testing are inherent in building a racing simulation. “There’s so much of it right from the get-go,” says Stephen Viljoen, game director for Project Cars 2 developer, Slightly Mad Studios. “Instead of ‘QA’ing’ a gameplay experience you have to QA the technology behind it.”
Does it sound like the real car? Did we capture the car or is it wrong? Does it handle like the real thing? Have we captured the essence of the car?
Rod Chong, Slightly Mad Studios
Project Cars 2 is built using an in-house engine and testing isn’t just reserved for the cars or the track. A lot of that detail has to be known by the testers before they can make sure that the quality of the game is accurate. “The technology has to be tested independently of the specific environment (like a track) and it has to be tested per environment as well. There is so much detail that falls out of the level of detail assimilation.
“The simple example is our spectators are dressed based on the weather, based on the seasons and then based on the era, and QA has got to pay attention to that. From our perspective, with my team working on the game design and specifying all that detail, we’ve got to make sure that the documentation that we handed over to QA is thorough enough that they know what to check for. Then the QA team has to pay attention to all of those little details. It’s not something that you typically find in many games or simulations.”
(L-R) Rod Chong and Stephen Viljoen
Another example of this is the pit crew in the game. With Project Cars 2 being a sequel, there are already some things in place that will help testing but the others, like pit crews and pit stops, have not only been entirely redesigned and animated, but they have to be accurate to the rules of the racing formula they are in, but also the era that they come from.
“If you go and do a pit stop while in the 1970s cars, the pit crew will be dressed like they would have been in that era. We can’t assume QA would know that, right? We can’t expect them to know that. It’s just a level of detail that a general QA team simply wouldn’t know. But that is one of many very important details that recreates the real world as close as possible.”
You could argue that because testing and quality is so inherent to the makeup of Project Cars 2 that QA probably isn’t the right phrase. “I don’t know if we would call it QA,” says chief commercial officer Rod Chong. “But there’s certainly a feedback process that we go through with the drivers or with people that are focused on authenticity. We have to look at each car and ask ourselves the question: ‘Does it sound like the real car? Did we capture the car or is it wrong? Does it handle like the real thing? Have we captured the essence of the car?’
“Like a Porsche 911 for example, with a rear engine, has some very particular handling characteristics and ways that you drive it which are quite unusual. We have to always be going through these feedback processes to try and strive for realism and authenticity and that’s a continuous process.”
One of the ways that Project Cars 2 has achieved the realism it has in the handling of the cars is thanks to the input of real drivers. Slightly Mad Studios currently has seven full-time drivers working on the game. Names such as stunt driver and former Top Gear ‘Stig’, Ben Collins, and former touring car driver Nicholas Hamilton have contributed heavily.
“We did two physics sessions with Stefan Johansson who was a Formula 1 driver in the 1980s, but he’s raced a wide variety of cars from the 1970s to the present day. So we spoke to him about his racing career and understood what cars he raced in what eras. Then we did tests where he drove about seven or eight different cars, all stuff that he’s driven and he gave us a lot of feedback.
“There were a couple instances where he drove a car and he said, ‘you’ve got this wrong.’ Like the Porsche 962C. He drove it and he said, ‘no this doesn’t feel right. You’ve got caster wrong.’ He said, ‘we ran this amount of degrees caster angle on it. Go back. Fix it.’ So we changed it.”
It is incredible that regardless of time, a driver can recall exactly how a car feels from memory. It’s even more incredible that a video game can capture that and present it to the driver authentically. It’s a testament, not only to physics or the improvement in game controllers like steering wheels but also to the dedication to testing and quality from Slightly Mad Studios. “The level of simulation now is so sophisticated that you can actually feel everything that’s happening,” says Rod Chong.
“You can feel the tyres, you can feel the suspension geometry and you can really sense what’s happening with the car now. For people that are highly technical, like these racing drivers, they can drive the car in the simulation and then know what has to change and give a list to the programmers. Then they can continue to evolve the physics and the handling characteristics of the cars.”
Racing games in this era of game development get more from car manufacturers than ever before, but historic racing is a much more difficult prospect. “When you have the modern cars, most of them have driven on test tracks or on race tracks,” says Chong. “They could be the same tracks and so we had laser scanned millimetre perfect versions of those, so we can get a data trace of the car going around a circuit.
“But when it comes to things like the older cars, some of which may not have been driven 20 years... What you don’t want to do is have a game designer, sitting at their desk thinking ‘Oh okay, Ferrari. How should that handle?’ And they’ve got some sliders to add oversteer, top speed, grip, downforce, etc. For us, this is unacceptable. We have to simulate whole aspects of the car, and that’s what we did. That’s the starting point.”
“We have this vast resource of data available,” concludes Viljoen. “But how you get that to the point where it’s accurately representing in the simulation, that’s the secret sauce.”