Arma 3 interview - AI, map design, mods, and a response to "Will Arma 3 break my PC?"

In advance of the Arma 3 alpha dropping on Tuesday for Steam pre-orderers, I launched a salvo of questions at Project Lead Joris-Jan van 't Land and Co-Creative Director Jay Crowe.

dropping on Tuesday for Steam pre-orderers, I launched a salvo of questions at Project Lead Joris-Jan van 't Land and Co-Creative Director Jay Crowe. We'll have in-depth coverage of the alpha when the embargo lifts on Tuesday, March 5.

PCG: We were elated when Ivan and Martin finally returnedto Bohemia safely. Are they actively involved in development at this time? How are they doing?

Joris-Jan van 't Land, Project Lead: The four months they spent in Greece were very troubling. It isn't easy to continue development as if nothing has happened. Welcoming them back on the day they landed was a surreal experience. They are both taking some time to get back into the game, absorbing the changes that happened, and doing what they do best in their respective fields.

Jay Crowe, Co-Creative Director: : Yeah, I'd agree with that, but, generally, it's just awesome to have the lads back! There was really no stopping either of them. I think Ivan could probably still work with a newborn baby under each arm, a beer in his lifting-hand, and a keyboard in the other.

It's been interesting to watch your projects cross-pollinate. What's something Arma 3 has gained--technically or in terms of design--from DayZ, Carrier Command, and Take On Helicopters?

van 't Land: It can certainly help us as a relatively small development team to be able to assimilate tech from other projects on the RV engine. Never is it quite a copy-and-paste job, but we don't have to reinvent the wheel every time.

Crowe: Exactly. Our programmers would kill us if we tried to suggest sharing tech was that simple! That being said—from Take On—there's a bunch of stuff like render-to-texture and the environments tech but, for me, the best cross-pollination is found simply in the finalization and release of other games. Particularly in the context of longer or delayed developments, you've got to maintain a real focus on “product” completion and have an awareness of the trials, tribulations and rewards associated with getting your work outside of the development bubble. Take On's development wasn't perfect, but it's tight, focused structure helped highlight these aspects.

van 't Land: Learning from TKOH certainly helped us to evolve the Arma 3 team. We needed to really start looking towards the finish line and stop adding new functionality and content. Doing multiple public releases within the project (last year's expo versions started this) helps train the team to deliver a final product. Doing this for the Alpha now, let's us get things out of the way for the Q3 release and should help it be smoother. We get some great support from our studio in Prague when it comes to distribution and publishing!

Crowe: One of the best things about the A3 Alpha is that we've dealt with some of these risks up front, and we're in a stronger position because of that.

What new AI behaviors does Arma 3 implement?

Crowe: With Arma's AI it will always be a case of carefully evolving it. We've given ourselves some quite big challenges with some of the new headline features. Now, it's our job to make sure they work and integrate well alongside the existing behaviors.

van 't Land: New functionality means we need to train our AI to do more things. We try to refrain from trickery and simulate AI behavior in all aspects; letting them walk around, engage in infantry combat, fly gunships, drive boats and do many of the things players do. It's a lofty goal that sometimes will mean the AI is seen doing less than intelligent stuff. Now we are teaching them to engage in underwater combat (this adds a third dimension compared to walking on land), to detect and avoid minefields and flames, to drive using the PhysX vehicles, and to use Under-Barrel Grenades well.

"New functionality means we need to train our AI to do more things."

Crowe: Little fixes like the use of UGLs makes a surprising difference--even if they do sometimes cause friendly-fire problems! We've also tried to take care of certain things that impact upon AI. We've improved some of our tools so the terrain developers can see how their compositions are traversed by the AI. I mention it because this kind of work improves AI behavior through appropriate design. It's something our programmers often shout at us about. I like to think we've started to listen.

van 't Land: A lot of work goes into the actual configuration of all game content. Only when that is carefully done, can we see which things still need tweaking in the AI technologies. Internally we've set up something we refer to as “Task Force Balance,” headed by our Lead Sandbox Designer, Lukáš Haládik. They are trying to find the best balance between authenticity, realism and fun to match our design vision. Their initial focus has been on the regular infantry squad and their weapons, and now they are moving on to specialized weapons and vehicles. We want to be able to proudly show off our advanced AI by making sure it's properly configured.

Crowe: Looking at it from the perspective of our Alpha, we can say there's still some way to go, but it's rewarding to see progress and the alpha helps us focus on our goals.

Will Arma 3 take advantage of the “ headless AI” that's recently emerged in the A2 co-op community?

van 't Land: We still harvest most of the changes and fixes done in the Operation Arrowhead beta program, and this could well be one of those. The success of DayZ has provided opportunities for our engineers to look at multiplayer performance, security, stability and functionality in more detail. This benefits Arma 3 as well.