Interesting posting about the IL-2 Series

Posted by Black Six on the 1C boards

And now, the State of the Project address from project’s lead Ilya “Luthier” Shevchenko

We could have probably avoided a lot of bad blood if I had explained some basics earlier. So let’s take a step back and discuss the specifics of video game development as it exists today.

2012 is not 1999 or 2004. Hardware capabilities and community expectations grow at an exponential rate. In the golden early days of the series – over 10 years ago – we could have written an entire cool new feature in a week, rewritten a complex module in two, or created a new plane in a month. In those days we could offer you something new and exciting to look at every Friday!

Things are not the same today. A single plane model can take a year of work or more! Complex tasks such as changes to AI or flight model take many months of coding and testing. The industry reached this strange stage where the end result does not seem to match the amount of time that went into it. If you compare a late-year 1946 plane with a CoD aircraft, it’s hard to believe that one took a month and the other a year to make. Does it really look 12 times better?

If you drew a diagram of time spent on task vs perceived quality, the curve would be very, very sharp.

It is what it is. It takes a huge amount of time to rise above the bar set by the original Il-2. The illusion of time spent vs quality delivered affects not only the layman but us professional developers as well. We’ve underestimated the amount of time it could take to “do better” quite a few times. We could do it exactly like in Il-2 in a month. But we don’t want to do it like in IL-2. We want to do better. And to reach the next qualitative level you need to spend significantly more time. In some cases, a year instead of a month, or even worse.

Secondly, and I hope this part is obvious to most everyone, games are developed by teams of people with very different skills. They rarely intersect. For example, there are team members who are great at making 3D models of tanks. They can’t just switch over and make a cockpit one day. They’d need many months of retraining. And even more than that, a tank modeler can’t just come in one day and fix a memory leak in the network code. All he can really do is make great tank models regardless of what a network programmer or a map maker are doing.

I’m really hoping this will cut down on cries of “why do we need x when we still have the glaring problem of y!” Each team member does the thing he can do at the speed it can be done. “Hire more programmers” is not a solution either. If your plane can’t fly very fast, putting more pilots in the cockpit won’t do much for its top speed. Especially when your plane is a unique one-off model with controls and other systems quite unlike anything any pilot might have ever handled before.

So we address each issue at its own speed. Some bugs can be fixed very quickly, and they are. We can update a cockpit lever animation or fix a type in a matter of minutes. On the other hand, locating a memory leak can take many long months of dedicated work by some extremely qualified programmers. All bugs are not created equal.

So our general status report is very simple. Everyone is working very hard doing the same thing they were doing a month or a year ago. Graphical programmers are working on graphics. Network programmers are combing through the network code. Plane modelers are building 3D models of aircraft. No one stopped. Nothing was abandoned. On the contrary, the team continues to grow.

I also have to say that under these conditions reporting on intermediate progress is very hard. Say, we’ve measured FPS in the new graphics engine. The engine still has some problems, solving which will have a definite impact on FPS – whether positive or negative is impossible to say. However if we were to announce that in a mission X we’ve increased FPS by Y% on hardware Z, then regardless of the size of the WIP disclaimer, we’d be eating our words for years to come if they turned out to be too optimistic.

So we continue to work, quietly and privately.

The progress is a constant. Graphics are virtually complete. Almost all of the newly introduced bugs are squashed. There are lots of other improvements. The new project as well is advancing at a good pace.

I can’t say anything more than that however, this week like many other weeks. It’ll be ready when it’s ready. And as we all know, nothing good ever comes out of talking about it before it is ready.

PS And I’d like to point this out one more time. There is no conflict between the old and the new. We have one team that works on a single overall task, that is, improving the Il-2 series. Whether it is a new sound engine or a new graphics engine, we don’t make them for CoD or for the sequel. We make them for IL-2 Sturmovik.


Comments are closed.