Valve favors updating Source to making new engine
Valve's Source engine is getting older, but the company is content to continue iterating on it rather than introducing a new "Source 2" engine.
Valve's Source engine is getting up in years, having debuted in 2004. Still, the company isn't keen on dropping the engine, and claims that updating and iterating on it makes for an easier workload than creating a "Source 2" engine.
"That's just the way we work with that engine, [we] just update it not just replace it," said Valve's Chet Faliszek.
While Faliszek told GamingLives (via PC Gamer) they may create a Source 2 "at some point," the current model is easier for the time being. "When you replace an engine you're replacing the tools and the way that people work, there's an expense in man hours and people learning and people getting up on it right."
One part of the iterative process was adding Mac support last year. According to Faliszek, testing on Mac hasn't added too much time to adding new features to the engine itself. "There's always some weirdness when we add new features to the engine, and we need to make sure when we do it still works with the Mac, but it's always part of the process."
The Source engine will be powering Valve's next releases, Dota 2 and Counter Strike: GO.
-
Steve Watts posted a new article, Valve favors updating Source to making new engine.
Valve's Source engine is getting older, but the company is content to continue iterating on it rather than introducing a new "Source 2" engine.-
-
-
-
-
Doing some but not all parts at the same time doesn't mean you can't make meaningful changes and have something after several changes that is significantly different from the past.
I'm not sure if you are aware but John Carmack has mentioned numerous times how he has changed the rendered in his various engines for his research while not replacing the entire engine. He has even had a voxel renderer working for Quake 2 or 3, I forget which.
In the same way, Source was designed from the ground up with modularity so that major changes could be made to sections without necessarily having to change all the others.
We know the pain of being limited in CPU upgrade if we don't also upgrade a motherboard, but there are also times when you can upgrade individual parts to the latest and greatest without having to change any other parts.
-
-
-
-
It makes sense though. There's no point in develop a new engine, that's based on DirectX, until DirectX 10 or 11 adoption crosses into the majority. Read an article yesterday saying that Win 7 has only just not surpassed XP in total installs. Steam has the advantage of being able to look at their hardware survey data, so they probably have more accurate numbers and I haven't look at them lately. But, it just makes sense that it's pointless to spend money on R&D like that until DX10/11 is closer to being mainstream. This is why Carmack went with OpenGL (and other issues).
-
-
DX11 still isn't industry standards. Blizzard hasn't adopted it, EA just did, Activision has been toying with it. Blizzard just added 3 things to the DX11 option in World of Warcraft. My friend turned on DX11 and we looked at each other's screens and couldn't see a difference between D3D9 and D3D11. StarCraft II released with a heavily modified DirectX9 engine where it looked like 10. VALVe does it the right way. Release updates so as to not force their large community to relearn the SDK.
-
-
-
-
id used to do that, but that was back when the technology was so vastly different from game to game. Wolfenstein used real mode, DOOM used protected mode (I may have those backwards), Quake 1 used polygon enemy models for the first time, Quake 2 supported 3D accelerators out of the box, Quake 3 required a 3D accelerator, etc.
And if I remember right, id made it a big deal that they tossed out and rewrote everything. This, I think, has cemented everyone's notion that you must occasionally sit down and rewrite the engine from scratch or else you're lame and will fall behind.
Today though, I think game engines are like web browsers - there's no point in writing a new one from scratch. There's nothing wrong with Source that can't be fixed. At some point people will stop writing engines entirely - you'll just use the latest build of the Unreal engine and call it a day (this is already what a number of devs are doing). Spielberg doesn't need to invent a new camera every time he does a movie, game developers don't need to reinvent the wheel every single time.
-
Valve LOADING... favors updating LOADING... Source to making LOADING... new engine.
(I have no problem with Valve updating Source, and it's a good basis for an engine. But don't cheap out and skip important functionality that everyone else has these days. I really like Source, but I don't think its evolution has been going fast enough.)
-
-
-
You know Bungie only had you load maybe once or twice in Halo and Halo 2. Then when Halo 2 PC and Halo 3 launched, we had a load screen every mission/map load. Do you know why it's a standard now? Because it lessens the load on your computer, giving the game that much more performance to the game rather than giving half to performance and half to loading.
-
-
-
-
There is nothing wrong with this approach, as long as its managed well with. Their toolset example (there are others) is the BEST way to describe the issues involved with gearing up with an all new engine, etc.
Carmack made a point about it during his Quake/Doom engines upgrades. His core of netcode/input/audio/graphics connections worked out really well, so he would keep a majority of it during the "next" engine development. The largest items he kept removing and updating was the graphics engine and I'm guessing they have added on the "texture" and potentially "physics" engine to the mix as well.
I'm really interested in the work Valve did with their particle benchmark and how they intend to use that knowledge with Source.-
-
This article from last week actually talks about some of the original Quake code being in those engines: http://www.1up.com/features/why-quake-changed-games-forever
-
-
-
-
-
-
It seems one of Valve strategies is to make sure that money pays and retains employees and expertise. Which is fine by me. You never read about major lay offs at Valve each time they release a new game like almost every other developer out there. They keep the same pool of talent and take on new projects and still remain profitable.
-
-
Source is a bit long in the tooth. I'm surprised you guys are giving them a free pass. They are considered the bastion of PC gaming yet have one of the most dated engines out there. With all the constant clamoring for games that push the hardware envelope, Source runs just fine on my 6 year old laptop.
-
-
My only technical beef with source is the lack of streaming and segmented map loads. It seems to be getting worse as time goes on if Portal 2 is any indication. It's quite bothersome in the HL2 and Portal games because they are set up to be continuous experiences, while many other games with separate level loads and breaks aren't.
Aside from that, the engine is fine. Runs well, feels solid, still looks really good thanks to artists and animators.
-
-
Why is this news? Right from the very beginning they explained how Source engine was designed in a modular way from the ground up so that they could update parts and keep it current for a very, very long time. They don't need to create Source 2 unless there is a major paradigm shift required (though arguably a MegaTexture style tech may require such a paradigm shift to allow for complete custom texturing of the world).
Their stated goal was to avoid having to develop a new engine every few years. And that is exactly what they achieved. The first notable example was probably when they released the HL2 Lost Coast level where they updated the renderer to add HDR lighting tech, and also added the commentary system. Since then there have been many changes and Source engine games of 2011 like Portal 2, look a lot different to HL2 in 2004.
-