Still I expect Valve to go even further by taking out so much need for animator intervention in the first place. The idea is that it truly is context that drives expression, rather than an animator having to step in and say "I think the character would react this way," and have to then manipulate sliders and such to get the right effect.
Just as with original Source, I believe that we can expect Valve to come up with something completely new to fit their unique set of demands for player immersion.
Let there be light(ing)
Half Life 2 really saw the birth of many modern lighting paradigms. HDR lighting, per-pixel shading on a massive scale including diffuse and specular mapping, detailed reflections and refraction, many of these things which are now household names in gaming graphics were not seen generally in games prior to Half Life 2. It was part of a new generation of games which at the time included also Far Cry and Doom 3. After the release of these games a rapid influx of new graphics technology followed, culminating is such revolutionary engines as UE3, CryEngine 2, Frostbyte and others over the next couple of years. At this point, graphics technology advances at, at best, an iterative pace. When was the last real revolution? Frostbyte 2? Please. It's amazing, don't get me wrong, but it's not a huge departure from original Frostbyte and doesn't improve much on CryEngine 2 either, though the latter saw its first action with Crysis in 2007. UE4? Yeah, it's the nicest dev kit anyone's ever put together in all likelihood, but nothing I've seen from it really screams the same kind of revolution that occurred around 2004-2005.
So when will we see the next real breakthrough? And what will it look like?
Speculation only gets us so far, of course, but the current APIs in existence have been far from maxed out by modern graphic artists. Largely this is due to the low overhead, or utter lack of modern rendering technology that they have to work with when it comes to current hardware on the console side of things. But if we were to explore rendering without those limits, we could expect at least the following:
Global Illumination - The Holy Grail of lighting. The thing current lighting models can only really simulate. A true global illumination model, simply put, is one in which all light acts like it does in real life, which is to say, being reflected and refracted correctly by objects in the world and creating both direct and indirect lighting situations (direct lighting: the object which is first contacted by the light-source. Indirect lighting: any subsequent objects that are lit by reflection off of or refraction through the directly lit object). Currently, "Global Illumination" techniques are a hodgepodge of various techniques which together simulate the effects of actual global illumination. Combining various types of light sources, simulating radiosity and ambient occlusion, altogether with hybrid shader effects and emissive textures and other such gobbledygook, we can approach the effect of a truly global ilumination model. But true global illumination system would be one in which light simply acts naturally, flowing, reflecting and being refracted in a natural way without manipulation through simulated techniques. When achieved, this will result not only in better looking lighting effects, but also in less work for artists to get a scene lit correctly and looking the way it should.
More Geometry - While the faces in Half Life 2 were fantastic, character models were woefully short on geometry. This is one of the few places where Half Life 2 absolutely does not hold up well with modern games. However, in today's world of more and more memory in our computers and right on our graphics cards, and API support for tessellation, we should see a massive increase in the geometric complexity of characters and environments.
Bigger Textures - Once again, with more memory overhead, combined with better and better compression algorithms, extremely high fidelity textures are right around the corner. Expect Valve to be at the tip of the spear here. They set great store by the immersiveness of their game worlds.
Extreme Performance and Scalability - Half Life 2 was famous for a lot of things, but one of the things that most set it apart was the fact that it ran ridiculously efficiently on current-gen hardware of the time and could be scaled back to run on almost anything. Like GladOS running on a potato. Put it this way; there will be basically no one on the planet that will be excluded from playing Half Life 3 due to the constraints of their hardware.
Water water everywhere...
There is considerable reason to believe that water simulation tech will be huge push for Valve as they continue to craft Half Life 3, and any other games, going forward. Half Life 2 and the Source engine have received great laud for the water effects Valve was able to accomplish. The water was simply beautiful, and everything that interacted with the water behaved exactly how you would expect.
Portal 2 continued to show Valve's emphasis on liquid simulation and it was a major triumph in that regard.
As for Half Life 3, it can be expected that it will continue according to its pedigree. But more so even than that, water in particular will likely be the star of a lot of a show that will revolve largely around a mysterious, derelict, sea-faring vessel; namely the Borealis (even if it is just suspended in ice somewhere).
We have so much further to push in this regard. Physics of water; volume, density, flow, almost none of those things have even been touched, let alone in any significant way. And of course, Valve will do it the Valve way, which is to make their water simulation to be not just a spiffy visual effect, but to have actual gameplay implications. It's time we move past Nathan Drake's wet t-shirt being known as the coolest thing someone has done with liquid simulation in a game. I mean, seriously? That's what we've got so far?
What goes up must come down, etc. etc.
If there's one thing that impacted me more, from a visual sense, than the graphics in Half Life 2, it was clearly and indelibly the physics of it. Valve put physics front and center. For the first time, real world physics became a part of how you interacted with the game world, and for that reason the game world began to feel more like a real world. Not only was it totally immersive to know that things in the game world would react to your presence and your actions in an accurate way, but it made puzzles that for the first time felt like part of the game world and not just part of a game. The puzzles in Half Life 2 are logical, intuitive, and never ever take you out of the game since all you're doing to solve the puzzle is interacting in a natural way with the game world. That barrel is in the way, so move it. That thing needs to float in the water, so stuff some of these empty, sealed jugs underneath it. It was blissful and immersive.
But the physics in original Source are quite limited. Some objects have physical properties, others don't. Buildings, walls and terrain are notably almost purely devoid of any sort of physical property in the game world that we would attribute to such in real life. You can break a pallet with a crowbar, but no amount of rocket bombardment will ever bring down a brick wall in Half Life 2.
Further, breakability of objects tends to be limited to pre-set portions of an object and lacks any sort of procedural dynamics. Ergo, hit a pallet with a crowbar and it pretty much breaks the same way, every time. Hit those pieces and they break into a predictable number of other pieces with predictable shapes.
I fully expect Half Life 3 and the new Source engine to fully break into a more pervasive and global manner of physics simulation. I hope to see, and on some level am confident we will see, a far more procedural and dynamic approach to physics. Destructable materials with specific properties that govern how it can break, not necessarily how it will. Add to that better tracking of the disparate parts of the broken materials (such as individual bricks from a shattered wall, or the bits of wood from a broken pallet), deformable terrain, and accurate simulation of a material's structural integrity and the result could be not only visually compelling, but really change the way players approach different situations tactically.
Well that's it. If you've stuck with me this far then, well, let's be honest, maybe we should both take a serious look at what we're doing with our lives. Also, good on ya. And thank you for listening to my ramblings for a bit. I assure you, it was a therapeutic experience for me indeed. And so will it be for you, gentle reader, as you post your own speculation below. There's a lot more to cover, including story elements, and many other technology discussions possible as I've only just covered a few of the visual advancements I believe we'll see. So, go ahead. Don't be shy. Hit up that there reply button and let'er rip.
Last edited by mason.s; 08-30-2012 at 08:23 AM.