Don't you love it when people talk about graphics engines, rendering techniques and API's so authoritively when they have very little knowledge of them, let alone experience coding a renderer for a game?
What "year" it looks like it's from is pretty inconsequential, after all many games coming out this year may not look anywhere near as good as certain games from two or three years ago.
The truth is, much as TerminX said, art assets are made to such high fidelity these days that their textures maps can be rebaked and the models re-processed from the high-detail sculpts. Shaders can be improved and refined to gain better speeds and higher quality rendering based on new techniques which further allow more detailed effects to make use of the freed up processing power.
Graphical programming is such a vast and complicated job that teams usually have people dedicated to it, thus while everyone else is doing gameplay code and the like, the renderer coders can carry on adding in all the latest effects and techniques they feel are needed.
Quote:
|
I'd say DNF would look like anything else on the market. Directx11 mostly unsupported in modern games, all other stuff have'nt changed much since 2007. Shaders are mostly the same, lighting model, shadows, particles, geometry... To tell the truth - we won't get any better graphics until realistic raytracing renderer will be introduced. Well, sure DirectX 11 have some neat stuff too, but it's kinda old comparing to overall computer graphics evolution (not realtime).
|
1) "all other stuff" has made drastic leaps and bounds since 2007. If you take a look at the average game's rendering path from 2007 and compare it to the rendering path of a 2010 game then there will be major differences.
2) Geometry techniques has made a generational leap with (duh) the generational leap in the hardware, allowing everything from native hardware support for blendshapes, hardware tesselation, skinned-instancing of animated characters, etc. Shadow mapping techniques have equally improved, just look at the jagged, blotchy shadowmapping in almost any 2007 title compared to the crisp and clean shadow mapping used in today's top games.
Lighting techniques have improved more than most areas, as proved by Crytek handling realtime radiosity, color bleeding and global illumination on DirectX
9(!!!) level hardware through their Light Volume Propegation techniques.
3) Graphical quality will always improve as there is always a more efficient, effective and/or graphically correct way of doing things. It's called "research". The sole reason why DX10 and 11 has such little commercial support is there's still a tremendous amount of ground left uncovered on DX9 level hardware.
Case in point: Consoles run high-end DX9 (or "9 turned up to 11" as some people say) level hardware and have been "maxing out" the hardware for the last three years. Each year, though, developers manage more and more out of the hardware. Again a key example here is Crysis 1 being "far too much for consoles to handle" but Crysis 2, with it's far more complex and realistic renderer, fits console platforms with performance to spare.
4) That "DX11 is old compared to non-realtime-techniques" line is possibly the most pointless and ignorant comment I've ever read in a discussion about realtime rendering. It's like saying "Sure, that V8 is good and all for a road car, but it's nothing compared to the energy created by the reactor at a nuclear power plant" in a discussion about a car's horsepower.
ALL realtime graphics techniques are old when compared to offline-rendering techniques. This should be obvious by the fact that they're both "rendering" scenes with a very key difference: one is realtime and thus requires techniques to be able to run in realtime on consumer hardware, while the other, lo-and-behold, is not realtime and so doesn't need it's techniques to run in realtime on massive farms of supercomputers. That's a bit of a clue isn't it?
It's just a slight bit ridiculous to think that realtime rendering techniques could
ever be comparable to non-realtime graphics. The very nature of the two means that anything that can be done in realtime can be done many, many,
many hundreds of times better not realtime. There's graphical techniques which are only now just being implemented into realtime systems which were based on algorithms originally discovered as far back as the 50's and 60's.